Skip to content

Instantly share code, notes, and snippets.

@patcon
Last active January 23, 2021 23:43
Show Gist options
  • Save patcon/8234e6f58d54428aaafc13700203e33f to your computer and use it in GitHub Desktop.
Save patcon/8234e6f58d54428aaafc13700203e33f to your computer and use it in GitHub Desktop.

Fork the government @ CES

https://www.youtube.com/watch?v=8tc_5bM0zSM&

Transcript from: https://paper.dropbox.com/doc/Fork-the-government-CES--AzmvdRn8ZkN5Zi5~cNWblZPTAg-2WkwonjChbKbl1qMVLGJW

Slides: https://www.slideshare.net/autang/fork-the-government

Host: OK, good afternoon to everybody. My name is Giovanni Allegretti. I am the coordinator of the Centre for Social Studies of the project EMPATIA, enabling multi-channel participation through ICT application. As you can see on the website of the project, we are today hosting at the centre for social studies a working seminar of a colleague who comes from Taiwan, whose name is Audrey Tang.

Audrey is a civic hacker that we met during a series of other international relation that our project is having with different country. I had the happy opportunity to meet Audrey, which accepted to explain us a series of interesting movement that have been happening in Taiwan in the past year.

We thought in the beginning to do like a web seminar. But Audrey is a person who has a lot of to share in person, so we prefer to have Audrey here today. Because we had several friends who asked us to be able to follow the seminar without being able to be here, we are transmitting live stream seminar.

We hope seminar would be the opportunity, especially for large discussion, on issue that could be of interest of whoever is interested to party a project in terms of cross-country issues related to data, open data, to civic engagement and to transfer of accountability toward a stronger commitment in civic issues. That's one of the reason why I would stop talking here, and I will leave the floor to Audrey which is doing wonderful job on screen...

[laughter]

Host: [inaudible 02:32] you people laughing on the other side of the screen. I leave to Audrey, so I will eliminate the possibility to grow so much, maybe it's doing a horn on my head.

[laughter]

Host: I have been turned into a cat. Perfect. This is the start, starting with a laugh, it's always something wonderful, so we are happy, and we're sure that it will be a wonderful seminar. If I'm not wrong, Audrey is usually open to discussion in every moment.

It becomes usually...it's yesterday? No, two days ago, that you were doing the conference in Paris that we came like sort of collective workshop, so we hope that it will be the same today. I left the floor to Audrey. Thank you very much for being here.

Audrey: I'm very happy to be here, and since this is a very small room, I would encourage people to just interrupt me at any time, you don't have to raise your hand, but if I happen to be looking elsewhere, you just wave your hand or something. The idea here is, that for next two hours or so, I will go through five different talks, and I have no idea which of those five talks are of more interest to you.

The five talks in very quick order, is introducing the political context of Taiwan and the Taiwan's, what we call the "Gov Zero" civic movement, and I will use one particular example, a crowd-sourced dictionary, to illustrate, and I'll talk about the Sunflower movement. Is there anybody here who have heard of the Sunflower movement?

OK, we occupied the congress for 22 days and it's one of the very few Occupies that is successful, defined in a sense that first we reached our goals, and also we have a stronger consensus after the Occupy compared to before the Occupy.

Then I'm going to talk about the national level politics that changed because of the Occupy. In particular I'm going to talk about how we use the same technology that empowered the Occupy, to talk with trans-national issues like with Uber, and with AirBnB, and with the other globalized factors. Before I begin, I would like to know how everybody prefers themselves to be called or recorded, so I'm Audrey.

Q1: I'm Michelangelo, I'm also working in [inaudible 5:11] department. I can use my name Michelangelo, it sounds artistic.

[laughter]

Q2: I'm Valdemar, and it's a pleasure. I come from Mexico. No, no, I'm from the [inaudible 5:31] . And [inaudible 5:33]

Q3: I'm Bruno and I'm studying sociology, social movements.

Q4: I'm Jenna, I'm [inaudible 5:43] many. It's OK.

A: Great.

Q5: I'm Lewis, I'm also a member of the project [inaudible 5:51] .

A: OK.

Q6: I am also Lewis...

[laughter]

Q6: ...and the technical coordinator of the interview project. I'm very interesting in trying to understand from someone who also has some technical background, how this government and cities and their involvement. How can we handle this in the city. I'm very very interested in receiving your feedback and your experience.

Q7: I'm Andre, not a noob but similar. I work as a project manager here at [inaudible 6:40] and I also work with the [inaudible 6:45] project.

[crosstalk]

Host: Well you can call her "Wisdom".

A: "Wisdom", yes.

Q8: I'm Vanessa and I also work on the interview project.

Q9: I'm Penn, and I'm a researcher here at [inaudible 7:08] interview project too.

A: So, without further adieu I'll just go into my talk and at any point please just stop me if I start using three letter acronyms. [laughs] Or if I'm talking too fast.

Q10: [inaudible 7:25]

A: Yes, we could have that fail, yes. I'll say if you are interested in going into more details into any particular slide we just stopping that slide and start doodling. If I look at people and feel that you feel bored then I will just fast-forward that particular section.

"Fork", here, means in the ICT context to take something that is already here, not eliminating, not countering, but doing to some direction, but we take it to another direction. When we take to another direction, we experiment. You may fail, you may succeed. But the most important thing about the name "fork" is that we are also open to the possibility of the original maintainer. This is calling merchant-back our work.

Because the way it was made to work in the ICT industry was by people abandoning part or all of their copyright. This week in Paris is fashion week. In many jurisdictions like in the US and in Taiwan, actually, any fashion designs cannot be copyrighted because it is a craft. It's something you use every day so you cannot copyright this design, you cannot copyright this type of sleeve or something.

So anything that shows in the fashion week gets copied the next week by other designers because there is no copyright connection. Exactly because of that we see a lot of forking going on in the fashion industry. Anything that catches on, be it a color, a style, something, it becomes experimented in very different ways. Then, if some of the good ideas emerge, then it becomes just part of everybody's wearing. Not designer clothes, but it comes the fashion of the year or something. That is how the fashion industry already works.

The open source movement in the ICT industry is trying to use the same model as the fashion designers do to make the open source work, so that people who write programs to do user designs and so on can also experiment with all different directions based on existing work. Only the good ideas would be merged back to the next version of its original project. This is a very interesting idea, primarily used in the ICT industry, but the way we work in Taiwan is that we apply this idea to the government, to the society, to the governance procedure.

I'm literally from the future, I'm eight hours in the future, which is Taiwan and this place. I'm very happy to be here, jetlag notwithstanding. The point here is that in Taiwan, because we are a very young democratic country, we lifted martial law in 1989, and then the first presidential election was in '96. So basically, people only have less than 20 years of experience with representative democracy. They're not very good at it.

When we start to introduce the "Fork the government" idea with direct democracy, with deliberative democracy, with participatory democracy, it was not like, "Representative democracy has 200 years of tradition, and now we're introducing someone with 20 years, like a challenger." But it's like, "Representative democracy has 20 years, and direct democracy also has 20 years, so they are on par with each other." So that we can say we take a better idea here. The government is much more willing to listen, because they don't have a long tradition to uphold.

Something about myself, I've been working in the ICT industry for 20 years, and retired in 2013. That means that I started working in 1994 as an entrepreneur. When I retired, I do what retired people do, I start to work on charities, voluntary work, [laughs] caring about the community, making dictionaries, [laughs] things that retired people do.

Because my ICT career was built around open-source and free software, naturally, I do also my volunteer work in the same way as I do my ICT work. This becomes a very interesting factor for me, because then I started talking to the very vibrant community in Taiwan, what we call the voluntary sector, which is people, not based on taxing and redistribution, and not on exchanging of money to services, but by people donating their time, is the voluntary sector.

The magic thing with open source is that when I start making, for example, a dictionary, as an open source project, which I'll talk about in the next talk, people in the first sector, academics, could, very easily, take the product and then make it part of the Oxford University Press Dictionaries, which is a non-profit, academic endeavor. Or, when I make other deliberation platforms open source, then the National Development Council in Taiwan is free to take it. That is a subject of another talk.

But not only the first sector. The private sector is also free to take whatever research we did as part of this, and make it, too, so that Siri speaks better languages in part of its process. Or there's social text, social computing company.

The idea is that while I work all in the voluntary sector, because my work is free for everybody to use, I was able to build much stronger connections to the first sector, the public sector, and the private sector, as well. That is the basic mode amount of the cross sectoral partnership.

Is that OK? Am I making sense? OK. That's great. [laughs]

I started learning computer programming when I was eight years old in my [inaudible 13:53] . When I got my first computer as a gift from my dad, he then went to Beijing for the first time in his life, to cover this very interesting student movement that was going on. We all know how that student movement ended, the Tiananmen Massacre. He got back to Taiwan in time, so I still have a father.

He then took a strong interest in civic movement, democratization. He decided to do his PhD in Germany, studying the dynamic of the Tiananmen Movement. When he finds his professor in Germany, and he visited Berlin, something else happened in the same year, the Berlin Wall fell. Some people say it's because of the Beijing Massacre that the German people decide not to repeat the same mistake. They were somewhat peacefully democratized.

I also moved to Germany to study with my father, because that's his PhD thesis, is about interviewing all the people who were activists in Tiananmen, who flew to Paris, to Germany, to other places, when they continued their studies. They were just students. They're studying all sorts of different sciences, but was debate on how to make use of what they learned to help the democratization process. The way that they choose to, in the first, did not work, so we talked a lot about what kind of ways would work.

I come back to Taiwan in '93. That was a very interesting time because that was the first time in Taiwan when the Internet access was made available to everybody. But that is not specific to Taiwan. The entire world was getting on the Internet in the same year.

But the thing in Taiwan is that we have an education system that I was never fitting in, but I found on the Internet, this very interesting project called the Gutenberg Project. It is a bunch of people typing the books, usually in the public domain, published before the First World War, and they digitized all the books for free for everybody to use and to read. That becomes my primary education.

Once I started to learn this way, the textbooks lose their attraction to me, because that was how knowledge was being generated. In '94, when Tim Berners-Lee invented the Web, introduced to everybody, I found that all the researchers who work on those classics, are online the same time as I am.

They were publishing their preprints on the World Web, and had published their email addresses. Because the web was so young, everybody was very eager to know each other. We worked a lot, all sorts of different things, from linguistics to AI, to all sorts of different philosophy, everything, mathematics. The beauty of it is, across the Internet, nobody knows that I am a thirteen-year-old. [laughs] They treated me like a professor to another professor. We just did work together.

That becomes so addictive that I decided to quit school, because school takes 10 years [laughs] for me to reach that level, and also it would take another 10 years for the cutting-edge research to become a textbook to be taught in a university or something, from the preprints to the dissemination of knowledge.

I quit school and I helped to build the World Wide Web. Because I took so much from the web, I wanted to give back. I'm sure that only the most geeky people will recognize all the projects that I've worked on. [laughs] But I think this one, in particular, made a lasting impact that probably everybody knows about, is the Wikipedia project where people use the same open source idea, but use it to produce human knowledge.

There is one common thing in all this different project I worked with, is that it's facilitating a safe space, where nobody could censor each others' speech or drive a tank, or something, to stop other people from talking. In this relatively safe space, we can learn from each other, and try one a time and many times, until we have something that's working.

Before Wikipedia, there were at least 10 different projects like Wikipedia, that I participated in personally. None of them worked, but the Wikipedia somehow clicked. It's OK, because over the Internet, it costs nothing to fail. It is not a scarcity economy. We just keep trying until we get something that is acceptable to the world.

This is my colleague, [inaudible 18:48] , who is the primary trainer of facilitators of participatory, and also, deliberative democracy in Taiwan. She said, "Behind every technology we should identify the values that identifies why we're pursuing this kind of technology." My value was very stable for the past 20-something years. Is just this value of the early Internet that it was built through [inaudible 19:15] consensus, so that's all about me, is that OK? People are generally OK. The first talk I like to share with you is the GovZero story.

In Taiwan in 2013 there's maybe 90% of Internet users on Facebook so it's the most popular Facebook place in the world. There is a prediction that says by the end of this decade there will be more Taiwanese Facebook users than Taiwanese people. That means that the more.

[laughter]

Q11: We have the problem with cops we have more cops.

[laughter]

A: Exactly, yeah but at least Facebook accounts doesn't create, you know, congestions.

[laughter]

A: The idea is that people have more than one account, that's first, because everybody is on Facebook and steers a lot of activists people, bloggers, civic media people. They are very influential like the author [inaudible 20:15] here. Anything that he writes about the politics on Facebook gets any number of likes and shares and so on. Of course "Wired" interviewed him saying, "You know you are a very famous writer and sometimes a political commentary reaches any number of people, so do you think you will be a catalyst for civic participation?"

He was very cynical because he thinks that people who share and like his political writings are not the same people who will go to the streets. When he calls people to the streets the call to the people to the streets posts gets hundreds of thousands of likes but 10 people came or something like that. [laughs]

So the conversion rate is really, really, really low and he says that is because people are so lazy they would spend only at most one minute of their time on Facebook in response to any call to action. Calling them to go on street takes more than one minute so they will just click like and share and feel as if they have participated. This is what we call "Clicktivism", right? He thinks the idea is that we need something practical that allows lazy people to engage in the action that makes a difference. This is called one minute limit.

The GovZero is basically a movement that builds ways for lazy people to engage in real action and I'll take one very concrete example. This example is called a captcha. I assume everybody knows what a captcha is. This is the way to tell whether you are a human not a robot by typing in some random numbers or text from an image. It works until maybe last year because this year AIs are better than humans for this so this doesn't work anymore. It used to be that this tells a human from computers. What we did was that we built a web site that asks people to just type here whatever they see from the captcha box and then click enter.

We say when you do that you are saving the capture because what this is is a campaign finance record of all the elections that came before in Taiwan which were kept in a paper only form in this building the corrective auditing [inaudible 22:40] . The law that mandates this kind of sunlight campaign finance records was done in an error with only papers and Xerox printers. The law reveals that anybody can walk in to this building and require a photocopy of the type of campaign finance record but you cannot download it, you cannot take a USB, disc or anything.

Of course people propose change to the law so that we can download it over the Internet. If you think about the only negative stakeholder of these changes are the parliament legislators. While the entire nation wants this, the legislators, they schedule it but they never really debate it. It's always the last bill to be debated by the end of the session so there's nerve really actually voted on, so it's kept that way. Instead of rallying or protesting or something we decide to do something. We send people there to print these out and then we scan it and we ask people to digitize it.

To do what the government would do but do it ourselves, this is the idea of forking the government. This takes only five seconds and you [inaudible 23:50] the capture. Now when we take the print out like this, A4 paper. I tried it would take may be two minutes or three minutes to type it as a cell or something like a spreadsheet. It's too much if we ask people to do this over the Internet nobody will come -- well we know because we tried. Then we used technology open CV to cut this in to byte sized tasks what we call Tofu and then for each one you will just take five seconds.

It's the same amount as a like or a share as a comment on Facebook but instead of seeing more, cute cat pictures you can feel you are saving the country. That draws a lot of people and in fact when we build a Gamification Website of this data, if people here have played FarmVille or Candy Crush or any of those games, you know that as long as you have a progress bar in the counter, in the count down, it says how many people are playing with you. People will spend all night not sleeping doing some very repetitive task just to see the progress bar reach 100 per cent. This is human nature so basically it became very addictive. [laughs]

People were calling each other to save the country by playing this game. Then we have a lot of designers who made very cute banners of Tofu to call people to action. In the first 24 hours the first batch that we brought out from the corrective building, like more than 300,000 records were digitized by 9,700 people in 24 hours. This kind of OCR technology we call it the Otaku Character Recognition. "Otaku" is a Japanese word meaning "nerd", this is geeks, basically geeks who have nothing to do but just [inaudible 25:51] or something.

Help to do the character recognition and now we have to complete campaign finance record of the past elections. Now of course when we have this data and publish this data the corrective said again, "You know this is not so good idea that you do this because why you can say every Tofu has at least three people looking at it, two of them must agree and so on." You cannot be 100 per cent sure. "You cannot be 100 per cent sure this is actually what was printed". What we said was, "OK so you publish it, that way you can be 100 per cent sure."

Before you do this we will keep doing this civil disobedience because there really is nothing illegal of this kind of work. Now that they feel the pressure we started doing a lot of data analysis based on this kind of data. We can correlate any legislators with the kind of donation that came in the individual donations, how it correlates to their portfolio, their stock options that they have purchased, the voting records that they did. Also when the campaign finance comes from large corporations, we also have a network that says the holdings of those corporations.

We also because each legislator in some counties have this recommendation, where they could recommend the building and constructions, so we also correlate the building of constructions. The owner of those companies versus the company that have donated to the legislator. This became kind of useful so that you can decide what kind of legislators you really want for your city. Then at the election of two thousand...

Host: Excuse me I haven't gotten on the issue of the timeline so what you were analyzing first, was a system of the nation so the campaign. The other one that you show us the last one was about during the legislation what has to be done with the public money. You had to take like five years' time that means because you have to analyze what the legislator did during the mandate.

A: Exactly, exactly yes.

Host: You took some years.

A: Yes, that's a great question. When we do the voting guide, as SGO said we would only really do this before and after analysis for people who aren't going to be re-elected. So that we can correlate their actual performance. For people who are going running for the first time we cannot do the same kind of analysis for obvious reasons.

Many people run for legislator but before they are running for legislator they are running for the county or country counsellor or [inaudible 28:46] counsellor. There's still a track record on the national level even if they are running for the first time for the parliament they also already have a campaign finance record level on the local government before.

Host: You started backwards from the ones that were candidates?

A: Exactly.

Host: OK so you choose the new candidates and you see how many of them had passed. The lucky ones were those that were running for the first time didn't have any control on [inaudible 29:14] was on the [inaudible 29:16] .

A: On the other hand that also was the more disadvantaged ones, usually right? When you are running for the first time?

Host: Yes.

A: This kind of oversight is a net negative to peoples' political capital. There is actually as you said these people are lucky but when they are doing their campaign finance planning is the one they were doing under the pressure that they will be compared with the people they are running with. At the end of the campaign within 30 days we will publish everything that they have done during the campaign. They must be very careful because otherwise it will look very bad even if they are elected.

Host: It's funny because we have a prejudice against the representative politics, I mentioned that somehow you were rebalancing the problems of the newcomers. You were reducing the [inaudible 0:19] of transparency of the one of the previous government.

A: Exactly, the [inaudible 0:24] .

Q: There were, for example, cases in which you could show that there were, for example, no direct relation between the funding and what they had done. Did someone emerge as particularly honest or not [inaudible 0:41] involved with this [inaudible 0:42] ?

A: That's a great question. For example, for the city level voting in the end of 2014, there's a record number of independents running. Even an independent won the Taipei, the capital city, mayor, was a surgeon who never participated in the politics before.

The environment was such that, when each county or each city click into here -- you see 18 precincts and 87 people running -- when you click into it you will see the analysis as we talk about, but also a discussion board under every candidate.

People would crowdsource extra information based on what the public record and things like that. The primary use of this is that there were one legislator who said -- actually the one that I was showing on the photo -- [laughs] who said he is unbiased, he is bipartisan. He is not a party member. He'll often vote against his party line.

Then we go back to the legislator record and found that he never really did that. He was saying this on public television because before, in mainstream media you can get away with saying anything because people are not fact-checking it. If people are fact-checking it, it's already past the news cycle anyway, so some people won't pay attention to the corrections, even.

But with this kind of technology, as soon as that person starts saying this, the commenter started saying, below, saying that, "But that is not the case." His office has to issue a correction saying, "Oh, we not, we don't mean by the previous, because he was, you know, five terms. We don't mean the previous term. We mean the terms before that."

Then we speed up the digitization. In the 24 hours after that, we digitized everything in the previous terms and found out he really never voted against the party line in any of previous terms. So, that becomes a tremendous pressure on the legislators to watch what they speak publicly about their past records.

Whereas before you would take days or weeks in a news cycle to correct or to find the flaws in their speech, now it's a matter of hours or even minutes, and that changed the dynamics.

People would click into it and say, "OK, my precinct has 22 people running for councilor, and after I see this website I now only have to choose between two." Because people become much more informed, and informed in a very quick way of what kind of legislators they really want. This negates somewhat the mainstream media on a representative democracy.

We can also see which legislator received campaign donations from their own parties. We can analyze the Nationalist Party versus the DPP party. The red one was a, I must say, ex-mobster. Because he's not really a mobster now, [laughs] a ex-mobster who is very rich so he can finance his own campaign. The idea is that we make it very clear who is getting how much money from what.

This kind of crowdsourcing of campaign finance records, of the news and so on is not limited to domestic policies. When we're doing campaign finance records we get contributions from outside Taiwan also, because you don't really have to know Chinese. If you can read digits, you can help us digitizing the campaign finance records.

When the international community needs help, we also help. For example, this is working with the Humanitarian Open Street Map Team. Open Street Map is like Wikipedia, it's a crowdsourced map.

When Nepal had this earthquake, all the major maps like Google Map, Bing Map, or Apple Map only have the mapping data around Kathmandu and the major city connections. The street car doesn't drive to those rural areas, but those were the hit most hard by the earthquake.

What the H-O-T team did, in conjunction with the local chapters, was that we divided the satellite image that was taken before the earthquake into very small, again Tofu. Then we let people who have never mapped for their everyday lives take a half-hour course, and then start to look at just one tiny piece of the satellite image and mark the roads and the buildings on it. That's all we ask them to do.

Just like the Tofu OCR project it has to be two or three people working on the same grid and then mapping expert will do the review cycle and so on. But all this was done in 24 hours. After 24 hours of the earthquake, the satellite company donated the post-earthquake satellite image. That is the first time after this recovery we get donation in such short time-frame.

So for the second day, people focused on the post-quake imagery, doing exactly the same thing. But now marking which wells are broken, and which buildings or camps have appeared, whereas they were nothing before on that grid.

On the third day, when the United Nations, the Red Cross field team came, they have a open street map on their mobile phone that shows which roads are broken so you can enter there, and where are the refugees' gathering, so you best deploy your logistics there.

This is something that couldn't really be done with their ordinary helicopter, or any this kind of work. This has to be done on the satellite level. So, of those 2,000 mappers, maybe 200 of them were from Taiwan. Our President-elect, Dr. Tsai Ing-wen -- she was still only a candidate, presidential candidate at that time -- she helped launch a Facebook and other social media to call for her supporters to do this kind of humanitarian work.

What I'm saying is that with this kind of tool, the boundary of nations no longer exist as long as we turn something into a crowdsource project. We can do this kind of crowd source work anywhere and we can map all the buildings and all the roads that was impacted by the Nepal earthquake. Using these as examples, I want to say that g0v, this movement is really way for three different kinds of people to learn from each other, something that they have been missing in their previous lives.

The core people, the first of our people who started to register the domain of g0v.tw had a very simple idea. Other government websites in Taiwan ends as gov.tw. So, for example, the environmental agency is this. Now if you change in your browser the "o" to a "0" you get into the shadow government, that is built by g0v.

It shows exactly the same data as the environmental agency. But where as the environmental agency shows just tables, and readings, and very boring things, in the g0v environmental agency, you see pretty pictures. So [laughs] this is the air pollution level at this point.

People have use that in news broadcasting, in every day because this was very useful. It's very, a lot of fun. You can see the PM 2.5, the O3, the rainfall, whatever. Because it's open source, you can look at the data, and the code and if you don't like the color or you don't like the font, or if you don't like how the way that the progression was spent, you can contribute very easily.

Q: You can customize the picture.

A: Exactly. Yes, you can just click the edit button. It didn't start this way. It started in a much more geeky way, but then we have professional designers who are very much into this Japanese comics and manga called Neogenesis Evangalion.

He used that font from the Evangalion fonts, so that it now looks like something from a cyberpunk design. [laughs] That was the idea, anybody who want to make contributions make contributions.

This is much cooler than the government website for obvious reasons. [laughs]

Q: The data were the same? Were exactly the same source?

A: Yeah, we've web scrapers or whatever to take data from the government, but we present it in a way that's open data, and that allows everybody to customize.

Q: So while in the previous thing you showed us, you created new data, for example, in Nepal, in this case, the data source was not put into that. You were accepting the data, and just helping them to be more clear and visible.

A: Yeah, exactly. And also mashing up it with other sources of data. It works with other major government organs. So for example, the legislation is "ly" and so if you change to a lyg0v.tw, you see all the bills being debated and it's like a shopping cart.

You can see a progress bar [laughs] of where the bill is and you then you can see a difference that shows in red and green, the before and after bill. The way geeks like it. [laughs] That was the initial group of people.

It is not just the geeks because while we are the open source hands-on hackers, we were actually not the main experts at civic participation. The traditional NGO people, the traditional mobilizers, organizers, people like that, they were very far from us. They were very little overlap between people who actually understand environmental campaigning like the Greenpeace people versus the people who do this environmental visualization.

We actually got a lot of things wrong in our first tries because we were not environmental experts, so we had to make contact. We made contact first with the people in the civic media like the bloggers that I mentioned, who has this very cynical view because they thought this software cannot really change anything. Facebook doesn't really motivate people.

Then we say, "If you go to our hacker's zone and present your ideas, we will help to amplify your idea and then to reach more people into actually make a count, a real impact." Basically, we teach them this hands-on spirit of not just writing blogs but doing something.

We engage with the social activists, who are again, very hands-on, very public-spirited but their main problem was that they don't trust strangers. This is a fact. [laughs] There's a lot of, I wouldn't say schism, but doubts of people going to different directions or a misleading data or whatever. The first thing they ask is always, "How do we know that our opposition movement will not feed us fake data or things like that?"

The way we went with them is that showing that with sufficiently number people and ICT technology we can build a peer-review system that is safe against vandalism and doesn't have a [inaudible 13:09] care about. So we teach them the idea of open source while they teach us the idea of the public spirits, the areas of concern.

Because g0v is using this domain, we are not limited in our projects because for every government agency or ministry, there is some social activists working at that area. It's not really limited to just elections or democracy. It could be environment or agriculture, education, whatever they named it, there is a ministry for it. There is a g0v website for it.

This is three very different kinds of people, but because we learn from each other, we generally build something that is useful for all the three different groups of people. The way we do it is with very good food.

We hold every month hackathons and large hackathon that happens every odd months. Like in March, it's March 5, the open day that day. We have anywhere between 100 and 600 people in the same building, in the same room.

We use the standard, what we call the Open Space Technology. You'll be aware that I don't use three-letter acronyms. [laughs] Open Space Technology, it's spelled out in full. We have this large room where we invite people from all walks of life to join and to share food with us.

When people ask about, "Where do g0v gets the money?" Well, it's our campaign finance. We say that we only spend money on really good-quality food. It's not that much money. We don't spend money on anything else. We use free software. We don't buy commercial solutions. It's all zero or very near zero cost.

We spend a lot of effort and time to think about very good food because a month after a monthly hackathon, most people forget about all the projects and all the people but they will remember the food. If the food is particularly good, they will be inclined to join us in the next month. If the food is so bad, they say bad words about our community. [laughs]

Q: So you rely on low instincts of man?

A: Yes, the very basic instinct of people. Very good coffee and so on, we have a special domain called g0v.cafe [laughs] where you can issue and print and get this very high-quality instant coffee with g0v printed on it is our souvenir. If you go to this web address, we ask for donations for hackathon's coffee and food.

That's our fundraising way. When we raise funds, we spend it on high-quality food. We spent everything immediately so we don't keep capital. When we run another months of hackathon, we ask again for donations.

For your donations, the only thing you get is a guaranteed ticket to the hackathon. [laughs] This is purely without commercial interest. Because the hackathon's were sometime very popular, they got sold out in hours. A guaranteed ticket to hackathon is still very valuable. People are willing to sponsor for that, for the good food.

What I'm saying is that with this way of the thing in common between the social activists and the free-software people and the civic media, is they love good food and good coffee so they come for the food.

When they join us early in the morning, in the hackathons -- as I explained, the large ones every two months. It's 100 people. The every other month is maybe 50 people or smaller -- you will see a bunch of stickers on the large hackathons. Those stickers, each one is like this high, you would take the sticker that describes you.

For example, maybe I'm good at storytelling, maybe I'm good at making music, and maybe I'm good at coding Python or something. You take the stickers that represents your interest and put it on your shoulder. There's also write-in stickers called "Nobody." [laughs]

If you have a proficiency that we don't have a listing here, you can write it yourself. If sufficient people use this, then we'll print it on our next version of the stickers. This goes through many iterations.

Q: Yes. This same view of your stickers, have two issues that could interesting for us. The first one is that in some parts of the [inaudible 18:05] Germany, people has stickers to declare their belonging to lobbyist or potential...

[crosstalk]

Q: ..for example, membership of an association of shopkeeper, trade unions and all that. In order to when they speak, when they're talk in public people know from what position are speaking, if they have one.

The second issue is that normally Open Space Technology have the two feet law. That means if you cannot contribute to a self-organized group you go away in the next one. But here, you can see what kind of people there is there. If you think there are too many engineers in that area or too many cook in that area, you can move not for filling or not [inaudible 18:58] the group, but because you think that the group is too homogeneous.

[crosstalk]

A: Exactly, exactly, this is a diversifying way of the walk-in role. This is not voting. This is diversification.

When you join for the first time you would take a deer sticker. Then when you're here for many times, a veteran, you would take a Taiwan bear sticker and put it somewhere prominent. So what are those stickers for?

The process is all day, sometimes two days. At the beginning of the day, everybody who has an idea, "I want to do campaign financing. I want to do recall campaign. I want to do whatever," right?

Then I go on the stage, use PowerPoint or some other tools to make a pitch for three minutes. At the end of the pitch we ask everybody to declare how many people of what expertise do they need for this project to function.

For example, for a public finance campaigning project, they will obviously need one engineer, and one designer, front-end designer. They will need one legal people to negotiate with the [inaudible 20:16] . They will need a storyteller to turn this into a public design for social media. That's the initial four talents that we would need for it to succeed.

Now after maybe 20 projects, each present their ideas. People start to play musical chairs. That is to say, they crowd around the corner where the projects need those declared number of people. You can see it at once that this project already has an engineer, or that this project doesn't really have a storyteller, so as a storyteller you would join them. By the end of maybe 10 minutes, 15...

Q: Certain markets of talents?

A: Exactly, exactly, and then, by the end of maybe 15 minutes...

Q: [inaudible 21:03] .

A: Yes it is. There will be a lot of deer caught in headlights, staying in the same position, not sure where to go.

There will be a lot of bears coming one on one, and joining to the side of this first-comer, and start to talk with them, saying, "What is your concern? What brings you here? What is your daily life? What kind of issues you care about? Walk with me."

At the end of that walk they will find themselves in a project. That is the kind of mentoring that we do.

For a hundred people at hackathon, usually maybe 40 people or 30 people will be first time in the hackathon. Currently the demographics is about maybe 20 percent engineers, 20 percent designers, and then other people -- story tellers, news media people, there's a lot of public servants now and people from all walks of life.

So when they have a project now, which we call [inaudible 22:14] meeting a gap, the reason why we don't call it a project is that people who are founding projects, in Chinese at least, have this notion of project leader, project commissioner, project organizer. But sometimes people just identify a gap and they walk away, because they just really have an idea. They work on some other thing.

So they should not have authority or control on the people who actually fill the gap to do the actual work. By rephrasing things this way, by identifying a gap in reality, where we have to hack on, we erase this kind of top two-button function of organization, so anybody who walk into this gap is a contributor.

Then we hack for an entire day or two days. Now after a day or two days, every project has five minutes at the end of the closing day to present what they have built over the day. Usually they will have a prototype already, because they have the right talents.

Then their presentation would say what in the next month, what is their participation policy. Some projects would say, "We meet every Friday after next month." Some projects would say, "OK, we meet in this chartroom in IRC or Slack." Some project would say, "OK, this is a long-term project. We just meet again at the next month's hackathon."

Every project is different, but this is very important because then it connects people who already connect to the project to future meet-ups. Many projects have this weekly meet-up where it's just three people, five people, seven people, but they do sprints to make the project actually happen between the large hackathons. When you participate in those sprints of meet-ups, you will meet more people. Then you will tell those people that hackathons are a great place, so they will join the next month's hackathon and then identify more projects. [laughs]

It really is a circle, and g0v is not an organization, this is just a way of doing things. Its just a habit, a way of living. Anybody who comes is a participant, and who contributes is a contributor. We don't have a leader. We don't have a spokesperson. It's just space, online space and offline space.

Q: I can show yourself one thing?

A: Yeah.

Q: My question was quite simple. Just give us example [inaudible 24:52] recently, give example of this [inaudible 24:57] mechanism. do this kind of project having strong ICT content [inaudible 24:60] project are completely...

A: Yes, that was my next line. [laughs]

Q: Oh, my question was different. You often say "we" and so I was trying to understand the "we" you say what is referred to. Because now, you're talking of something which is a space of encounter and not a movement, so when you say "we," you talk about what exactly?

A: In order to be a g0v project, the only requirement is if it's a coding project to use a open-source license, meaning that other people can use your product without asking your permission. If it is a non-coding project, we ask people to use Creative Commons, which again, is a way to say people can copy your work without asking. When I say "we" and people who agree to this protocol of social, I would still say it is a movement.

People say open-source movement, the Creative Common movement, the free culture movement, but this is not specific to Taiwan, or specific to g0v. It is a global -- not even global -- this is a both global and also on the Internet mixed-reality kind of movement that is happening all around the world.

G0v really is like a gateway into this wider, world-wide movement that is defined by the open source and Creative Commons, free culture movements. When I say "we" I mean the people who see the way of this way of doing things, and is willing to contribute or at least participate and their [inaudible 26:50] .

Any other questions?

Q: Do you think that others also use "we?" So you feel a strong identification with those principles. Do you think that this is spread around so you all behave in the same way?

A: Yes, exactly, because we don't really have a spokesperson. There's nobody who speaks for g0v. When a media -- this used to confuse the media to no end -- they would say, "OK, I would like to interview your leader," and we were like, "We don't have any leaders."

There was a very early motto of g0v that says, "Don't ask why nobody is doing the work. Admit first that you are that nobody." This is a combination of many different slogans before, but this means that it is OK to start doing something imperfect.

So when the media people says, "Why is nobody working on the project?" we would say, "OK, be that nobody. Come to our hackathon and you start a project."

So this is a way what we call, "Worse is better" which is a core open-source tenet. This is an example for you. This is zeros g0v hackathon. We count zero-based, we're geeks.

On the zeros hackathon, there is a logo of g0v that was designed by two coders, two friends of mine. I wasn't joined in g0v at the time. I joined two months later.

They were brilliant coders. They were master hackers, but they sucks at visual design. Anybody here can design a better logo than this. It's very difficult to get a uglier logo than this, especially with a jpeg art effect.

This is so ugly. They had the guts to print this in A1 and hang it in academia [inaudible 29:09] open space as a welcome banner, "This is the g0v hackathon. Come and join us."

So for the 100 or so people joining, it became a very sore spot because it's so ugly. One of the visual designers wrote on social media that, "This is so F-ing ugly that I cannot do anything productive, unless I make a better logo." So we infuriate a visual designer. He just looked at this very ugly logo, feeling completely outraged, and produced a better logo.

So this is at the end of the day at the hackathon. This is his only contribution, because he is immobilized. He cannot do anything else, so he makes something better. This is like discussion, inspecting, voting. This is actually a lot like your logo. This is a lens that is viewing the society, and then this is melting. It has some culture in it, and it's much easier on the eyes.

Because he abandons copyright and the Creative Commons Zero, people were free to then iterate, to improve on this idea, which is very important, because this doesn't look so well on mobile phone. If you look at it in a very low resolution, it doesn't look like G-O-V or G-0-V anymore. It will look like G-Q-V. We registered GQV.tw just because of this, [chuckles] because people type in the wrong way.

Then it was iterated. A much better visual design came, which looks like this. On low resolution, we'll just show the square, which is very identifiable as a zero.

Without the two shameless people who published their ugly work, they would not infuriate a designer. If they do not infuriate a designer, we would not have a better logo. With very minor things like this, we overcome the Asian culture of what we call "losing face."

There is nothing to lose while doing something imperfect. There's this Leonard Cohen song that says, "There is a crack in everything, and that is how the light gets in." Without this imperfect thing, nobody will come and help, but if you do something a very ugly way, everybody will come and help you. [laughs] That is the basic operation way that g0v works.

That's the first hour. Is it OK with people?

Giovanni Allegretti: Yes.

Audrey: Now a larger project. This is a project that brought me into g0v and it is a dictionary. It is the Ministry of Education Dictionary.

Giovanni: Is that the project that was born inside the hackathon?

Audrey: It was, but it was on the first hackathon. It used to confuse the media, because the first hackathon is the second hackathon, because the zero came first. [laughs]

On the first, which is really the second, hackathon, we started this project called the MoeDict. By last year -- this is an old slide -- we have seven million visits per month, and we have half a million Android, iOS, Windows phone, Symbian, Blackberry, or whatever users.

People use this to teach in school, Chinese, because Chinese in Taiwan is spoken in many, many different ways. There's Mandarin. There's Taiwanese Holo. There's Taiwanese Hakka. There's also the Taiwanese Austronesian upper region Amis. There's also Tibetan, because of the Tibetan Buddhism influence, there's people from the mainland China, who came to live in Taiwan, and so on.

There's just a lot of different ways Chinese and Austronesian language is spoken in Taiwan. This dictionary website, this project, integrates everything together. You can type in French, in German, or in English, see the Chinese word, and how the Taiwanese Hakka, Holo, Mandarin people pronounce it, the strokes of how it should be written, and so on. It's a very useful website.

It started in the first hackathon, by my very old friend Ping Yeh. He was a physics professor, quantum physics, in National Taiwan University, but he joined Google Taiwan, to work on Google's cloud center in Taiwan. After working in Google for a few years, he moved to the Valley. Now he works on Google Analytics, I think.

When he moved to the US, he brought his children with him. He found that it's very difficult to teach his children Chinese, when he was in the US. Learning Chinese is hard enough, [laughs] but learning Chinese in a foreign country is very difficult.

The way we learned Chinese in our generation was through the Ministry of Education Dictionary, which was available as a website in the Gopher protocol. Many people here will not remember this, but before the World Wide Web, there was Gopher. There was Archie. Those were the pre-World Wide Web kind of World Wide Web. That's how we learned the dictionary, because it was published on Gopher.

Of course, after the World Wide Web came, everything becomes the World Wide Web, so it has a website, and that's how we learned from it. But this website was built at the dawn of the Web. Nobody really knew how to do websites at that point, so it was an absurd website. But the content is top class. It is the definition of Chinese language, in classical Chinese.

Because of Cultural Revolution, Mainland China doesn't really retain that much material about classical Chinese anymore. It's like the Latin, or ancient Greek of Chinese. This dictionary has all the citations, all the etymologies, and everything about classical Chinese, and how it's evolved into modern Chinese. This is linguistically a treasure.

But, because this website was built in '96, there was no idea of bookmarking or a permanent link. So you can not bookmark and visit again. It won't work. And because it was using legacy encoding, there was no Unicode at that time. All the difficult characters were done in 24 x 24 bitmaps, which you can not copy and paste.

It doesn't support mobile phones, because there were no mobile phones [laughs] at the time. At the time, if you view source, you will see it's best viewed in IE-5 or Netscape 4.7+. The "+" here is redundant because Netscape has discontinued after 4.7. [laughs] This is a very, very old website. I'm just trying to give you this feeling of a ruin, a living fossil or something of a website.

Because HTTP 1.1 has not been invented at that time, the idea of a keep-alive connection is not invented, so it automatically logs you out after 30 minutes of inaction, to conserve server resources. The funny thing is that there is no login button, so you will get redirected to the home page after idling for 30 minutes, saying, "Sorry we had to log you out."

"But there's no login button, why are you logging me out?" [laughs]

All the dictionary websites after this took this as the spec, because it then became as procurement as part of the functionality spec. All the modern websites built by the ministry of education in the next 10 years have the same function, even if it makes no sense now. It became a ridiculous joke of a ICT technology, even though it's a great dictionary.

Having basically attended a hackathon from abroad...

Giovanni: Was it a [inaudible 07:43.14] that uses the function?

Audrey: Switching to engineering-ment, as a geek, HTTP was invented in a very draft form where it was not possible to automatically keep the connection between the browser and the server. This is why we call it, "stateless," so you make a request, like ordering something from the menu.

Then browsers evolved. Netscape 2.0 or something introduced this idea of a keep-alive connection where, when you make a request, it doesn't close the connection. It will keep it open in case you want to click somewhere else. But, because the server software were written in '96 before the HTTP 1.1 spec, it doesn't know that after five minutes of inaction, I just terminate this browser connection.

Every browser connection was consuming one process, one resource on the server. Because the server cannot auto-disconnect, it will become overloaded if too many simultaneous connections are kept.

Giovanni: It happens when the [inaudible 08:50.22] searching.

Audrey: Or something like that, yes.

In '97, most of the modern Web servers has this idea just called, "keep-alive:closed." If a web server says this, the browser will not keep the connection alive, but because the website was not updated, there's no budget for it after '96, so a technical problem in '96 lived on for 20 years, even though all they have to do was to upgrade to the newer version of the NCSA, later Apache Web server.

Nobody was around to do that, so the same function [laughs] was there for the next 20 years. It was basically "out of maintenance." All the staff they have, just know how to buy more hardware, things like that. The programmers, they all went away.

I imagine this is not our only problem. When you contract out the ICT solution, the funding is just for one year or two year, and the team disbands, and it's not open source, it becomes very difficult to get the second team to carry on the work.

Sometimes they just rewrite everything, and if they rewrite everything, they ask for more budget, but if they ask for more budget and the ministry doesn't have it, what you have is a legacy system that runs for 20 years.

Ping Yeh attended the hackathon from the US and that is when we started to do live streaming for our hackathons, because they had to attend from all over the world, not just Taipei City.

For that hackathon, we had Taipei, Tainun, and Taichung, that's three cities in Taiwan, simultaneously using this kind of tele-presence. He outlines a vision saying that I need data collection, data cleaning, structured data and so on, other the requirements.

Within 24 hours, with the hackers downloaded everything from the dictionaries and Ping Yeh designed a JSON, that is to say structured data, that matched HTML. Some other hacker wrote a program that converts the website into the structured data.

Some other hacker converted the relational database, and some other hacker turned it into a website, and an input method extension, and also an online dictionary and offline dictionary, and a mobile phone. All of this was done in 24 hours. [laughs] . This is called rough consensus, because we don't need anybody's permission. Everybody just works on whatever they need or their children need, with out any niche coordination.

About those 24x24 bitmaps, where we need to identify the Unicode for it, we set up a Google spreadsheet that asked people to look at the pictures, and then using handwriting input method something, to try to identify the Unicode for those characters, again, within 24 hours.

We had participants all around the world, about 100 people. We thought it was a lot of people at that time, about 100 people identifying all those difficult words from pictures into the Unicode components, so we can copy and paste the definition in the dictionary.

The fun thing of this is that we brought down Google Spreadsheet, because there's too many people editing in the same time and there's too many pictures. his is going to become a trend. Any time we try a new service in g0v, we will bring down their service. We're like a scalability testing team for new services on the Internet.

For this case, I had to actually build a new spreadsheet system called EtherCalc. EtherCalc is like Google Spreadsheet, but because it's free software and because you don't have to sign in, the server overhead is much lower. People can just do whatever they want on it, without incurring the same server load of things.

Now we'll face the legal problem. The reason why the Ministry of Education, nobody was doing this before, was that they say, "All copyrights reserved" on their front page. It says, "You may not link to individual entries in the dictionary." Actually, you cannot link to it anyway, but they said you can just link to the home page and "all copyright reserved."

In Taiwan, the Fair Use law says you can use a reasonable part of a government-produced Web page, as long as you're using it for non-profit purposes or for educational purposes. I imagine it's the same doctrine anywhere. But we're not using a part. We're using 100 percent of that data.

Correspondingly, we have to relinquish 100 percent of our claim. At that time, there is an invention firm created [inaudible 14:01.15] called Creative Common Zero. When people use CC0, they say, "I relinquish even the attribution of copyright." It's like it enters the public domain immediately, without waiting for a me-too guy and 70 more years.

With CC0, we said all of our code in the dictionary are CC0. All our data that we converted are CC0, meaning that we're really just doing data conversion work for the government. We're not really claiming any copyright on it. We argue this is a fair use, because it's proportional to the proportion that we use -- zero, 100 percent [laughs] .

This is a very interesting legal case. While the lawyers in the ministry are debating this subject -- they took a year -- we try to say, because it's not one individual doing in, it's 30 hackers doing it, it's called "civil disobedience." [laughs] We maintain this fair use, peaceful doctrine by that.

Now we have structured data, what we call five-star data, meaning that every word has a URL, has a website address. So data, [Chinese word] , in Chinese, everybody knows it's website address in MoeDict, because it's just moedict.tw/ [Chinese word]. There's no need to remember. This is again, the hack as g0v.tw. You don't have to remember our website. It's just government website change.

With linked data, Open Linked Data, where you mouse over, you hover over any word in the dictionary, it will show a cross-reference of the dictionary definition in other dictionaries or in the same dictionary, in a linked data kind of way. This is how we did permalinks.

People started showing each other definitions of words on Facebook. We did for Facebook, Twitter, and Google+ what we call an open graph cover image. The cover image means that anybody who ties moedict.tw/ [Chinese word] , mean open data workshop. Of course, there is no such entry in the dictionary, yet.

Instead of saying, 404 Not Found, we say, this is the definition of Open, this is the definition of Data, this is the definition of a workshop. Then we use beautiful calligraphy to produce an image that is whatever you have just wrote.

This became the sweetheart of mobilizers everywhere, because on Facebook, this virality is 10 times more than compared to if you just had a message without a picture. If you have a picture that doesn't match your message, it could be a counter-influence. But if you have a really good high-quality image, maybe it's copyrighted, so that's another problem.

The mobilizer spent a lot of time searching for high quality images that matched their message. Now with no MoeDict, they don't have to do that anymore. They just say, "Go to the streets tomorrow," and then you have a banner that says, "Go to the street tomorrow." It looks like this. Then you can use whatever message you have and, because we abandoned copyright, nobody will sue you. This becomes the preferred banner tool for online mobilizers in Taiwan.

With this kind of technology, if you think this calligraphy is not fitting you message, we offer you a whole menu of open source or free fonts, so you can tailor-make the font to suit your message where you want it to look very violent, very peaceful, or very classic, whatever. People use this for most mundane things like, "I feel so good today," or whatever. They just let their friends know their feelings.

In a sense, we hack into Facebook algorithm with this kind of dictionary technology. When you see people share like this, you can then re-share it on Facebook, Twitter, and Google+. It went viral very easily. Now with this kind of virality engine, we have 7 million visits per month.

When you have 7 million visits per month, whenever we put a call to action on the website, as MOE Dictionary, even though maybe 1 in 1,000 will donate their time, that's a lot of people. We will start to ask people's time to digitize old dictionaries made in the '70s, that we only have very low quality print-version now.

We do a scan and then we ask people, again just like the [inaudible 18:43.18] , to adjust for the OCR because the OCR of low quality sometimes makes mistakes. You look at the OCR result, you correct the mistakes, and you send it out. Again, we have the progress bar that lets everybody know that you are contributing to preserving the culture of the Aborigines, you're doing a good work, thank you.

And it worked. The Amis-Mandarin-English Trilingual Dictionary was digitized in just two days and a half by a lot of those people, and then we have an electronic dictionary. People who love words spent a lot of time on this just doing free work, and then digitizing the dictionary that they care about. They don't even need to know Amis or Francais, because it's just typing in Latin characters.

While the ministry lawyers were debating our case, they were having an activity where they asked people to spot problems -- typos, errors -- in the MOE dictionary. They're finally, after 20 years, trying to do a new version.

Within those 18 days, we started to campaign on our version of the MOE dictionary. We use a program to identify two entries that cite the same source, but differ on one word. Meaning, one is probably a typo. A computer knows which one is different, but it doesn't know which one is correct.

We asked our readers to Google, see which one is correct, choose the one there that is correct, and send it out. We collect more than 5,000 errata from the dictionary this way, and sent it all back to the ministry. They have maybe 6,000 contributions, and a majority was from the MoeDict.

On the day of our sending of this dictionary errata in this huge spreadsheet, they gave us an award and say, "OK, what you're doing is very useful." If they sue us, it's not against 30 hackers any more. It's against thousands of language lovers, teachers, high school students. They cannot risk alienating any of those people, because they're the core constituency of that Ministry.

When we involve all the teachers, all the students, all the linguistic scholars, and academics, the civil disobedience becomes a national thing. Then all they can do is, "I give you an award. This is very useful."

Now we digitized the aborigine dictionary. This is the Amis [Chinese word] , meaning a square. The way we do this is not because a project leader like me or Yeh Pint, knows Amis, Français, or Hakka. We don't actually. The way we do is that we work on a language we know, and then we open-source everything. Then any other language, just take it, and building their own MoeDict website.

This is different from how the Ministry used to do things. The Ministry used to think in a coordinated way. They have a committee of five people, and when the six people join, they have to know other five people, each on representing one language, a community. There's a lot of fighting of which council member represents that minority.

Some Aborigine will say, "We are actually two tribes, not one tribe, so it's not fair," things like that. Committees have a lot of problems exactly because human beings cannot really know 30 people in the same room, and have the same share of time.

The way MoeDict is doing things, which we call collaboration, we have started something that we share. Then any other language is free to take it in whatever direction, but sometimes, like in Amis, they will have very good idea. Then the Mandarin dictionary will just merge, because everybody relinquished their copyrights into the Mandarin dictionary.

Then other dictionaries will follow. If some ideas are so fringe that it's only useful for their community, it's still OK. They just fork the project and maintain an independent website, specially tailored to that community.

This is what we call a rough consensus. As long as people agree generally on the direction, even people who collaborate with enemies, are able to work together by going on their different ways. That's another half hour of talk.

Giovanni: What are we doing? Do you want to have a break?

Audrey: Yes.

[crosstalk]

Audrey: Let me finish with just one last slide. Ping Yeh registered this domain. Now you know what it means -- edu.tw to 3du.tw. After two years, the Ministry of Education finally says, "All our dictionaries, past, future, are released other creative comments. You don't even have to do civil disobedience or fair use. We now join your movement." It took two years and a half.

For example, we have an open data portal where we wrote an open data license that we will want to see from all the communities in Taiwan. The government, the National [inaudible 24:19.14] Council, saw it as a much better license than the license they were using, which is not 100 percent open-data definition compatible.

Because they merged back the data ported from g0v, suddenly other levels, the city level and the national level, are in an open-definition compatible license. Because of that the Open Knowledge Foundation Network, OKFN global, opened their index. Taiwan became, from g0v when it was first founded, to 36th place, to 11th place, to now, this year, the first place.

That is not because we produce more data, but because all the agencies merged, the gutsier way of doing open data including the license and the infrastructure.

I think we will have a pause that will be 10 minutes. People are still free to talk and ask questions.

A: We will continue for at least an hour. And in the beginning of the hour I will start talking about one particular case. That the Gov-zero, it's actually the Zeros Project of Gov-zero, it's [email protected].

The magic about this project was that it was the initial saying that promoted the creation of this domain. It started with a television advertisement. It's how one government has, at that time, what they call an economic boost up plan. It's like one of those five year plans.

The plan is very complicated so to spend the budget wisely they reallocated all the different ways to spend public money on a national level. Now they filmed a movie, a advertisement just five minutes short movie, that they broadcast on YouTube.

It was the first YouTube account movie that they published to the [indecipherable 01:25] in addition to television networks. That advertisement which I will spare you is basically showing the economic power-up plan on top and then five people looking very puzzled to this banner.

Then the voice-over says, "The economic power-up plan is a very complicated plan. We would like to explain in five minutes, but it is not possible. So we will just tell you that we have everything figured out. And in the five minutes time, I would ask you to not question the government's decisions and just go on with it. Because to do economy, we don't need more debates, we just need actions to go ahead and do it."

Q1: So trust us, and also don't criticize.

A: Yes. Trust us and don't criticize. It is a complicated plan and you must trust us and we don't trust you to understand.

[laughter]

A: This is so insulting that the director of that advertisement was replaced promptly. Also, that sufficient people used this YouTube feature called report as spam on that YouTube advertisement so that the Taiwan government become the first government to be marked as a junk mailer and to be removed the account from YouTube. People were just saying this is spam.

Of course all it takes is for a letter to YouTube to restore their account but one of the founders of Gov-zero, [indecipherable 03:17] was at the time in the Yahoo Hackathon. They were doing a very commercial hackathon where people present commercial ideas.

They were trying to do some e-shopping, this kind of hackathon. Because of the YouTube account was restored that day it did feel very much insulted. They changed their topic on the very last minute and decided to download all the PDF and word files of the national budget.

Saying OK, you don't trust us to understand it because it's too complicated and cannot be explained in five minutes, but maybe the problem was the way you share your data. Maybe the problem is not with our brain, maybe the problem is with the way you show it with 500 pages of PDF files.

They took those 500 pages and made a website, budget.govzero.tw, that shows the same national budget but in a way that could be explained at the end of the hackathon in five minutes. As existential proof that this could be understood in five minutes and they had to register the link for it.

They registered this domain and that was the beginning of Gov-zero because they got some minor award from the hackathon and they used the money to buy very good food for their fellow hackers. That was the creation, so to speak, of Gove -zero. [laughs]

Q1: No food, no hacking.

A: Exactly. Exactly. Now from the environmental agency, I will now take you to the budget side. This is the Taipei City government. The web address is budget.Taipei. If you type budget.Taipei in your browsers or something you will go to this website.

This is saying very prominently on the top saying, this is a fork from the Gov-zero central government website.

Right, it's right there on the top. It has any member of Facebook likes. What this is doing, is it's saying to you that the education budget is the largest among all the city government budgets. However, it's getting cut by about two percent, compared to the previous year because of the color. Everybody could see that right?

Then the environmental...

Q1: The pink is the cuts?

A: Yeah, the pinks are the cuts. The red are the severe cuts. The green are the increases, year by year. The one that was a red circle around it, are the maintenance costs. Meaning that this has to happen every year. You cannot cut this.

Yes, any questions?

The non-red ones are investment meaning that it is subject to change year by year. Now, you could see that in a glance, that for example, social housing, social security is getting more and more and so on. If you want to look at the entire budget you can also do that. [laughs] The ones that are getting increased float.

Q2: That is organized by areas with cuts and area with...

[crosstalk]

A: Exactly.

Q2: The colors would [inaudible 06:50] .

[crosstalk]

A: Yeah, that's what the color is saying. You can see that the top one with the red circle, I have no idea what it is, actually. Oh, it is the pension because it is that obviously, is letting go of all your private chauffeurs for the city officials in a way to reduce costs or something and other benefits that the Type A officials are having.

They're now just taking the metro like everybody else. They need to pay the pension to those private drivers that were driving for them before. Anybody in a few seconds can see what this is doing.

Q1: In the United States, I don't know if you know. There is a thing called debt in taxes, which has been done by a marketing agency. There was a representation of the American budget, which is in the studio of each one of the members of parliament because it is the only way.

They did like an experiment and now it's a three-meter-long poster they have in a building, which is very interesting. It was an experiment of marketing but it became the only way for the member of parliament to understand the budget.

A: Exactly, to visualize. Now, here what we see is four buttons, meaning that you ask for more of this budget. You do not understand the explanation, you want to cut it or you want to delete it altogether.

Whenever you click that, you will be asked, why? The interesting thing here is, all of this in Chinese obviously but I will go to maybe social security, labor, whatever. Social, OK or I think labor is more interesting.

Q2: You have to be logged in to do it?

A: No, to comment of course. For example, in labor where you have a re-education and whatever. You see a tree map, you also see all the fine details. You can see for the labor insurance, for the labor union or whatever, how they exactly are spending the money.

This is what they send to the council. This is exactly the same work as they send to the city council. With software, we can highlight the part that gets increased or gets decreased and the reason behind this.

When people have a discussion of this, sometimes we want people to understand exactly how this is like. For example, when we switched to the units that is perhaps iPhone 5. I don't know why iPhone 5 is here. It would tell us how many iPhone 5's is this amount of money and so on. When people...

Q1: Calculate maybe iPhone 5...

A: Yeah, exactly. It's just...

Q2: You can calculate the fund, for example.

[laughter]

[crosstalk]

A: Yes. You can use very creative units.

Q1: Can I ask just one question?

A: Sure, yes.

Q1: Can you show, for example, what is the amount of we're paying the debt in this budget?

A: Yeah, sure.

[crosstalk]

Q1: For example, in Italy we have big bubble that is corresponding.

Q3: So, it's possible?

A: It's here actually.

Q1: You can choose several things that you can use to calculate, can you choose here or the iPhone 5 the only option?

A: No. [laughs] The other ones are in Mandarin.

[crosstalk]

A: I will translate for you. These are in order, again, which is just in Taiwan dollars and this is a lunch, how many lunch there is.

Q1: Oh, OK.

A: The average national salary. How many minutes of space travel? [laughs]

Q1: Business trips.

A: Yeah, business trips. How many bubble teas? How many Icelands?

Q2: Shocking gun.

A: And how many Icelands?

Q3: How many Icelands?

A: Yeah. I think they're using the total Iceland's bailing out money as the unit, I think. When Iceland went bankrupt.

Q2: Ah, OK.

A: Yeah.

Q3: Iceland.

[laughter]

A: Yes. As a land. Yes.

Q3: There's no one from Iceland here.

A: [laughs]

Q3: I know how it is.

Q4: Could be of use of you.

Q2: How many ports is that?

A: [laughs] Yes, and you can see this, how many relative size and things like that. To answer your question, the debt is here. It's in the other expenses, which is not that much actually. Also the pension and also this is disaster preparation. This is like a temporary workers actually, which is not so much. It's OK. Yeah.

Q2: So the main issue is it is a multi-level readability. You can adapt to the culture of the reader different level of understanding?

A: Exactly. So if you are a expert, then you debate on this level. But if you're just a lay person, you still have a basic idea to what the budget is about.

Q2: In iPhones.

A: In iPhones. Exactly. [laughs] Exactly in iPhones. Because each of this has a discussion board, people could click on it and say I want more, I want less, and so on and have a Facebook discussion on it.

After one month of people pressing like and saying whatever they liked or disliked about the budgets sitting communist surprise all of them by having all of the office reply to all of the comments directly.

This is amazing. We use the focus conversation method to reply first to the fact questions. Like, I see the stadium being invested by, I don't see the construction what's going on? Then the reflections. Like, this should be more. This should be less. And then ideas, right?

People were very surprised when the city government people replied to them in this way. But the reason why they do that, as I've said, the mayor was a independent. He doesn't have a party. The entire city council is his opposition. When he want to do participatory budgets, by allocating some percent of the city investment budget. The council was against that because there was no party backing the mayor, right?

Then it was seen as they shouldn't have attacked our representative democracy from the direction of direct democracy. The mayor then said, OK. We bypass the council. We now recruit people who can make sense out of this information who made interesting or useful contributions.

We'll send them invitations through their local communities for them to attend classes to educate for them about budgeting. When they get 12 hours of education, they get a credit card. When they get 24 hours, they get additional training on how to talk with civic servants. Civic servants specialties.

If they enroll in this another 12 hours of training, they'll train to become facilitators. How to hold meetings. How to take records. How to do cross sectoral stakeholder analysis. The very basic facilitator training, and they call it Shahid deliberators -- city deliberators.

By the end of the 36-hour training, they get this metro card that has their name on it. Saying that this is the biz-B card. You are now a biz-B of the Taipei city that is like a civil servant but from the civic society.

When they have completed training of maybe 300 people in three different batches of this kind of daily breakers, they now have the same kind of counter-expertise that could rival the budget committee of the city council.

Now, they see the same data and they have more or less the same level of expertise and knowledge. Then the mayor could start doing participatory budgeting. Before that, he doesn't really have the buy-in from the city council.

The city civil servants would be very scared, because then maybe all the budgets gets cut when the participatory budget process runs at the end. It still needs several approval, right? From the city council.

Now, with this threat, or carrot on a stick of saying, you will get bypassed if you don't buy-in into this process. The mayor could now get much more buy-in from the city council, who sees now their role as the leaders of those civic daily breakers.

Now, they're much more aligned in value. But this is because we have people who speak the civil service language or like the council member language through this kind of public education for at least half a year. That's how Taipei does its PB plan.

It's based on the national budget work, like Gov-zero did as its zeros project. I hope I'm making sense. Any questions?

Q2: That means his strengthening are his weak political position through a program that was matching transparency a civic training, in order to create precondition to have a society that could support externally the lack of political support into him?

A: Yes. That's a brilliant analysis. Exactly. Yes.

Q2: So that was because you were there to help him with this?

A: Exactly.

Q2: Without your system, you wouldn't have helped him?

A: Right, because the council would have to approve a budget to buy a system to replace him, which will never happen. [laughs] The fact that we offer this system for free is, of course, for the cause.

Q1: Because we are trying to imagine what kind of things [indecipherable 17:58] can offer. It's attractive because it's for free but at the same time, obliged to a larger level of transparency and responsiveness by the institution than what it exist now?

A: Yeah.

Q1: We understood that also from your perspective the effect of offering for free tempting things. Although this tempting things includes a responsibility due to your transparency, they continue to be attractive.

A: Yes.

Q1: Our discussion is exactly on that, on how we can get attention and try to contribute...

[crosstalk]

A: Also, when Taipei City did this, the [inaudible 00:11] did a southern city analysis immediately that they would also publish their entire PDF in an OpenSpending format to join this platform because they don't want to be seen as less administratively progressive than Taipei City, because they're fighting for the capital position, actually. [laughs]

Q2: All this budget platform relies on the publication of budget data and open data from central and local authorities in all the [inaudible 00:43] ?

A. Exactly, yes.

Q2: I mean that's the end of, when you started to radio of this source of open data?

A: No.

Q1: Or the platform was also able to let's say activate...? Because one problem for example, we met with another project which is like a clean or cozy project which is also funded by the UN, it's called "Open Boundaries," and focuses exactly on providing semantic, blah, blah, blah, blah, for analyzing code.

[crosstalk]

Q1: The point is that it originally relies on the fact that it's based on a whistleblowing principle so that citizens will provide this kind of data bottom-up.

Which somehow it's a mechanism that can be enacted just when there is a certain kind of critical mass, I would say, that pushes others to imitate it with something that exists, but I'm not sure, how does it start? What was the original for that?

A: I'm reminded that I should not write acronyms so I will write Freedom of Information Act. I think most countries have something like that where you can ask your government things and they must publicize it.

The problem is, this Act is read-only. The public, in most civil-tradition countries, meaning that if they give you an A4 paper, or a PDF file that is scanned, the only thing you can do is to read it. You cannot sell it. You cannot change it. You cannot even put it on your website sometimes. All you can do is to read it, maybe aloud to other people, but that's the only thing you can do.

[laughter]

A.: Then, again, what it gives, is information, meaning, it's understandable by people, but not necessarily computers. If they give a very low quality scan, there's nothing you can do with it on computers, without human. The Freedom of Information Act is a start, but it's never strong enough.

In Taiwan, where we're pushing for open data, we make a very, very clear distinction saying, "When you go from public-read only to open read-write, you enable people to make things like this because you can now change the way the data is presented."

You can make tabular data into treemaps. But in the original public Information Act, you cannot do that because it was not licensed using an open definition or a Creative Commons license, that enables remixes and creativities of this kind of thing.

We sell the idea of open data not through data policy, but through the openness that allows people to do the convincing, the translation, the visualization, everything, reporting, storytelling work for the government, for the civil servants.

For the civil servants this is very attractive, because that makes their position much more important than the elected officials, because they are the provider of this information, which then gets converted for free to reach more people. This is the first thing.

Now the next thing as you said, sometimes with whistleblowers or with Freedom of Information Act, all we get is low quality information, but we want to turn it into machine...

Q1: The civil servants in a specific area provide good information, but you cannot [inaudible 04:45] .

A: Exactly, so it's not machine readable. If it's not machine readable in its entirety with context, then it's not really data. You can call it raw data but this is fairly stretching it.

To make information into something that is also machine readable, we rely on the international community like the OpenSpending Community, the OKFN, the [inaudible 05:10] , the usual suspects to define the international formats.

That we say, "If you convert your information to this data definition that's being maintained by 27 countries," which is true, "you automatically get visibility to 27 countries." Because Taiwan is not part of United Nations, it's not part of WIPO, all the elected officials are very interested in getting our visibility anywhere in the world, [laughs] because it was like a hidden country.

Under OKFN where we were, the first place, they had to change this column. This used to say, "Country," but now it just says, "Place." Because of that, that Taiwan is not in the United Nations, they had to change for that multilateralism way of thinking of sovereignty into multi-stakeholderism where Taiwan is 23 million people of a civil society, or some other things like that.

The idea is that elected officials buy into this, because when they publish information in a way that it is also compatible with international data, then they get international visibility like the top space on the OKFN index, which is very good for publicly elected officials.

This way, we convince the public servants and the elected. This way, we convince the elected officials. Together, we change the norm from the public information to open data. Now, any information system that costs less than one million euros are open data Creative Commons by default.

They cannot even argue. They cannot even refuse. As long as this is built in under a million euros of total budget over the past three years, they must be open data in a machine readable format and a Creative Commons compatible license. This completely changes the role of civil servants. I will explain how we get to there in the next slide. Is that OK?

As I said, all the technologies we worked that I showed so far is from the government to the civil society, and asking for the feedback. But the civil society is not satisfied with that. What we really want is agenda-setting power, saying, "What kind of things must the government think about."

This is because the civil society, although it has solidarity linking, whatever, is never getting the same amount of early-stage decision data as the private sector lobbyists are getting within the government.

Maybe individual academicians, scholars, committee members have some representation in those committees, like the Environmental, or Budget or Development committees. But the problem is that they don't have this natural, what we call the "industry chain" connection that the lobbyists have.

A lobbyist in one industry naturally has affinity with their vendors and their customers in industry, so their natural interests are aligned. But our individual committee members don't have this kind of natural chain.

The reason of this is because where they can share a lot of information within the lobbyist network, there's no comparable network for the civil society to share those early-stage decision-making information.

As a result of that the protesters on the street, [laughs] even though they could mobilize a lot of people, they're not really speaking the same language as the lawmakers are speaking. They could escalate however they want, but it's not the same kind of process. This is a general enough graph, I think this applies to pretty much any democratic country on the planet.

One of the ways we turn this around is by Occupy. I imagine all of you know how Occupy works, so I will not explain the Occupy or the hand gestures. The place where we occupy is the Legislative Parliament, the Congress. Why we occupied was that it refused to do its job.

The background which could be very easily explained in one minute is this. In 2014, Taiwan is about to sign a cross-strait trade agreement deal with Beijing. When Beijing, China agrees on this much better than the world trade organization term, they offered, cross-strait, very, very good deals about the service agreement basically giving a semi-domestic monopoly, so to speak to Taiwan-based companies.

Normally, when we sign something like that, with say, New Zealand, or with Japan, or with Portugal, there is a process. The Parliament must hold a hearing. All the impacted industries must send representatives. They debate case by case. They do an impact analysis. This is the same as in any other democratic country.

But, constitutionally, Beijing is part of Taiwan in the Taiwan Constitution, because of a loophole in the Constitution, because the government that occupies Taiwan was the government of China. Constitutionally, they consider Beijing part of Taiwan, Mongol, Mongolia part of Taiwan, Tibet part of Taiwan. [laughs]

Any agreement that we sign with Beijing is like the national government signing a deal with the Taipei City. This is a domestic agreement, and a domestic agreement is administrative business that is nothing of the Parliament's business. Because that's how these things work.

If all the Taipei City or Taichung City budgets must go through the national Parliament, the Parliament doesn't have to do anything else. This is too much for them. Because Beijing is a local government, [laughs] a Taiwan institution, this kind of trade agreement is when the President and Administration want to sign it, the deal doesn't have to go through the legislative.

The legislative, when they sit on it for 90 days it automatically gets passed. There's no way to not sign this agreement. Clearly, this is against the intuition of everybody in Taiwan. But this is part of the Constitution and the Constitution defines the function of the Parliament.

The Parliament cannot really function, other than saying, "OK, we cannot talk about it. We cannot deliberate about it."

By the date of this automatic expiry that it automatically goes into effect, there was this large protest outside of the Parliament building, where I was applying the Internet connectivity for broadcasting. This is what we call the, "0th Sunflower Digital Camp," because this is the first time, as a demonstration, it was not done on the street. It was done in the Parliament building.

The protesters were not doing the usual kind of mobilization, where they were just doing counter-power. They were demonstrating in a demo kind of way, "How should we talk about service agreement like this?" They were doing the legislator's job in the legislative building for 22 days. That was the idea of the Occupy.

A few months after the Occupy, something very similar happened in Hong Kong called the, "Umbrella Revolution," or, "Movement." Again, it's self organized. Again it was, just like in Taipei, caught by the world's media as they occupied this space, which do garbage recycling for themselves.

I was in Dusseldorf, Germany at that time. I was joining with telepresence, typing the message which was projected on the Occupy Central walls in Hong Kong.

Q3: I was in Hong Kong.

A: Really? You saw those projectors, yes. Then, one of the people there tweeted saying the website of the Occupy Central has got to be the most technological advanced in history. Now, one of his friends said, "I have seen this website before."

[inaudible 14:14] , one of the Gov Zero founders, saying ,"Yeah, because that's the Sunflower movement website. They just forked the Gov Zero Sunflower Movement website."

For people who have not seen the website before, this is crowd-source bookmarks, here, everything is crowd-sourced. You can add a bookmark here. This is all the live video feeds that is seen here, and this is the real-time map, so to speak.

This is, "Newsroom," and this is, "Logistics." In finer detail, the highlights show you the medical areas, the barricades, the concentration of police, a wall map, basically, that anybody who plays video games knows how to use. [laughs] This is to tell rumors from facts. Any rumors get triaged by other fields, "Photo," and it's done in a timeline map.

This is also a time maker and again a Gov Zero export so to speak that correlates those rumors and news to the place where they happened. Then, this is all the cameras that people brought in real-time videocasting. You can view four of it, or nine of it, at a time. You can correlate the news with real-time transcripting service that people do next to the videos.

When I was in Germany, I can see that now there are maybe 12 different places occupied. For every Occupy, this is the last update date. They say that they need water, [inaudible 16:01] , sweats, drinks, ponchos, towels. We see these places are being gas-attacked, so they need N95 masks, and things like that. Any new supplies they know where to flow.

This says, "Urgent releases," and what kind of extra supplies they have, so that they can also repurpose the supplies for other nearby Occupy areas, and so on. This is a very useful application, obviously, but this is what Clay Shirky calls a "situational application," a sit-app. Because if you use only Twitter, only Facebook, only Google+, you cannot do movements this way.

This way of doing movements requires the hackers go for Hong Kong, in this case, being on the field, changing the software every day, responding to the need of the occupiers that day, and deploying it in a matter of minutes to everybody on the occupied areas.

Without hackers at the front line of an occupy, these kind of occupations could never happen because Facebook and Twitter, as experienced as they are, they were not designed for Occupy. The Zero's prototype of it happened in the Anti-Nuclear Fourth Plant Protest.

A year before this, in 2013, there was a very large, almost quarter million protest because that was after the Fukushima nuclear plant event in Japan. Everybody was very against nuclear power at that point. Taiwan was doing this fourth nuclear plant, and people went out on the street to say, "We don't want that."

When people were on the street, the news media came, and they found that all the cell phones were down because they could not report on the field, because the 3G network got overloaded with that many people on the street.

None of the real-time media people could send out their footage. Even though there is very many people you can only look at still some next day newspaper, which decreased the bargaining power of the movement.

That year, in 2014, they don't want to repeat the same mistake, so they searched for experts, that is to say Gov Zero people. I'm not a CPR expert, but I know something about software. We worked with the cable power radio experts in the Gov Zero team to a hackathon that we called the, "Parade Hackathon." The Parade Hackathon takes place outdoors.

Then we issued for a 50 megabits line to a nearby building, "We want to broadcast. We expect a quarter million people to come and we will give all the journalists, real-time footage they could stream everything." We had everything planned out.

But now on that day a typhoon came [laughs] and it was raining cats and dogs. Not even quarter a million people, there's maybe 50,000, people or even less. It became a very small kind of show. People were ready to help and it was raining so hard.

We had a very high speed fiber optic line, but we didn't know what to do with it because, people could just use their 3G network, there's not that many people. But then YouTube, just opened its YouTube Live platform a few weeks before that.

We have extra bandwidth and we have a high definition video connection from the stage, where the shows, and the protests, and the speakers, and those bands were playing. We just repurposed that line connecting to my computer with a Thunderbolt port and broadcasted it through YouTube online.

It was not announced because we did not expect that we have bandwidth to do that. But now we have all the extra bandwidth, so we do live broadcasting. Now people feel guilty for not showing up on that day because of the weather, so they spread the news very quickly, as soon as we posted the link to the YouTube broadcasting.

In a matter of minutes, there are more people watching than people are around the stage and because we're just broadcasting the camera, the people watching don't really know that it's just a few people there. It looks like still a very large festival, event and whatever.

People feel like they are in a virtual parade so to speak that still has some kind of influence. We worked on the protocol of how to do this kind of live broadcasting on that day and just 10 days after that the Occupy Parliament happened.

We set up the same kind of gear here and we expect to demonstration for a night and people would go home. But one of the young students, who lent me his laptop, said, "This is my administrator password." This is a laptop of 16, 17 inch a large laptop. He says, "You can use it as your broadcasting station and I'm not going to use my laptop anymore."

I'm like, "This is a university student. He's not going to use his laptop?" But it turns out that he went to the other side of the Parliament building and broke through the window and occupied the Parliament.

What we learned that day is that all the occupiers, there were maybe 50 or so students, they are only allowed to carry MacBook Air. Anything that is heavier than MacBook Air they cannot climb over the walls with it, so they have to leave it down the street. [laughs]

It makes sense, and iPad, of course. When they were in the Parliament building, nobody was expecting it, so there were just a few police. It was very civilized. They don't even have to fight with police, there were no police.

Then the video team that supported this Gov Zero live video, happened to have this very high quality video camera with this long stick, and so on. They covered the entire process of breaking into the Parliament building.

Once they are in the Parliament building, they set up this so called [inaudible 22:32] broadcasting station, which has been broadcasting whatever happens on occupy area for the next three or so days. We have three video sources at that night.

The police soon came and surrounded this place. But because we already have people watching the live video, people want to take buses, take taxis to support and they counter-surrounded the police, and it was like 10 to 1 ratio.

The police dare not move and the people were just counter-surrounding the police, so they knew the police cannot join the other people. It became a very interesting situation, where new people also cannot join the inner occupiers, but they were very interested in participating.

When they do listen to our YouTube views or the students Ustream views we asked people to type whatever they hear into this hacker system that Gov Zero uses all the time. This is like Google Docs. We brought the hacker in immediately. They brought a new cluster in EC2, or something specifically for Gov Zero, because otherwise their other paying customers couldn't use their service.

Then we used this to correlate the transcripts with the translations, which was at the time, 12 different languages. This is basically a media apparatus that is like any other media, but it was built by civil hackers over the first 24 hours.

This is important because the news cycle. We occupied at the night, and the next morning, all the printed papers saying, "These are monsters. There were drunk people. They were breaking things," what mainstream media do. But the agenda-setting power is done by the civic media already before the morning papers were printed.

People see on YouTube with their own hand how the breaking in happened, how there were peaceful negotiations with police, and so on. By the time the morning paper printed, people knew these were lies. Like there were no fights and so on.

Then it became we set up the hack folder, that was the prototype of the Hong Kong system that you just saw and then the same designer who designed the Gov Zero logo, designed the main visual identity in every crowd-source and all the bookmarks.

For the next few days it becomes a war between traditional media and civic media and it's a war on agenda-setting, on virality, on everything. We easily won that war in three days. Because the agenda setting power when you see it with your own eyes, the transcripts are accurate and the transcripts are translated and broadcast overseas, this is a reach that the traditional media just don't have.

On the third day, we run into something that all the occupiers, when they are more than a week or so, run into. It's the spreading of rumors. Because when there are more than one site, for example, here we see people rumoring that the people in the Parliament are being attacked by the police, so they want to escalate the fight with the police, for example.

The leader has to come out and say, "No, this is not actually the case." Because while people could fact check in their phones, the rumors still spread faster than the time it would take to check their phones.

What we did was that we brought our own projector and we set up this temporary projecting screen and we worked with a stenographer in the Parliament building.

She types everything she hears in the Parliament building, which is then broadcasted to the outside wall and the wall of the Parliament building and then brought to the ICT experts this very long Cat-6 line that's 350 meters long, and they deploy this as Intranet to other three different sites of occupation.

It becomes a [inaudible 26:44] so to speak, so that we have sub 10 millisecond view of everything that happens in Parliament and people who don't have the time to check the screen can see the stenographic transcript and say, "That is correct."

It's as if the police that is between the occupied area and the streets don't exist. As I said, you can see straight into the Parliament building, so rumors spread now slower than facts. Because it's such a good idea, the people, in the Parliament buildings soon set up their own projectors that shows the projection on the streets. [laughs] It really just wired in a lot of people.

We provide this as a neutral role because during the occupy aside from the students and the protesters there's a lot of other protesters on the street representing roughly speaking the Separatists from China, independence.

Environmentalists who protest against agreements with ecological impact, and the Leftists, who protests against the delaying of the trade agreements that would cause the loss of jobs or the loss of life quality of labor people.

These three kinds of people are considered also protesters. But the other three kinds of people are considered neutral roles. They are doctors who upheld the right of health, so they will treat police or protester or anything, and lawyers protecting the right of due process.

The Gov Zero people stick a sticker on their shoulder. We call ourselves, "Upholding the constitutional right of communication." We were the communication experts, so anybody there who has anything to say, we are there to support their right to say anything on the Internet.

This is important because it's only with this kind of neutrality that people could trust us that we're actually representing people inside and outside of the wall in a neutral fashion.

This comes to test just two days after we declare ourselves neutral. The training force a bunch of students decides from the occupied Parliament building and the streets, three streets near it to attack also the Administrative building. That's just the next street. This is very dangerous because the Administrative building is the executive power.

The police went and tried to evacuate the people who went there and at the same time as the action which was not pre-announced we got cyber attacked through our crowd-sourced bookmarks and our crowd-sourced transcripts.

It's like a coordinated attack under the infrastructure of communication. Because we use only free tools, because its open-source, because it costs nothing, it took us only one hour to recover another platform. We just changed the CNAME and for the DNS provide it. That's the only time we need to wait.

But on the same time we send people with the WiMAX, that is to say, a high-speed connection, and as many battery packs as we can muster, and an iPad to cover, as a real-time stream, the attack on the administrative building from the occupiers.

Now the police, who went here, and then the students, who went here, behaved very peacefully. Because they know they were being watched by 60,000 people online. There is a counter used for them. They broke the glasses, maybe, there was some shouting or something. But they behaved very civilly.

But on the other side of the administrative building there was police brutality of some degree. It was very, very brutal. That's because there was no camera filming them. On that night we learned that people behave differently [laughs] on the camera versus not on the camera.

Of course we all know that, but we learned it very painfully. We decided then...OK. I still have time...to be deploy not only the stationary cameras, which by that time is in the dozens but they were stationary.

We decided that we need mobile cameras everywhere, around not just the streets but anywhere was any kind of possibility of escalation. We built a website. This is a civic journalism batch generator. All you have to do is to write your name here and upload your photo here.

It will print you a batch of your desired size that will fit your firm or your iPad that declares that you're a journalist. Then we print on the other side of the badge, or the flip side of the badge, this QR code. What is this QR code?

It's a link to a Supreme Court's ruling a few years ago, that says, "The article you love in the Constitution protects not only speech but also news gathering. News gathering is not limited just to professional journalists, but also protects any ordinary person who gathers information with the aim of providing people's news worth information to supervise the job."

Whenever the police stop a civic journalist, we ask them to scan the QR code and read the Supreme Court's decision. That says, "Any area where the mainstream media can enter, the civic journalist who printed this badge must be able to allow to enter. Otherwise we will take you all the way to the Supreme Court."

They assented because this decision was done in an unambiguous way. Everybody voted for it, all the Justices. The thing is that with kind of civic journalism badge, suddenly we have dozens of mobile news feeds on the ground.

After that day, there is no injury, no fighting, nobody missing. It became a totally non-violent protesting. Now when people have become non-violent, we can actually do some deliberation.

Beginning at March, 29 we went public this eight months old project from Gov Zero, called, "Are you Affected by the Cross-Faith's Agreement." This is a website where you can be with your mobile phone. You can enter the trade you're in.

Maybe you're in the IT industry or you can enter your company name. Then you will show crosslinking to the company registration database the five trades your company's working on. Then clicking on the kind of work that you do, it will show in a three-panel comic how exactly does the CSSTA affect you.

It will show you a male in Chinese coming by person or just the money, whether you can also do it to some province or the entire mill in China and how much impact would have on your neighboring industries. Or if you are not in that service industry, it will show you that you're not affected by the CSSTA.

Instead of reading through hundreds of pages of pdf files, which was the wrong term we're working with, we correlated it with registration date. Was the UN data I was WTO data, was the male in Chinese laws to show everybody in five seconds how exactly do they affect them.

To show the support of the occupiers on the street, because on that day when we started the deliberation in place of legislatures, the president said we do not acknowledge the result of this deliberation.

We do not think, even though you can convince thousands of students, they are representative students in anyway. Half a million people showed up on the street of Taipei and says this is not right. We must deliberate and the administration digitalization must accept the result.

Then, what do we do with half a million people on the street? We group them. Again, according to existing streets into that independents, separatists to that deliberation which talk about in sovereignty or the relationship with China of the Chair Agreement.

Then the ecological, the green people talked about the ecological of the land, the farm land and whatever impact. Lefties people talked about the labor rights and so on. Our own ICT people started using Lumia, which is specifically designed for the occupy situation application, where people could share in their local area network how to reach consensus.

One of our topics here is how do we ICT volunteers from people who are just here to use this neutral, fast lane to get a very good view of the parliaments. Because GOV Zero people, because our logo is creative common zero, anybody can print it and put them on a badge and says, "We are a radio-powered technology expert so we are neutral, so please get us in."

But what they really want to do is to check in on Facebook. Basically, we need to tell the ICT, fun and nice ICT people. There were a lot of proposals being proposed on Lumia. Starting from very stupid ideas. We could ask for their ID card or something like that, or credentials, or something.

We would ask them to get on hub and commit to prove or something like that. None of this are very practical in this kind of on-the-field setting. The beauty of Lumia is that we can have multiple stage of straw polls.

Whenever a new idea come, we can do another strange of poll until everybody agree, converge on the consensus. Our final consensus was that whenever a new person shows up with the Gov Zero badge saying they are of the ICT team, we ask them what is 2^16.

If they can answer this mathematical question, 2^16, they're probably a geek. If they're not, they are probably not really the ICT team. 2^16, that was the consensus. It was really effective.

Using the same technology, we captured the deliberation that was happening on the street in a different part of CSSTA. That was our first encounter with the people doing deliberative democracy or participatory democracy and to lend ICT as their cause.

That is how we met those people and that's how Gov Zero, as a whole, gained a whole new dimension. We want to set a mediation space where everybody could trust us and for the private sector to sit down with the civil sector about the thing that government's not having the entire agenda-setting power.

I think it's time for another 10-minute break. Are people OK with this story? Any ideas, thoughts, comments?

Q1: I actually have a question.

A: Yes.

Q1: I was thinking about the participating budget and the website. I am from Turkey. It's not a comparison with being how many millions of people we have in Istanbul during the evening traffic.

Still, I was thinking about the other question that you really need to have a critical mass to be able to infect the politicians. My question is how many people follow this website? Is it really active in the society at large?

A: Yes, you mean the PB website? The budget in Taipei?

Q1: Any other Gov Zero website?

A: Gov Zero website, yes. We have had a million people on the street. They show up, but it's easily 10 times this number online. This is because everybody is concerned about the CSSTA.

Now, the Taipei budget, I don't have the members. But in a Ballpark, the day when the mayor, he said something very interesting. It was that without this kind of educational tour, NEPB is just populism? Which is true, because people are voting without knowing what they are voting for.

Disqus closed it on the national media. On the first day, we get, I think, almost a quarter million people who joined this new cycle went very viral because...OK.

Q2: Someone want to knock and a big transparent... [hammering sounds]

A: Yes, exactly. Yeah, but I cannot speak Morse code so we will need an interpreter. In any case, to answer your question directly. The number of people who hear about Mayor Chu, the Taipei City mayor, is a lot nationally because there is a tradition in Taiwan that for the mayor of the capital to become the next president.

Mayor Chu was unique because he have never done politics before. He was a professional surgeon. He was generally seen not a leftist. He was generally seen as somebody who things Singapore is a really good idea. Modern, efficient, you know?

Something a surgeon would like. Clean, and things like that. With this comes your partnership of participatory budgeting. This is something very, very important. Because he contradicted himself just a week ago. That's another things he likes to do.

He would say something. Then a week after that, or even a day after that, he will say, "I was wrong yesterday. Sorry." He admits publicly he is on the autism spectrum as an Asperger's person. It's very natural for him to just speak the facts.

His child was diagnosed as an autist. He has this special quality of not feeling any shame or losing face, of saying, "OK. I was wrong. Now I think this way." Which is great, for a direct democracy.

Then, what he said essentially was this. Democracy is very young in Taiwan. But we can either roll back and go back to the authoritarian single rule model, or we can go forward. We can bring it to the 21st century and do something that nobody in Asia has done.

Then for participatory budgeting he buys into this idea, exactly because nobody else in Asia is doing PB this way, as the type of he's doing it. Based on open data, direct participation, bypassing the council entirely. Having a counter-expertise to the council. Things like building his own council, basically, and things like that.

He sees it as a democratic innovation. If a potential future presidential candidate says that, it gets national attention very quickly. I think everybody in Taiwan knows about this. But people check at his website, only maybe people living in the north region of Taiwan.

Which is slightly fewer, but it's still a large number of people.

Q3: [Inaudible 13:03 to 13":08] which is this idea that...we were able to put cameras on certain side of the school. The number of violence decreased automatically once transparency was projected in the public space.

The public was pretty specific and transparent. That's something that I was thinking about. I don't know. I was seeing, for example, in Turkey. Last year we had a lot of public demonstration, a lot of cameras on the street.

Still, the level of violence on the people who was protesting was really strong and really high. I'm wondering, in what measured way...I think, the way in which you frame transparency and attribute the center of political background, entrenched in the concept of transparency, is not completely legal trial.

With respect to the effect that transparency has on the society, where we would implement it? That implement, I don't know if I am completely on this point.

A: Yes. There is two levels. There is transparency and there is reflectivity. Because the important thing is not the camera, it's the projector. It's the thing that people donated two-story high projectors installed on the street.

Everybody sees everything that is taken by the on-field camera in real-time, as if they're tele-present in the same space. The important thing is, not the nation sees the street. The important thing is, the street sees itself through a mirror, knowing that the nation is seeing it. Do you know what I'm talking about?

Q3: Yeah.

A: Basically, this is a Truman's Show, so to speak. We make the space itself a space of mirrors, a reflective space. Without this kind of settings, with just cameras, people are not being made aware that they are inside a panopticon.

Q3: I've seen them on that.

A: We have an entire social computing theory about this.

Q3: An example of the differences. I think, even if the [inaudible 15:34] could be used...it always depends where the poverty relation is. What you said to Don. I was thinking, for example, the rally of Donald Trump. The other day, they were clearly knowing that there were cameras on a position.

They used this situation to draw black people away from public space where Donald Trump was speaking. It was a clear political message to the people that was attending, and to the people at home. It was.

A: Yes. I have more slides that talk about that.

Q3: [inaudible 16:08]

A: [laughter] Yes. Exactly. But yes, I agree with this analysis. This is completely true.

Q3: I'm not saying that you asked transparency as an neutral plan. It is just that its been...

A: Yeah, exactly. Which is why this is not enough. This setting is not enough. Because if the private sector interests are much larger, then what I draw here. Then, the civil society...this becomes a message of exclusion.

Because this becomes what the political analysis are talking. You are here. Everybody can be here. But it doesn't really matter, because you're still a minority. We evolve this idea in my next slide. Any other thoughts, ideas, questions, comments, tweets?

Q3: We have some people at home, they're now asking questions. We can ask them instead.

Q4: If someone at home wants to ask questions, you can...

A: You can do so.

Q$: ...you can do it.

A: One of the relay people will relay for you. Now, evolving this idea. After the Occupy, the politics in Taiwan changed completely. After the Occupy, and the Occupy Central, there was this reelection of the city level government.

The Nationalist Party, which was the dictatorship party, morphed slightly into a democratic party, lost completely in that election. I'm glad to say, all the legislator voting guys, and everything that I mentioned, maybe played a part in this.

A lot of mayors from the Progressive Party in the independence, were surprised they were elected. Their numbers went like 10 percent more than they predicted, because a lot of the Nationalist supporters just refused to vote.

After that landslide victory of the counter-power, the ruling cabinet, the prime-minister resigned immediately. Then a new prime-minister came, still working for the Nationalist Party.

But knowing that he only has one year, until the election cycle that changed the president and the entire administration and legislation on a national level. When he was tasked with directing the country's agenda, when he has only one year left and he has essentially two bosses.

The president that's going down with an approval rate of nine percent, and the president's that coming up controlling all the cities — but not yet the national parliament or national administration. The new prime minister is in a very difficult position. But he is an engineer.

The day he went on office he was quoted saying, "An engineer has no right to say no to a technical problem." Which is very good mentality, which I probably had to agree with. He then set a completely new agenda for the next year that will not alienate any of these two bosses.

The three agenda are open data, crowdsourcing and big data, because they are infrastructure. All these three things are infrastructure. No political elected official could be against that. Plus, it is the same agenda, the Gov Zero people has ever been advocating about. They know they have natural allies.

He appointed the vice prime minister, a Simon Chang, an ex-Google engineer. He know something about data. Then, Simon has recruited Jaclyn Tsai, a legal expert, an ex-judge who worked for IBM Asia Head of the Legal Department for a few years.

Again, she knows something about the ICT industry and the legal things about it. It becomes, for the first time in Taiwan's history, a cabinet lead by engineers, or technocrats if you are against them [laughs] . They identify for the next year the topics they want to talk about, that is neutral to any of their left or right or whatever. It has to happen.

They started with the open data definition. Then, an equity-based crowd funding, and closely held LLCs. Then, taxation, privacy and de-identification, security, tele-work, tele-medicine. Again, none of this is partisan. Whatever ideology you have, this is the bridge between the civil servants and the rest of the Internet. That become the new agenda of the year.

The first thing they want to do is to talk with people doing this kind of mobilization on the Internet and saying, "OK now we finally want to tribute as a peer, as like face to face. How do we use the social media rules for law making? Because we know that you can mobilize any number of people."

We cannot say, "How about we start talking about these kinds of things?'' But the problem is that the government-initiated deliberation is completely different from an occupier-initiated deliberation. The occupiers notice that the people on this are already interested topic. Otherwise, they won't show up.

But the government don't really know who the stakeholders are. The stakeholders don't know what the government is celebrating about. If they set up a website, people don't know how to actually comment effectively, even they have all the open data and things like that. They are often trolls, they are often practice an attack ad hominems.

But even if you have a really good moderator, people always still face the information overload. Because, there is a lot of three-letter acronyms in all of those legal code and things like that. It is like coding. When you change one line, you change the entire system.

But if you are not a legal expert, you don't know how the entire system will get changed. You see a very difficult problem. They lost the election here, and then...a few days after that they brought some Gov Zero people, and said, "We now want to start to e-deliberation, the way you do in some of our movement."

Then we select a topic that concern all of you. It is called the teleworking for the start-up company that don't have a physical office address. Which is a good topic. This is a topic that obviously concerns everybody on the Internet.

They say the way administration do things is to hold a public hearing. When they do a public hearing, they want representatives from the associations, from the guilds and from the labor unions. That's how they do things.

They ask us, "Is there any union for all the teleworkers in Taiwan, who could speak as a representatives for everybody who work?" This is ridiculous because a programmer who work at home and a musician working at home, and a designer working at home, is completely a different kind of trade.

Nobody would dare to speak for other kind of people, because the entire work flow, the labor and the company relations is completely different. Nobody would stand and say, "I represent all the teleworkers."

They change codes they say, "But how about the companies, early stage starts-ups who only have an email address and maybe they go on Kickstarter or some other you know confounding website and registered their company in Cayman Islands, and they'll have no physical address. Is there an association of such early stages starters and representatives?"

Of course there are not. They have problem paying their next month salary, why would they form associations? This is ridiculous. Then, they were faced with a problem because if they ask only the startup entrepreneurs and the teleworkers they know, everybody else will say this is lobbying, this is just closed deal with the people who are so close to administration, and so on.

They were very afraid that they would get occupied again if they do that. It is always this sword hanging over their head, around that side in Taiwan. If they do the deliberation wrong, they will get Occupy again.

They need a different solution. They want to talk with people who already register at Cayman Island or work for a company register in Cayman Islands, and find those people and ask what kind of loss of teleworking, and of startups, their company's loss will we take for you to register in Taiwan. Why are you registering in Cayman Islands, and why are you working at home?

They want to reach everybody who would do that. This is an engineering problem. The first thing is that Jaclyn joined the g0v hackathon as a civilian. We call her Jaclyn, not minister. Then, she took three minutes, just like everybody else, showing this design saying, “We want to reach people who register their companies in Cayman Island."

So this is an engineering problem. We need one editor, a few engineers and so on. It's played by g0v rules. Then, we gather around the white board, using open space technology and work on the website that will enable this kind of operation. That was in December.

It took actually only a month for us to build this system and this slide was our first case. The company law change. The way we did this was modeled after IETF in the early 90's, in the previous century. This is saying, "We've held a mail list, a public forum and everybody can join as long as you have an email address. Then, we welcome discussions."

But we are not saying anything about law. We are asking for stakeholders to identify themselves. Are you a startup lawyer? Are you an accountant? Do you have something to share? Any experience that you share is very much appreciated.

When people share something useful or contribution, they get marked as being valuable. We invite those people to the administration building and hold after-hours meetings. We are not wearing a suit or anything. But it's in the administration building with the ministries of economy, finance and something. With the leading scholars of civil law and case law, and all the local government people working on a registration of companies, to talk with those people.

Anyone who contributed on the mail list in a constructive fashion were invited to attend. Then, it was done via the same technology as the civil right movement. It was captured in real time, transcripts broadcasters outside of walls where they were.

The way this works is now we have identifies all the concerns everybody have. Then, we ask people who make contributions to form a working group. The working group is responsible to produce a document that's called, "Request for Comments." That is a request or commentary on the closely held companies.

It use the language of the IETF saying, "If Taiwan want to make a law on the closely held companies, like the Cayman Islands, it MUST allow multiple votes per share for example. It MUST NOT limit the crowdfunding venues and it SHOULD, for example, allow a telepresence shareholder meeting."

Then, it SHOULD NOT, MAY or MAY NOT something... Because that is language the Internet engineering task force. We produce this and then we send it to the ministries to translate it to legalize, to legal pair. They agree when they translate to crosslink back to this suggestion.

This is the first bill in Taiwan where every line was annotated with the demand or specification where it came from. They become our coder and we are like project managers. They have to identify the specification-implementation link.

The suggestion were then linked to the specific points in the video feed when and where people throughout this topic up. After this deliberation process, when the two parties are in the parliament filibustering each other, they couldn't pass anything at that time.

The season ends in June the 3rd. This is the only bill where none of the parties want to block. Because first, all their parties have participated already in the working group. Unless they discover new facts that were not covered by the working group, they don't have any rights to say we don't have consensus, we have consensus.

If they block this bill they were against the entire staff health community and all the teleworkers who paid attention, thousands of people who view this public consultation deliberation. They are not to block this, it was passed in record time and signed to effect and so now Taiwan has this kind of law.

A: To go into detail because you are experts in this. We deploy the Focused Conversation Methods We identify all the speech and our email is to this course.

We are separated into the facts, the objective layer, the feeling of those objective facts and the suggestions resulting of the feelings. We use a font with six different ways of the font showing the strengths of consensus.

Just in a glance, you can see the overall strengths of different options that has the support. We talk about one aspect of the standard law. Every slide using this pen and paper technology...well, a screen technology which is then broadcasted online. People will participate online.

We take 20 minutes to talk with face to face with the group members, and then we shift to here and then see what people online have to say. We give them also 20 minutes to set the agenda for this particular slide, particular topic, so they can express their consents, their worries or their total support with catch smileys, [laughs] and things like that.

These are all different non-verbal kind of communication going on. We have a professional deliberator online, Lu Chia-Hua, who takes care of that online part and carries it to the offline space. I was there in this for the term of offline space, and was using the centered technology to broadcast it to people who attend on the remote, so you don't have to be inside the city.

You're guaranteed the same amount of representation to working group meetings. So, this is very effective, and when the whole country is seeing this either from a recording or from mainstream media reporting, there is no denying seeing that there really is consensus being formed in this therapeutic process.

That is how we run the working group meetings. And again, every word, every sentence was captured on say its platform, so you can do a link to one specific utterance within context with the mainstream media people, if the system works. They don't have to think of a very contentious title or topic, or something. They just copy and paste, and make reports out of it.

The design principle of the system is this. To reduce the ignorance problem, or the ministries who propose things, like the Ministry of Labor, of Economy, they must first do a slide on SlideShare that explains the problems they meant very clearly, and we have amateurs trying to read it and make sure.

Then for every keyword "startup," closely hold up a cooperation teller work. We asked for 140-letter definitions because most the online deliberation is wasted on fighting over the definition of keywords.

When we start to define them, the way the dictionary does when you hover it, you see a definition. It saves 80 percent of people's time because now we say, "We know 'startup' means different things to different people but for the sake of deliberation it means exactly this."

When we do the initial stakeholder interview and agenda-setting, we make sure that one representative from elective official office, from Jacqueline's office, and one from public servants, one from the Information Industry Association representing the private sector and one from the ground zero civil society people and at least, four people, but sometimes five, six, or seven people.

These people are the people who built this website. We set the term of use together, the license together, everything together. When we do interview of the stakeholder and some agenda-setting, we lend each other kind of legitimacy that any other kind of initiation will not pass because everybody knows that other agenda-setting are being done in balance with each other sector's ideas.

We ask people's opinions. We say, "We're not asking you to vote, we don't do voting." What we are asking is an agenda for the face-to-face hearing. When we're running a public hearing on this, we promise we will only talk about the things that you propose to us online, where you reach a consensus on.

This completely changes the dynamic of the ICP of the virtual part of the deliberation. It's responsible for the objective, that is the fact, and the reflections, that is the feeling parts after deliberation process. We lift the idea interpretational part and then decisional part to the face-to-face but we use the e-forums or whatever to collect as wide as possible.

When you bring a useful contributor idea, then you are invited into the working group who do the idea and the decision. This is a self-selecting process.

This is a very geeky point but I always want to emphasize this. It's a safe space, meaning that when people propose their responses online, sometimes you see 10 sentences, nine of which is very useful at this closure as the useful contribution but one sentence is an attack, either to the agenda or to other people.

As a moderator, you're faced with a dilemma. If you censor this, if you kill these comments, people will say that you're jerking in or something. If you keep this, the next reply is not going to reply to the civilized part. It's going to reply to this one sentence that's attacking people.

Then you will get another reply that is 50 percent attack and then the next reply is going to be 80 percent toxic and the next reply is going to be a cat picture and then [laughs] once you reached the point of cat picture there is no return. [laughs] So the challenge then is to how you construct a safe space where you can have a discussion before the cat pictures arrive.

The way we do this is by editing the comments when anybody say, "Oh we showed very clear term of views, code of conduct" and say upfront your comments when they say "A harm attack" when there's toxic it will be deleted. But we delete only that part of the sentence, we keep everything else.

We delete may be four words and then we send a private message to that person saying, "You're a violation of our term of conduct you say, "Public servants are just pigs wasting taxpayers time. This is not OK, this is not constructive." We deleted that but we kept your other very useful sensible contributions.

Because it's version controlled, everybody who is so willing to dumpster diving can [laughs] see the original comments. But for most people who enter the first part of discussion and reply to those civilized points.

Q1: So you keep memory of all the original comments but you give like visibility to...

[crosstalk]

Q2: Even if you want to see if you're using just comment there...

[crosstalk]

A: Yeah, it says edited and it says version three so you can go back to version two and version one if you have too much time. [laughs] Yeah, exactly...

Q2: But for the troll was important.

A: Yeah, but because since people seem dis-proportionally interested in this, I will go into details. Trolls on the Internet are just people who crave for attention that they cannot get from the real life. Right and so from people who...

Q1: [laughs] I have a problem in your interference...

[laughter]

A: The important thing is that you must hug the trolls knowing that even if they say 10 sentences, nine of which are trolling, just one sentence is useful or it was not actually but you can read it peacefully, you ignore those parts that was toxic you maybe delete or moderated out but you may respond very enthusiastically to the part that was constructive, because they are just people craving attention.

They learn with seeing one or two exchange that bringing something useful to the table was the only way to get attention and because they crave attention, they now think of ways to do constructive work and so we reform the trolls very quickly with systems like that by giving them due attention to just a part of it that was of benefits to the community.

The safe space was not just a moderation thing it was also acculturation. We bring a culture of civilized discussion saying, "If you crave attention, this is the only way you're going to get it." I hope I'm making sense.

Q1: You said that most of the interaction you bring in the person online is really based on the idea that the...functioning which is literally online is based in the center and you will have the let's say find the development of the idea and the decision in person.

A: Exactly. Yes but we use their presence to extend what "in person" means of course. So that was the principle.

Q2: So you never vote neither in person?

A: We do straw votes the way have votes. Basically it's a vote that is never binding. It is a way to see what people feels and people can change their votes anytime. So this is not really voting, right.

Q2: Wait you're talking online or...?

A: Online.

Q2: When you're in person you're voting for the final decision or not and in case the idea as Fork, there are competing ideas which are not easy to be used.

A: The idea of rough consensus is that the deliberation never ends until we have consensus. So if we don't have consensus, we say we leave it to the next administration. We don't even vote. Yeah, that's the idea.

Q2: I've got a question online are there examples in our communication spaces of these more months of approach dealing with toxicity and trolling you?

A: Are there more examples to deal with trolls?

Q2: Toxics nowadays and trolling?

A: Yeah, sure.

[crosstalk]

Q2: ...is based off of this.

A: Yes. Yes. I see that. I'll show one about Uber in the next slide.

Q2: Show in the next slide. There is always a next slide with the scoops.

A: Yes. But I want to stress the idea about safe space. Because safe space means usually that anything that's violating the code of conduct is not tolerated, right? It means zero toleration. There is a difference between a safe space enforced by people, and a safe space enforced by algorithms.

If you have an algorithmic make safe space, the trolls don't play attrition game with mediators. Whereas before...like in Wikipedia. A Wikipedia editor, when it fights the trolls, or the revert words, or something, the operators must put in exactly the same time as the people doing the vandalism. Even with the help with bots.

With robots, like automated tools. Because the vandals also have automated tools. Right? So you have to put in, when you're playing by the same rule, exactly the same amount of attention. If the trolls outnumber the moderators, then they dominate the attention from other people nearby.

It could only be controlled, when the mediator proves a disproportionate amount of time and attention, so that it overrides all the vandals and all the trolls. What our innovation, I would like to say, is that we watch what the moderators do, and then we turn them into code. That is to say, we automate this kind of process of mediation of moderation.

So that any number of trolls are now faced with robo-mediators. [laughs] So, they cannot waste people's time. When trolls see that they're not even wasting the operator's time, they'll lose interest. Because there are much more interesting ways of wasting other people's time.

Grieving, that's how they call it, other parts in the Internet. So they go to those other parts of the Internet. I'm going to show you an example with another automated system in the Uber case. Thanks for the question. Let's go on, because we're almost...

Q1: Six.

A: ...at six, yes. So with the vTaiwan system, we spent a lot of time saying no to the ministries' proposals. Because the ministries always want to talk about the things, that they fuel like hot potatoes. Because they're civil servants. And civil servants in Taiwan are in this very unenviable position, as a new democracy.

If they do anything wrong, because we don't have an anonymous civil servants culture like in the UK, they get blamed for it publicly. If they do something right, the elected officials get all the credit. So the thing is they're in a very powerless position. With things like vTaiwan, public servants feel very much empowered.

Because they offset their responsibilities, but they gain credibility, because they interact meaningfully with netizens as part of deliberation thing. They bring us a lot of very tricky issues, like gay marriage. They say, OK. Let's have a national deliberation out on gay marriage. And we say, no. We must keep saying no to these kinds of thing.

Because vTaiwan was designed like a town hall. It must affect only people who are netizens as the main stakeholders. We're like a small town, of network-using people, negotiating with the government. This is how we get our legitimacy. Otherwise people would say it's just a technocrat, the elites doing the deliberation.

Who are so good at typing at a keyboard, or using a pen and pencil online, or something. Or showing up with telecommunication, deciding the fate of people who are not so good at this sort of thing. We don't want that. So we must ask all the ministries to prove a very high correlation between the people they are going to affect, and the people that's on the Internet all the time, the netizens.

Because the Ministry of Justice cannot prove there is a strong correlation between going on the Internet and being gay, [laughs] we don't do that case. This is very important actually. Then it must be something that is codifiable. Because otherwise it doesn't really work.

Q1: So this exclusion was not motivated by what related to call issue, related to minority and majority? We are not going to discuss an issue over minority in front of a majority, but what's more related to the internal coherence of the system of deliberation?

A: Exactly. Because, what we are saying, essentially, is that for the parliament, we're getting all the stakeholders on a multi-stakeholder dialogue, that gets all the facts, and all the reflections, and give a recommendation. If this stakeholder does not represent everybody that's going to be actually having a stake, because those people don't use the Internet, then we'll lose this kind of legitimacy.

Then the parliament could very easily find a representative from other fields of life. Saying, "But you're missing their voice, because they're not on the Internet." Then we lose the entire basis of this legitimacy, of this kind of system. I hope I'm making that...

Q2: The issues that sometimes stakeholders or sometimes there is some day when you get involved in the stake and in the discussion, and what is...?

[crosstalk]

A: Yes. Exactly.

Q2: Do you think, progressively, are you able, since you use for example a typical netizen, to start using a technology. This can include...

A: More and more people.

Q2: ...once it shows its effectiveness, a possibility to actually influence the public relation?

A: Yes. That's three slides.

Q2: Three slides.

[laughter]

Q2: Little distracted there. I think that someone is going to set us live...

[crosstalk]

[laughter]

A: Yeah. I'm very happy actually we get into so much detail. Because in Paris it was like one minute per slide. Even though it's taking up everybody's time, I think it's important...

[crosstalk]

Q1: ...space in this country. These are ideas that serve our worst.

[crosstalk]

A: That's great. That's great. Then our other requirement is that the ministry must not send us drafts of any laws or regulations. Because by that time the window of opportunities is so close. It's like we're building a nuclear plant, but the color of the wall is up for deliberation. [laughs] It doesn't make sense, right, by that stage.

What we ask is, at least, even if you have a draft, don't show it. Show your problem analysis. Show your stakeholder analysis. Show your cost/benefit analysis, whatever. For 30 days refrain from making any suggestions as public servants. You now only serve the fact-finding team, the online discussion.

After 30 days you're allowed to show your drafts. Everybody in the working group is also allowed to show their drafts, so we can have a competing version of the drafts. But if the ministries show their draft first, before we have 30 days to get everybody initiating this topic, then they have an unbalanced power in setting the agenda.

But after 30 days, when everybody knows about these things, more or less, their alternative draft is going to read at least coherent as the ministry drafts.

Then it's possible to bring them together. Otherwise it's impossible. It's just the public sector dominating the agenda setting. That's the power analysis with it, on that hackathon, for this system to work. And it really worked. We have one million visits per month. We have a lot of subscribers.

As you could see, just like in Wikipedia or Flashmob, or anything, this gets progressively down. Like, 10 percent, 10 percent, 10 percent. But this is normal. Yes?

Q1: I was thinking, this rule that you're describing now, has a meaning, for example, in the planning system of Tuscany. Traditionally, you had a project, and then you worked with the analysis to justify the project. The analyses were following the project actually.

In many cases of law happens this, you have a political, ideological idea. Then you do analysis to justify your idea. The fact of inverting, or obliging people to show the analysis, and then each one by its own think about the solution, then you can compare the solution after a certain time, inverts the way in which usually the laws are conceived.

A: Exactly. This is actually very much empowering the civil servants, but it's decreasing the power of elected officials and parliament members. We used to do all the initial setting of the jotted law, because they have think pank parties, and so on, who know the initial context of doing this thing. So this is weakening your mandate.

Because of three independent, non-nationalistic party engineers running the country, it is not a problem for Taiwan at that particular point of time. This kind of opportunity doesn't really happen all the time everywhere. Yeah. And we have a very good net promotive score. Meaning that people want to recommend their friends to go on vTaiwan.

Q1: Do you have any idea about...? For example, you have 30,000 subscribers, and 2,000 commenters. Do you have anything to check the degree of attention of the subscribers? Because the risk is, subscribers aren't a very good number to sell. Then you can imagine that they are completely silent, in the sense that they're not even reading.

[crosstalk]

A: Yes. That's the next slide.

[laughter]

A: It literally is the next slide.

[crosstalk]

Q2: ...and we got that in the next slide.

A: Let me just go to the next slide. As a trainer, this slide I can spend an hour on. But we don't really have an hour to spend on this one slide, so I will talk in very brief terms. This is all, either free software, or at least free of charge software. Or free under 10,000 users, or something like that. So this system is zero cost, is what I'm saying. It costs nothing to set this up.

To answer your question, we use Mailjet, to send a monthly newsletter to our subscribers. It includes a tracking pixel that tells me how much time they've spent reading the email. Whether they actually click the links. Whether they engage, in what kind of topics in the newsletter. We do actually AB testing.

I can talk for hours, but you do get that idea. Then we do surveys and registrations using Typeform, which is a very mobile-friendly kind of choice-making or voting platform. We vote for things to talk about, so we don't vote for things.

Q1: So you say that you have a strong analytic spot, in order to have an idea of how many subscribers are not just formally subscribing, but they...?

[crosstalk]

A: Yes. Then we know that people want to talk about Uber, and then Airbnb, and then Bitcoin. Then we know people do not want to talk about e-voting or something like that. This is because Typeform has an interesting thing called...a randomized...this is like a survey technology, right?

So we show people in random order, and ask them to pick what people think they're interested in, and things like that. We correlate that with the contributions they've made before, and the time they've spent on the website, and so on. So that we know, for absolute certainty really, when we talk about Uber, how many people will come in the first day?

Within maybe 10 percent of margin of error. Without this data analytics, we would not launch deliberation plan, is what I'm speaking. So this is Typeform. Then the forum software that we use a lot...people use that now actually. The Etalab in France use that. It's called Discourse.

The company start-up, who produced this open source tool, is called The Civilized Discourse Construction Kit Company. This is optimized to have a civilized discussion. Where you can edit people's comment. Where people who contribute frequently automatically are promoted to moderators. They get badges.

When you register, for the first three days, you can approach anything with pictures. You know, it has very, very tiny rules like this. But taken together, it enforces civility in a way that doesn't cause attrition on the moderator's slide. We asked all the ministries to publish their statement of analysis on SlideShare as PowerPoint slides.

We ask them always to use a sans-serif black font. The font weight must be more than 300. We put a restriction on the fonts, because people in Taiwan used to use fonts that are more calligraphic. That looks very good on paper, but very bad on mobile devices. That's how they always used to do things.

They do, I think, Times New Roman. Times New Roman will look very bad on a deliberation space online like this, when people are looking at a film. We even have a style guide for the ministries to publish their deliberation material. Then all the issues, agenda and settings were captured on GitBook, which is a way for people to write Markdown, to publish structured data online like a book.

Where people can download in PDF or EPUB, or something like that. Any three-letter acronyms in GitBook are defined in a Google spreadsheet. With the keyword, the English translation, the description, the cross-referencing, or else. All this are published using this free publication platform, called GitHub Pages.

Which costs nothing, and scales to hundreds to thousands of viewers, without us paying anything. Then we have YouTube, of course, to keep all the video records, and real-time interactions. We use Labhouse Inc., which experiments with us virtual reality recording techniques.

We're not really using it in the early stages of deliberations now, but we have a lot of pilot projects. Where in this room, they just throw a spherical camera up here, mounting here. So it captures everybody's number of expressions in the same time. It's called, that colony is called TripMoment. We do a lot of pilot projects.

We've done six like that. It changed the quality of the discussion. Because when you see it from the real-time, it's just like you're there on the ceiling, really. Then you can see what everybody's face looks like when I speak a sentence. So it feels less like, do your trick performance.

It puts less incentive to people to speak to the camera, and try to make a very good impression, because they know they're only going to be capturing this square. But it creates an incentive for people who really get counseling and understanding from people actually around the table, because the spherical camera is capturing this also.

So if everybody is rolling their eyes, people will see that. That is another important method. Of course we have stenographers, who published the transcript in real-time, that's the Sunflower technology. Hackpad plus SayIt. Then I explained that we use the FCM, the focus conversation method. That says we explore all the facts, before we ask people's feelings.

We ask all the feelings, before we ask for ideas. We ask for the ideas, before we do a decision. It must be done in this sequence, because ideas are sticky. People are smart. and once people have an idea, they ignore other people's feelings. We don't do that. We strictly follow this.

We follow this so strictly now, then when the ministry proposed their problem analysis, we asked them to rephrase it in an ORD form, and in this sequence. What facts do the ministry know? What feelings do the ministry have about the facts? And what ideas they have, and only in this sequence.

This is our tool kit, so to speak. I talked about Police, the robotic facilitator system as my last talk. But this slide is our entire toolbox, of the every kind of deliberation system. This is written on to mend. Many of this methodology is done by Cornell University Regulation of project. We improved on their methodology.

They had professors and students, post-PhDs as mediators. But we have ministries with their own user names on the forum software. We created a rule, that's showing on the log-in, saying, anybody who tags...that is to say, who press X Then after that, for example, MoF, meaning Ministry of Finance.

Or MoEA, Ministry of Economic Affairs. Any ministry that gets mentioned this way will get an email, saying, "You're getting mentioned." By the second they're getting the email a countdown starts. Within seven days they must provide an official public reply to anything that mentions their username.

This changes the dynamic completely. Whereas before they could reply privately, they could stall, they could say, "This loses face so we ignore it," we say, "If you don't do this the platform will not do deliberation on any of your cases," so in order to join you must do this.

With a four week deliberation period, that means four exchanges with anybody's concerns about fact-finding about that ministry. For many of ministries they really just print this for an email, and then they sign it, and with the initial idea they send it for approval for the head of the ministry, with a pen and paper and so on, having other people type it into the forum.

This is still OK. We allow seven days for this entire paper-based process to go through. The important thing is that it is still public. They cannot go retract their words.

They were forced to admit a lot of shortcomings that were not possible, if they don't have this guarantee. Whereas, in Cornell, the synthesizers are the professors, here the synthesizers, the working groups, are just anybody who makes an active contribution, IETF style.

Then again, any offline collaboration that we meet face-to-face, even the preparatory meetings are kept on record and published online. Any online forum, system, software, whatever, are printed to pen and paper to the ministry's archives. This is, anything that happens in one space, happens also in another space, in as much fidelity as possible.

This is the design principle that we adopt on top of the regular sharing of knowledge. Yes, we are expanding it to the general public. [laughs] This is because the National Development Council thought this such a good idea. They really want to talk about gay marriage. [laughs] They set up another website called, "Join Government Taiwan," that is government run.

I don't have time to go into a lot of details. But the innovation is where they have a toll-free number, where you can call and have somebody read to you the top concerns currently trending on that website.

You can tell the operator over the telephone your input. They will type it to the forum software and, "If you'll please leave your mobile phone or telephone number, when the government will reply to you, they will call you back." [laughs] This expands netizens to people with a telephone, which is a lot more people. [laughs]

Q1: It's a little bit more costly also, I guess.

A: Right. Gov-Zero, we are volunteers so we can only run things with zero cost, but this is the government speaking so they can hire telephone operators.

Q1: Call center.

A: A call center, exactly. [laughs] Yeah, a governmental call center. We only do early consultation. But because it's government, they also do e-petition, which is something we cannot do because they will request an empowerment from the administration obviously. They also do that, because they extend for anybody with a telephone. I think I'm of the last generation who draw telephone this way. [laughs]

Q1: That's my generation.

A: Yes. My childhood is also this.

Q1: My teenager had your childhood.

A: Exactly, yes. [laughs] Yeah, because we expand our work to anybody with this kind of phone, they now think they have sufficient legitimacy to talk about gay marriage, because everybody has a phone or something like that. Without going into too much detail, yes, this is being adopted as a national agenda by the National Development Council.

Q1: Why this temporary government?

A: No, no, no by the next government.

Q1: Oh the next...

A: This is at the end...

Q1: The temporary government you're talking about ended? It was replaced?

A: Well, the election happened in January, that is two months ago but it does not...The legislators are already in office now and the Nationalists kept 30 percent, 20 something percent, negligible amounts of the seats.

Then the equivalent of Podemos got five people in the Parliament, which is great, and then the Progressive Party, the main counterparty to the Nationalist government gets 60 percent or something of the Parliament seats.

It completely changes, it flips the seats. The new President of the DPP, she was elected but she's not in the office until May, May the 20th, because of a constitutional loophole. We have a lot of those things. [laughs]

Now we're in a waiting cabinet, so to speak who now answer to the new legislators but is not answering yet to the president that has already been elected. For those three months, this is a lot of fun. The Civil Engineer training them in IT, Mao Chi-kuo, he resigned as all the cabinets must do after a national level election.

But the new prime minister was the vice prime minister, so we have a Google engineer now as the prime minister and then the prime minister knows he only has three months of time.

He says, "OK, I will work with the new government administration team to transfer everything. But I have a precondition. Any meeting that the new administration is having with us, I promise not to destroy any records, any archives, but in exchange I will only do the transfer of power and explanation of the current ministry agenda and everything under a direct [inaudible 06:41] ."

He is basically transferring the power not to the opposition party but to everybody, so it means that everybody now has the same knowledge as the new president and they can now oversee the new president in a way.

But he could only do that because, we, Taiwan ratified this open leader policy, this open government policy saying that anything that the government does must be transparent in a reasonable way.

He writes on this saying, "OK, there must be no secrets between the party that's losing power and the party that's gaining power. We are now transferring the power to the general public," which is great.

It will take three months to do that. Join.gov.tw and [inaudible 07:23] Taiwan, are a very important part of it because then it allows technical topics to be debated or to be shown to the national populace. That's the policy-making from now on has a deliberative democratic spirit and attitude.

This is also because the new President, Miss Tsai Ing-wen, run on campaign platform that says, "Open source policy-making and maker spirit." To make good on her campaign that she also must agree to this way of transferring power, otherwise her platform means nothing, so that's a lot of fun.

To end my talk, which I will take maybe another 15 minutes, unless people are so interested, we'll take another half an hour, I will talk about something that is not domestic. The vetoing process or the joint process that we design after that, with me as an advisor, are only really good for domestic issues.

With Uber even if we get all the representatives, all the stakeholders, they sit down they agree on something, Uber shouldn't care, Uber wouldn't care. They don't even have a physical operating center in Taiwan. If the consensus is to shut down their business...Their business is already shut down, they had to find elsewhere.

The thing is that this kind of thing really is domestic and we are facing what the scholars say, "Post-democracy entities," where they could ignore the entire rule about democracy altogether, because they operate on a completely different domain of human behavior.

Because they are not part of any physical place, they ignore the multi-lateralism part of sovereignty. Uber basically is a symbol of a lobbying power that transcends sovereign power. I will make just one very quick example.

In New York when the mayor is not even deliberating, starting to think about a law that will limit the quantity of Uber charges, Uber introduced a new button in their app in New York. It used to be UberPOP, UberBLACK, and UberX as some people know of something about that. You can call different kind of things, even a helicopter in New York but...

Q1: Even a helicopter?

A: Yeah, even a helicopter and a boat in Amsterdam. In any case...

Q1: [inaudible 09:58] .

A: Yeah, so Uber introduced a button that says, "Uber with the mayor's proposed laws." If you call Uber, it's usually saying you must wait for eight minutes or five minutes. But if you slide to that button, it shows 50 minutes, one hour, five hours.

It's a very effective lobbying tool. It's telling all the users, "If the mayor's law passes," whether it's true or not, "then you will wait forever for Uber and this is not acceptable." From a social computing perspective, this is genius, this is wonderful operation.

But from a deliberative democracy viewpoint, this is a nightmare. We now have a non-answerable to sovereign power entity, who could engage or mobilize much more people than public servants could do. All the celebrities were retweeting this because it was sensational and the mayor has to retract even the idea of talking about this law.

Uber is Uber. It's super. [laughs] Here are the places. The green means they were legal in those cities. The red, meaning they're illegal, but they're operating anyway. The pink, meaning that they're controversial, meaning that they're being debated, they're of questionable legality.

Q1: In some cases it's at the state level and some at the city?

A: Exactly. When we did a poll to all the B-Power members, of button-up topics, not ministry topics, people want to talk about, that doesn't have a ministry support yet, Uber is at the very top. Any time we run this, Uber is at the very top. For Uber we had to redesign the entire flow of deliberation, because we know the existing process doesn't work with this kind of entity.

We identified the Ministry of Transport, who at that time has been fining Uber for violating the law, for over a million euros, by that time.

Uber says, "We are not under the jurisdiction. We will take it all the way to the Supreme Court," which they did, "to fight saying, 'We are really an economy, share-riding, ride-sharing, whatever, platform, so, we are under the jurisdiction of the Ministry of Economy. It's not a transport problem.'"

Then the Ministry of Economy does not want to propose their problem analysis, because their problem analysis doesn't agree with the one that our Transport Ministry is saying publicly. They were afraid of losing face. They will feel like one ministry fighting against another.

Then Jaclyn, the Minister who went on Hackathon, says what she really cares about is taxing and maybe insurance and this belongs to a Minister of Finance. But the Minister of Finance says, "We don't have the expertise. Indeed, we don't have the interest to do a detailed problem analysis. We think this is the job of the Economic and Transport ministries."

I think, Taiwan is not alone. Everywhere in the world we have the same dynamic. The three ministries playing sometimes against each other's interests, for challenges like Uber. [laughs] This is what we did then is we think professional mediators are needed. A mediation space is no longer enough.

We must have somebody who connects directly to all the taxi fleets, who already surrounded the Ministry of Transport, as other taxis in other countries did and also Uber itself, and Uber's competitors, Lyft, whatever. Then, through them, we should reach the individual, like limousine drivers who are in the Civil Society, but they don't have an association, which is why they are adopted [inaudible 14:22] .

Then, again, some independent drivers have an association, and we have to include them in the deliberation process, because outside of the fleet, they may want to join Uber, and they may want to join the taxi fleet, so we have them in the association.

Through the association we also reach the local governments policy-makers, like the cities where Uber operates in and also the other for-hire companies that's already locally operating somewhat like Uber, but in a much more smaller scale, like the mom-and-pop shops of ride-sharing, ride-sharing companies. Again, they don't have a representative, which is why they are adopted.

Our power analysis of this challenge, is then to find people who the Association would trust, who the taxi fleet would trust, and the Finance Ministry would trust and for those mediators to talk to each other, and to design a deliberative process that will please, at the same time, everybody who connects with them and gets the buy-in from the civil servants from the three ministries.

That took us two months. It's very, very difficult. But we did it. We designed a process that will allow us to reach all these people at the same time. Basically, all the drivers in Taiwan was the stakeholders. Then all the passengers of whatever taxi fleet, ride-sharing, Uber, whatever, who are also stakeholders.

We must prove that we can reach all of them before this deliberation starts to have a comparable mobilization power, versus Uber itself. Because it could very easily reach this amount of people, like in New York City. We'd have to reach at least this amount of people, and then more. The process we did for this is, again, we cross the agenda.

We use this idea, called, "overlapping consensus." That says, "People with different ideologies, they could never agree on ideologies. But if we make the issue specific enough, they would agree on the practicalities." I think this is well-known in the academic circle by now. We choose a specific.

It says, "What about, you don't have a professional driver's license, but you carry somebody as part of your driving to their wanted destination, and you charge them for it?" That's it. That is our issue. It says nothing about Uber. Not about sharing economy. Not about transport. Anything. It's just a very specific thing that anybody could do, and people could have that consensus on.

That is a very tiny slice of the entire Uber challenge. Then we are saying, "Everything that we collect from the people on this deliberating process for a month, will be published for independent analysis, as open data." We don't do any analysis, because no ministries are willing to do an analysis. Instead we ask what people feel about this thing.

We ask what they think about this thing. We record everything. We publish everything as open data. We ask people from the academic community that they're not scientists, Uber themselves, whoever, look at this data and tell us what your analysis is. That becomes the agenda, which we guarantee a deliberation amongst from that day.

This is the interface that we came up with. This is the designed for drivers, of course, and people on the taxi. They don't have a minute of time, which was the Facebook limit, of the old voting process. They don't have one minute. They have a red light, that's maybe 10 seconds. We did say, "No voting while driving." [laughs]

The idea was that they could park to the street a little bit, and spend maybe 15 seconds on our deliberation interface. This interface only demands 10 seconds of their time, and then they could start riding again.

This interface is very simple. On your phone, which you will get this link on the same hour of the day, in one specific afternoon, we announce the URL to all the mediators, to those different interest groups, so they get the URL at the same time.

Then they spread this URL through whatever channel they have, telephone, SMS, line, Uber has its own, you know, closed route of instant messaging and the things like that, Facebook, whatever. But the point is that on that specific day, when they go on this place, they see people in four or five different groups already. They don't feel overwhelmed by only one kind of people dominating the discussion.

When they go on it, they see one simple sentiment. This is picked at random from a poll. They just click, "Agree," or, "Disagree," and that's it. When you click agree or disagree, you see your avatar moves.

Q2: It's agree or disagree with friends with a sentence?

A: Yes.

Q2: Now with their belonging to the group?

A: No.

Q2: OK, because the group is based on the sentence.

A: Yeah, we have three sentences that says, "I have a professional driver license," and, "I have a driver's license," and then, "I have taken Uber before," so, you can say, "Yes," or, "No," on these, and...

Q2: The groups are created on the cross of the three questions?

A: No, the groups are created automatically based on any number of questions. This is like a survey, except all the statements aside from the nine, which we prepare initially, when you tap nine times, you will have answered all the initial questions, then you will feel that your interests is not being reflected.

We ask, "What do you think? What do you feel?" Then you can say, "OK, I feel passenger liability is actually very important and you didn't actually mention it in any of my questions." This becomes then the thing for other people to vote on, actually. Any sentiments that they write must begin with, "I feel," or, "I think," which is the same word in Mandarin.

They become then things for other people to agree or consider on. When you agree or disagree on something, you see your position gradually change based on your answers. If you login with Facebook or Twitter, you don't have to, or email, then you have the capability of writing new sentiments. When you login with a social account, you see your friends.

Otherwise, you'll see people like celebrities, or people with more followers on Twitter instead. But the point is, first, you see at a glance that there are four different groups with very, very different views. Your friends are among all those groups, so these are not your enemies, these are your Facebook friends. It's just you don't talk about Uber usually.

Now for a different time, you know how they feel about...It creates a non-antagonistic relationship. Second, you see people's positions can change. Third, we use algorithm it's called, "Dimensional Reduction," so in a nine yes or no question, this is a point in a nine dimension space.

People are really grouped in a nine dimension space, but we don't usually think in nine dimensions. As more people contribute, we are now in a 40 dimensional space, which is a very large dimension.

But when people have a natural grouping on some of the most divisive, that is to say, this is the most controversial [inaudible 22:46] , some questions will naturally group people who answer yes or no that would become the determinant of their other answers to other questions. Those are called "primary factors."

The algorithm could identify the two primary factors that's currently going on. Using the primary factor as the X axis, and the second primary factor as the Y axis, so that you can see on your mobile phone, how people really are grouped on those two most divisive issues. Am I making sense? This is technical but...

[crosstalk]

Q1: How many questions can it take, this kind of system?

A: Seven, it takes seven to start to do the grouping. Then it could scale to dozens of questions.

Q1: Is there not another...?

A: There is not a limit. But there is a limit...

Q1: But still it focuses on the two most controversial opinions for asking questions?

A: Yeah. The thing is that once you answer, because you only start to show up in the group after answering seven random questions, so by that time, you probably already know what has been determined before. Especially, you can click into it to see the current consensus.

People will only write when they have new sentiments to contribute that they feel is not being reflected on this space. This is Open Space Technology with a reflective projection, but it's carried online. This is our simulation of the sunflower reflective space, but on the online space.

Whereas we were occupying, people who are leftists joined these streams of deliberation, people who are ecological greenists joined the stream of deliberation. They vote with their feet but it was still reflected through this projection, and then captured online. We're trying to recreate this kind of experience on a mobile phone, basically, that requires only 10 seconds of [inaudible 24:45] .

As they are doing this, the system rewards consensus by only showing within each group the highest agreement. People are motivated to convince their neighbors. At any given point, we can run a generalized linear regression, or you can run a primary factor analysis, or run any kind of those geeky words on this data to find why people are voting the way they are voting.

In the first week, we had four groups. Uber drivers, taxi fleets, Uber passengers, other passengers. Then we've seen the group. They now try to convince each other because we show, again, only the top ones with consensus.

Group one initially was under [laughs] consensus saying, "They are criminals. We should cancel their registration immediately." The other group was united under this sentiment, "When I'm not in a hurry, even if taxis are passing in front of me, I will call an Uber." As you can see, it's very polarized.

Initially in the first week, people couldn't really agree on anything. But they agree that the other side is their enemy. But when you do a multiplication, these are minorities. They convince nobody on the other three groups, and even within the group, when you [inaudible 26:17] the group they represent as an overall population, they are not even 50 percent of the entire population.

When you click into the majority opinion tab, you see nothing. There is no majority opinion. Because people compete for the intra-group agreement, they start to propose sentiments that are more moderate.

Group one after five days converged saying, "This is not about Uber. The Minister of Transport and Communication should fine any unlicensed personal vehicles. The fact that they are only fining Uber and not other ride-shares may be a problem. But this is not about Uber, it's just them doing their job." This is getting much more consensus in that group.

In group two, invented this kind of thinking, saying, "Many of the large city taxis are joining in the taxi fleet, because the fleet are capitalistic, they have advertisement, and they have other ways to earn money than the taxicab. There is economic pressure.

"Now, Uber is a way for an independent driver to join many fleets. They increase their labor union bargaining power against the taxi fleet." An ingenious argument.

[laughs] When they proposed this argument, not only did they honor the support of everybody in the Uber sympathizers, but they gained two percent of popular. Some taxi drivers jumped groups [laughs] just by seeing this sentiment.

Q1: It was not that they had to be in one fleet only?

A: Yes, exactly. Like, "This makes some sense." [laughs] They compete for their intra group. Still, when you time this, it's still not a majority. This is barely 50 percent.

Now the first majority opinion appears on the second week. That is also the week where the four groups become two groups. The Uber drivers convinced their passengers mostly, and then the taxi fleets convinced the non-Uber passengers mostly.

This one, everybody agrees, "Laws and regulations should progress." But this is so general. You cannot really act on this. But this is a true feeling that the regulations are not set in stone. We show, of the 400 or so people that have seen it at the time, most have agreed. Now, once the first majority opinion appears, people now compete for the majority opinion ratio.

They try to improve on this score. People now compete like this, "Review is important. We must balance the interest of riders and drivers, but safety is the most important." Who would disagree with that? [laughs] They get even higher score on the scoreboard. That's the second week.

By the third week, we have a winner. This is the highest score that anybody has ever got. This is from Alvin of Mozilla Taiwan, a Firefox developer. [inaudible 29:31] .

He said, "The government should leverage this opportunity to challenge the taxi industry to introduce the same five star rating system that Uber has to guarantee quality. Because now if all the taxi cabs, independent or in a fleet, must answer to the same five star rating systems, where the rider can also rate their passengers, then we'll get a very high quality," and so on.

Everybody agrees. It is a very good idea. [laughs] Then This opinion which was like 60 percent for in the first week saying, "We don't argue with criminals," basically started declining on the third week. On the third week more people than not saying, "OK, even if they are criminals in my opinion we should still sit down and have a deliberation with them."

That number only increases. That is to say, more people are in favor of deliberation on the third week. On the fourth week, we start getting real suggestions.

Our cut-off point is 80 percent consensus in any group, so it has to convince at least four people out of five in all the groups. These suggestions survived this test and most of them only appear on the fourth week when people already are done with competing from majority feelings.

Now, it says whatever the law draws up, it shouldn't be because of Uber, and the taxation, you must have a good story about taxation. This is a new thing.

All the UberX things must register. They must have a U or something on their windshield, knowing that they're operating for Uber. When inside it, you must check the photo, and number and so on, so that you know it is the same person driving than the person that's showing on the phone. This is also very sensible.

The other sentiment is saying transport is like food and medicine, so it makes sense to be more stringent, because it is a matter of not only economy but public safety. Everybody agrees.

People are saying some people say private-persons' vehicle when they do ride sharing, they should not be taxed. It's OK for them to not be taxed, but if they do that, they should be limited to be two shifts a day, meaning that I go to work, I go back to work, I ride people, ride-sharing platform. Why not?

It should be limited in numbers so that if you want to evade tax, you can do it as part of business, right? Even if you do that, passenger insurance should be mandatory and the insurance, company should take this kind of insurance. Finally, people should be able to join multiple fleets. If you join Uber, we must not [inaudible 2:22] you from joining a type of taxi or other fleets.

As you can see, this is very reasonable. This is actionable suggestion. Then we use those suggestions to do a comparative analysis internationally. Then say, for suggestion one through six, this is how other countries are doing it. We show it to everybody before deliberation, and now, we have the deliberation.

We show the consensus sentiments. We show the consensus suggestion, expectations, and the six criteria. And we show it like a progress bar. We try to extract promise out of everybody who showed up. This is the scholar's ministry. The three ministries, Uber -- Uber Hong Kong, Uber Asia and also Association of Private Drivers, the fleets and so on.

They all sit down and we look at the consensus. Then we ask everybody, "Do you agree with those consensus? Are you willing to make compromise, accord of those consensus?" For example, the Insurance policy. The CEO of Uber Taiwan said, "We will help people claiming their damage if they are by the Uber."

I asked, "How many claims that's happened before in the two years or one year and a half?" [inaudible 3:48] said, "There hasn't been any cases." Somebody asks, "OK, so what is the insurance terms? Can you show us the terms?" The Association leader says, well, he doesn't have it with him today.

But then their lawyer said, "OK, just after this deliberation meeting we will send it to Jacqueline's office, our exact terms of private insurance, so that is checked." We asked the minister of transport, "What kind of legal basis there are on to find Uber?"

Then the taxis fleets saying, "If we are allowed to do surge pricing, that is to say, a higher end of taxi cab, we can compete with Uber on their own terms." The minister says, 'OK, we will deregulate that so you can compete with Uber on that term."

Then the Independent Driver Association says, "The reason we are not talking with Uber is that they take 20 percent of the cut. If they are willing to lower to two percent, we will start driving for Uber tomorrow." This becomes a matter of negotiation so we extract promises.

At the end everybody in the nation sees that Uber is checking four of the six marks but it's not willing yet to register a local company that pays the local tax and the insurance and because of that, the ministry is still fining them. If a local government agrees to take on that registration, it could be made legal.

Everybody knows why it's not yet legal and there is no need to fight because they are not fulfilling the consensus of all the drivers and passengers in Taiwan. This is what we call an empowered sense.

This is why we say all the ministries agree saying, "We share our early stage effects and reflections. Each of us agreed that the agendas that empower does not belong exclusively to us."

Then we bridge it out to this cross-sectoral space where we know beforehand, the private sector and the civil society will enter on equal terms on the same day and then we empower the space to decide the political Uber in Taiwan.

This is how we redefine the deliberation process for Uber. The very good thing about this is that the initial nine questions, "Who is involved?" "What do they want to know?" "How shall we respond?" is then carried verbatim for Airbnb.

The Airbnb people has been watching Uber, real-time transcripts, everything from the very beginning. We didn't know that. [laughs] They know that we're using a timeline based on Wikipedia.

Airbnb sympathizer has edited the Wikipedia page that we're about to use [laughs] to reflect better about Airbnb. They sent email to everybody that has used Airbnb before in Taiwan because they have their email address saying, "Go Airbnb Taiwan and place your opinions in support of Airbnb to keep us legal in Taiwan."

When you see three groups of people, all these people responded after the Airbnb's call and there are three groups of people. One-third says it has to satisfy for checkmarks.

Another group says, "At least we should ensure that when people say 'It's their home' it's really their home. One must not have 10 homes in the type of city renting on Airbnb, always identical photo that is not bed and breakfast."

Group three says, "The government, stay out of it," but they are in the minority. All these groups of people, they came at this point in time. Before that, we have pretty good consensus of people who haven't used the Airbnb before.

But once the Airbnb sent email to all their members, [laughs] we got the explosive growth in participation. This is why we don't call ourselves a voting platform, because if we call ourselves a voting platform we would have lost at this point, because it is so unbalanced representation.

Because we say we are just a reflection and objective fact-gathering platform, we were able to say, "OK, we know there's actually not that many people are using Airbnb in Taiwan so let's assume that this is their consensus degree." People have not used should have at least, the same importance and this is their consensus.

We were able to show it side-by-side, not to determine by the sheer numbers but by common sense. [laughs] The main contention, regardless of whether they have used Airbnb or not is that they're in favor of doing Airbnb sharer but not a tenant, because a tenant doesn't have insurance but a landlord has insurance according to Airbnb regulations.

That's the main contention. Their reflections are generally positive and their expectations are generally positive. Airbnb is not a troublemaker because they are really just another evolution of Agoda, or booking.com or something but still those three points are important and they would still get a comparative analysis of the same format.

Now, we have the hostel association, the youth hotel association and Airbnb, Hong Kong Airbnb, Korea Airbnb, co-founder even flew to Taiwan and the scholars, the administration and so on. The magical thing is that the Airbnb people have said, "You know, send things to our members. We are a company who respects our members."

We see this is their consensus so every point that I show on the screen, this set, we have discussed this and we agreed. We agreed that how long will be the first place where we work with the authorities.

I'm going to be showing that no people rent 10 identical houses. We will ensure this kind of tenant insurance. We agreed with everything, in short.

There's nothing for deliberation. By the end of the deliberation, the hotel chain association who was actually protesting before, the deliberation process and went up into the Airbnb.

It was the book of all the thousands of legal hotels in Taiwan and say, "OK, I see you agree with everything so I think of you as a good guy. Please help us because Agoda and booking.com takes 15 percent cut. I understand that you only take 5 percent, even after taxes, so please help us to bring all your hotels on your platform."

That is how the deliberation ended. This is because we made the example out of Uber. It is like a checkbox that it doesn't fit so it is not legal. Airbnb played by the rules. Of course, we thank our contributors. That is the last light.

The ideal after just Airbnb is, of course, try to think of not as sectors of people specializing in the public-private or a civic sector, but think of a fluid role where people when they share with money, they become the private sector, if they share and volunteer their time, they become civic.

They're texting [inaudible 11:14] government, that they can play many roles, even within a day, even within a deliberation because the space were designed by robotic algorithm mediators who don't care where you're coming from, as long as you can make convince each other of their consensus.

This is how we transform people who could only like and unlike, to share our links which takes maybe 10 seconds, to do questions and answers takes one minute and to do real discussions takes five minutes, and deliberation takes an hour or so. Finally, if they make contributions, we invite them to a face-to-face agenda-setting meeting, where they spend two hours with us.

This is my last slide. Thank you so much.

[applause]

[crosstalk]

A: No, it's fine. What about this kind of process? Any other...

[crosstalk]

Q1: It's not easy to answer because there are so many things to digest in terms of the small variation into any other. The general structure seemed to make sense but the few knowledge, for example, I have of the tools that you use in each one of these spaces so the tools are tracing the cluster does not allow me, at least -- I don't know about the others -- to judge the feasibility, for example, in a case here.

It convinced me the idea of starting with feelings and slowly arriving. I mean, the timeline you set for the second, the third and the fourth week.

Then the deliberation space, to make totally sense in relation to the progressive opening of people to the switch of people from a different [inaudible 13:07] to the other. Obviously, there are so many questions in my mind that I'm not even able to formulate them.

Q2: Probably it could be used. That plus one combination you use in that kind of setting, which is program [inaudible 13:25] deliberation.

It would be interesting to test it in a local setting and [inaudible 13:33] . That's something, in my opinion, what is really interesting is that it is on one hand, it is not gamified. [inaudible 13:46] people...

[crosstalk]

Q2: It also takes you to think a little bit. It's not so pure gamification. You are not a fake person that gain points and the [inaudible 13:55] fake identity.

It sounds to me what we were talking about, the kind of serious game that takes you in the direction, this could the same. Find a nice way to attract people, at least at in the beginning. Create a stream of opinions and after that apply more classical maintenance...

[crosstalk]

A: Yes, exactly.

Q1: ...role of informations.

[crosstalk]

Q1: I'll also make free the...

[crosstalk]

A: Thanks for staying so late and also across the Internet. [laughs]

[crosstalk]

Q1: Maybe the people on the other side of the Internet, they are eating away. They are not doing it so...

Q2: [laughs]

[crosstalk]

Q2: We had 11 people last time we checked.

A: For municipalities, it will depend a lot on virtual reality, which is another three-hour workshop but not for today. [laughs]

Q1: No. It's possible that the positive result you had is driven on the -- what's it called -- the emotions that are catching around these two issues. The existence of tensions in society, if I think of how many times I got to the station in Portugal and I found the taxi driver protesting against Uber.

You have also to think about how much the issue you're proposing intersects the interest of the general population, which can happen in a specific moment. My question is, these methods seem also very posed.

My question is, you have the setting then now you already tested in some cases. Is it possible to decide to apply from one day to the other, if you have the very hot issue, a hot topic around this that you did in some time?

These two issues were [inaudible 16:26] being discussed in society. A sort of instant poll. I don't know, a fire disco. l'm thinking about a story that happened in Romania that involved the government to...

[crosstalk]

A: We use this tool not in a national empowered space but as a civil society, as a test-drive long before it is Uber or Airbnb. We did a death penalty deliberation with this after a random killing in Taipei.

Taiwan is one of the last countries to still have the death penalty and a lot of people...

Q1: Last country among the democratic one because the undemocratic ones... [laughs]

[audio cuts]

Q2: Do you release also all the data you collect?

A: Yeah, of course.

Q1: They do not release in an open forum.

A: Yeah, of course.

Q2: You find a place that you would...

[crosstalk]

A: Yes, because otherwise we cannot run an independent analysis. That is a basic...

[crosstalk]

Q2: [inaudible 17:31] analysis.

A: Yes, that's right.

Q2: Was the [inaudible 17:36] ?

A: Right, as you can see this differs from the usual the Taiwan ministry proposed because we asked ministry to provide their initial analysis. For topics like this, nobody is willing to be the one to do the initial analysis so we asked the entire [inaudible 17:47] community.

Q1: Yes, because in Uber case, it's not the level of violence between cab driver and the Uber driver.

[crosstalk]

A: Exactly.

Q2: Your rival was...

[crosstalk]

A: Right, so if the ministry used one wrong word they'd get occupied tomorrow.

Q1: I remember in Milan, [inaudible 18:10] month the vice director of Uber Italy, after a conference and they start hanging things outside the home of this person. [laughs]

[inaudible 18:21] It was a [inaudible 18:24] on your map. Legal after that.

[crosstalk]

A: After we did the Uber deliberation, there is still the Supreme Court ruling and things like that, but we don't see people on the street anymore because people now know what are left to be done. Before that's done, the MOTC, the Ministry of Transport kept fining and people generally think it's a good idea. That's it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment