joejag · December 14, 2015 23:28
diff --git a/jez-value-notes.txt b/jez-value-notes.txt
 http://www.youtube.com/watch?v=ESOaDiv3lXA

 Was not a product manager, read about Scrum & product management in it.
 Scrum says: work out what to build, prioritize & estimate, give feedback shows
 Other: Competiive analysis, pricing
 No longer a product manager

 It's super easy. Once you have an idea, you can get loads of stories. (priortized and esitmated, then actually build it).

 Came up with a product over 6 months, got 2 months feedback. Next two years a blur.
 Scaled agile exists: Assume a person is in charge of the product.

 Agile manifesto: What is value? No satisfactory answer. The one thing you can be certain about is that the PO will be wrong about what is valuable. This is what happens in reality.

 What is value? Shareholder value (from theory of the firm), maximise profits? CEOs are legally bound to do so, or they can/will be sued. Firms that focus on profit end up being less profitable than companies that do not.

 Roger Martin - fixing the game: 8 fold increase in CEO compensation on decreasing rate of return. Clear who the winners and losers are here!

 Elon Musk (found of paypal, made a bunch of money). He then made SpaceX, first private docking to ISS, sent cargo up there in less than 10 years. "I want to put people on Mars". Also on the side made Tesla motors, set out to prove that electric vehicles could be awesome. He's a millionaire, we can't all be Elon. Then again he just got divorced again recently.

 Jake Andraker - 15 years old - made a diagnostic for detecting pancrecatic cancer. Costs a few cence to build, way more accurate. He was told "go figure it out yourself". He got really into the scientific method.

 Jack Welch - Chairman of GE - says about sharelhler alue = dumbest idea in the world. Your main constituencies are your employees, customers and products.

 How to create value for customer - (read book on lean IT). "Users are unable to articulate exactly what they need, yet they seem insistent about what they don't want....once they see it.". Biggest problem is Scrum, build then say "look at me!" then get told "that's crap!". We don't know what's possible, new stuff is happening all the time. We don't know!

 We are building complex systems, you can't plan ahead, you have to try experiments. Saying you are going to plan all this stuff then build it is majorly flawed.

 Along comes Eric Reis: have a vision (universally true, could be called a goal), make a business model, build MVP, test/iterate, pivot when you reach a local maximum. Goal isn't money, it's a means. Funding proposals are miserable to write for. First thing you do is test the biz model. Most expensive way to prove a hypothesis is to build it. He released after 6 months despite bugs, despite being scared, and no one bothered to download it. Could of got the same result with a static html page with a 404 to the downlaod link! Work out how to test model, then build that. Not a new idea, but he's packaged it well. It's really hard to do this and put your money where your mouth is.

 Optimize your process: Build Measure Learn loop. Optimise for lead time for the loop. People normally optimise for busyness. Very easy to show with queining theory that optimising for 100% utilisation is inefficient. Image airport: checking in, loads of desks with one person, massive queue! If there's loads of people, much quicker, but the people won't be fully utilised. YOu need a bunch of people who aren't fully utilised to get good throughput.

 Easiest way to optimise lead time: reduce utilisation (put slack into the systems), reduce batch size (wip, less people queining at the desks), work in cross-functional teams (sit together, no other groups). This fundamentally challenges the concepts we have about projects. Book by Don Reinetson on the prinicples of product flow.

 There is a problem! Tradiontional focused people say it sounds like crap. Engineering should be like buildings! they don't fall down and kill. Barcelona Gaudie building. He studied flowers, all people did straight at the time. He used parabolic forms and hyperbolic structures. Very radical, people were very worried. It isn't a waterfall process. He was building models all the time. Most building of bridges is Trust bridge, put it into a model and get a bridge. If you are building something new then you have to experiment and iterative thinking. you want to optimise for doing this fast.

 Other people who don't behave in this way. Apple! People don't associate them with agile. Apple's first product was the Apple 1. Built in a shed in 76, It was really advanced for it's time, but consumers didn't want it. Very minimal, very viable. Macintosh changed to being a consumer product. They didn't argue about software ideas, they tried them,and kept the ones that worked.

 Where is the customer? Gaudie's church still isn't finished. He said "My customer is not in a hurry". Most of us don't have that luxery!

 Early Apple stuff was built for themselves, they were their own customer (or Steve Jobs). We have develop empathy for our customers. 

 An exercise is to make personas on a target (primary consumer is 1 persona, then 2, then 3). People never want one "they want to dominate the sector". They deliberatley make the target big so it can't be put away! PO said they can't relate to his primary persona, was asked "how many of these widgets are you buying? none. Why should I listen to you? That's the area of your process I'm still having trouble with.". 

 Naturally we want to build for ourselves. If it isn't you have to emphasise with the people you are builing it. you have to find a way to measure the value you are delivering to them. This is where you apply the scientific method. 
 Tehniques: A/B testing (junk mail isn't all the same), show proto to real users, measure biz metrics, surverys/social media. nigeria 401ers have delbieratly got worse via A/B. They only want people that are massively gullabie. So deliberatly make it less gramatical. Google Analyrics do A/B tests for free out of the box.
 If you can't use real world. Get NDAs, film people interacting with you product. Once a user has it, they can give you great feedback.
 SAAS, measure biz value. Etsy: Measure capacity, conversion rate, measure each feature on conversion rates.

 POs should be working out what to measure value, then work with the team on how they are going to do that.

 Grockie defintion of done: "user stories were not considered done until they led to validated learning". Was it a good idea to do this? then you are done. One of your acceptance criteria should be "how can we measure if this was a good idea or not?"

 CaseStudy: Votizen: A/B got to local maximum. Pivoted biz idea to sustain biz through revenue. Your need an idea and a bunch of ways to get there. But you can't pivot too much as it becomes too much work to understand. At least with the data you can make rational decisios about if you are there or not, have we made it?

 True measure of value (for novels: no one wants to read your shit): will people use it? Using a product is a transaction as the user donates their time and attention which are valuable commodities which you the creator must five something worthy of that gift. This is worth bearing in mind while building stuff.

 the point isn;t the money, it's about acheving your vision. If not one is going to pay or use it it doesn't matter. If managment buy it but the users don't use it (it doesn't matter). Classic example are expenses apps: if users are mandated to use it (IDM) then you have to, so all your measurements are going to be off.

 The PM role: What is their role? Create & Communicate share vision. Work with team to define measurements. Run experiments to learn if your ideas are any good. Optimise your process for cycle time. Quality is everybody's responsibility.
 Everyone has to be excited about what your doing, and they are contributing to it (book:drive). Use scientific methods to build/measure/learn.

 Lifecycle for product dev (from the Don): 3 stages: inception, mvp, delivery.
 inception: get together, define vision, biz model, decide on mvp to test the idea
 deliver: make mvp, get feedback
 keep delivering small increments
 What's missing? Creating an enourmous prioritized backlog. Fundamentally waseful, as it estimation (book:how to measure anything) when you look at products over their lifecycle, cost has a very low info value, in terms of what you should build. the high value info is the  value something delivers over it's lifecycle.
 The reason we do estimation is that it is easy to do. It's substituion. We've taken a difficult question (what's the ROI, what's the value we are going to deliver?) and substituted that with an easier question with how much is it going to cost? What it costs isn't very important information, esp if you are building a MVP. It's more important to timebox the MVP and say we'll deliver something in a few weeks, certianly not more than three months. How can you get feedback more quickly?

 Individual productivity: how much has an individual contribute? Previously used poor metrics like LoC (liability not an asset) and hour worked.
 "The cost centre pattern fills the vacuum of our inability to define, model and measure the value most workers create for their organisation" - Ken Judy
 Produly said we have 10m lines of C++. high cost on maintainability. Some people use LoC as an asset on a balance sheet via SVN hook. 
 These things create bad behaviour. In my experience it's the same people working heroic hours that fucked it up in the first place. Working long hours makes you stupid.
 Measuring the value of people: Toyota: how do you create great people? Coursees - nope, people train you - nope. how do you do it? I don't know.
 Primary role of leaders in Toyota is to teach people. What do they teach? How to do experiments, fast, and always trying new things. Measure the value delivered.

 Jack Andraka (15 yo): "Make sure to be passionate about whatever it is you get into, because otherwise you won't put the right amount of work into it". Insightful about building anything. This is key to everything we do.

 Value is what gives life meaning. Find out what you care about. If you could do anythng and not fail what would you do?


 Q&A: Problem with estimates if that they are wrong. Esitmates have negaitve value: they are wrong and will give the wrong expectations. Biggest value is shared understanding of what to build. You get an idea what/how you will build it.
 Managers still ask: You don't know what you are going to build.

 "I'd say: start out _every_ new project with a team of 6-10 people, and set them a time limit of 2-3 months to build a POC, release it to production, and incorporate feedback from real customers. If you decide to continue, you can start up more teams. Make sure your architecture supports teams working on vertical slices as described in "Steve Yegge's Platforms Rant" (Google it)."

 Pricing model: http://www.youtube.com/watch?v=czYDshjmys4#t=6m20s
	http://www.youtube.com/watch?v=ESOaDiv3lXA

	Was not a product manager, read about Scrum & product management in it.
	Scrum says: work out what to build, prioritize & estimate, give feedback shows
	Other: Competiive analysis, pricing
	No longer a product manager

	It's super easy. Once you have an idea, you can get loads of stories. (priortized and esitmated, then actually build it).

	Came up with a product over 6 months, got 2 months feedback. Next two years a blur.
	Scaled agile exists: Assume a person is in charge of the product.

	Agile manifesto: What is value? No satisfactory answer. The one thing you can be certain about is that the PO will be wrong about what is valuable. This is what happens in reality.

	What is value? Shareholder value (from theory of the firm), maximise profits? CEOs are legally bound to do so, or they can/will be sued. Firms that focus on profit end up being less profitable than companies that do not.

	Roger Martin - fixing the game: 8 fold increase in CEO compensation on decreasing rate of return. Clear who the winners and losers are here!

	Elon Musk (found of paypal, made a bunch of money). He then made SpaceX, first private docking to ISS, sent cargo up there in less than 10 years. "I want to put people on Mars". Also on the side made Tesla motors, set out to prove that electric vehicles could be awesome. He's a millionaire, we can't all be Elon. Then again he just got divorced again recently.

	Jake Andraker - 15 years old - made a diagnostic for detecting pancrecatic cancer. Costs a few cence to build, way more accurate. He was told "go figure it out yourself". He got really into the scientific method.

	Jack Welch - Chairman of GE - says about sharelhler alue = dumbest idea in the world. Your main constituencies are your employees, customers and products.

	How to create value for customer - (read book on lean IT). "Users are unable to articulate exactly what they need, yet they seem insistent about what they don't want....once they see it.". Biggest problem is Scrum, build then say "look at me!" then get told "that's crap!". We don't know what's possible, new stuff is happening all the time. We don't know!

	We are building complex systems, you can't plan ahead, you have to try experiments. Saying you are going to plan all this stuff then build it is majorly flawed.

	Along comes Eric Reis: have a vision (universally true, could be called a goal), make a business model, build MVP, test/iterate, pivot when you reach a local maximum. Goal isn't money, it's a means. Funding proposals are miserable to write for. First thing you do is test the biz model. Most expensive way to prove a hypothesis is to build it. He released after 6 months despite bugs, despite being scared, and no one bothered to download it. Could of got the same result with a static html page with a 404 to the downlaod link! Work out how to test model, then build that. Not a new idea, but he's packaged it well. It's really hard to do this and put your money where your mouth is.

	Optimize your process: Build Measure Learn loop. Optimise for lead time for the loop. People normally optimise for busyness. Very easy to show with queining theory that optimising for 100% utilisation is inefficient. Image airport: checking in, loads of desks with one person, massive queue! If there's loads of people, much quicker, but the people won't be fully utilised. YOu need a bunch of people who aren't fully utilised to get good throughput.

	Easiest way to optimise lead time: reduce utilisation (put slack into the systems), reduce batch size (wip, less people queining at the desks), work in cross-functional teams (sit together, no other groups). This fundamentally challenges the concepts we have about projects. Book by Don Reinetson on the prinicples of product flow.

	There is a problem! Tradiontional focused people say it sounds like crap. Engineering should be like buildings! they don't fall down and kill. Barcelona Gaudie building. He studied flowers, all people did straight at the time. He used parabolic forms and hyperbolic structures. Very radical, people were very worried. It isn't a waterfall process. He was building models all the time. Most building of bridges is Trust bridge, put it into a model and get a bridge. If you are building something new then you have to experiment and iterative thinking. you want to optimise for doing this fast.

	Other people who don't behave in this way. Apple! People don't associate them with agile. Apple's first product was the Apple 1. Built in a shed in 76, It was really advanced for it's time, but consumers didn't want it. Very minimal, very viable. Macintosh changed to being a consumer product. They didn't argue about software ideas, they tried them,and kept the ones that worked.

	Where is the customer? Gaudie's church still isn't finished. He said "My customer is not in a hurry". Most of us don't have that luxery!

	Early Apple stuff was built for themselves, they were their own customer (or Steve Jobs). We have develop empathy for our customers.

	An exercise is to make personas on a target (primary consumer is 1 persona, then 2, then 3). People never want one "they want to dominate the sector". They deliberatley make the target big so it can't be put away! PO said they can't relate to his primary persona, was asked "how many of these widgets are you buying? none. Why should I listen to you? That's the area of your process I'm still having trouble with.".

	Naturally we want to build for ourselves. If it isn't you have to emphasise with the people you are builing it. you have to find a way to measure the value you are delivering to them. This is where you apply the scientific method.
	Tehniques: A/B testing (junk mail isn't all the same), show proto to real users, measure biz metrics, surverys/social media. nigeria 401ers have delbieratly got worse via A/B. They only want people that are massively gullabie. So deliberatly make it less gramatical. Google Analyrics do A/B tests for free out of the box.
	If you can't use real world. Get NDAs, film people interacting with you product. Once a user has it, they can give you great feedback.
	SAAS, measure biz value. Etsy: Measure capacity, conversion rate, measure each feature on conversion rates.

	POs should be working out what to measure value, then work with the team on how they are going to do that.

	Grockie defintion of done: "user stories were not considered done until they led to validated learning". Was it a good idea to do this? then you are done. One of your acceptance criteria should be "how can we measure if this was a good idea or not?"

	CaseStudy: Votizen: A/B got to local maximum. Pivoted biz idea to sustain biz through revenue. Your need an idea and a bunch of ways to get there. But you can't pivot too much as it becomes too much work to understand. At least with the data you can make rational decisios about if you are there or not, have we made it?

	True measure of value (for novels: no one wants to read your shit): will people use it? Using a product is a transaction as the user donates their time and attention which are valuable commodities which you the creator must five something worthy of that gift. This is worth bearing in mind while building stuff.

	the point isn;t the money, it's about acheving your vision. If not one is going to pay or use it it doesn't matter. If managment buy it but the users don't use it (it doesn't matter). Classic example are expenses apps: if users are mandated to use it (IDM) then you have to, so all your measurements are going to be off.

	The PM role: What is their role? Create & Communicate share vision. Work with team to define measurements. Run experiments to learn if your ideas are any good. Optimise your process for cycle time. Quality is everybody's responsibility.
	Everyone has to be excited about what your doing, and they are contributing to it (book:drive). Use scientific methods to build/measure/learn.

	Lifecycle for product dev (from the Don): 3 stages: inception, mvp, delivery.
	inception: get together, define vision, biz model, decide on mvp to test the idea
	deliver: make mvp, get feedback
	keep delivering small increments
	What's missing? Creating an enourmous prioritized backlog. Fundamentally waseful, as it estimation (book:how to measure anything) when you look at products over their lifecycle, cost has a very low info value, in terms of what you should build. the high value info is the value something delivers over it's lifecycle.
	The reason we do estimation is that it is easy to do. It's substituion. We've taken a difficult question (what's the ROI, what's the value we are going to deliver?) and substituted that with an easier question with how much is it going to cost? What it costs isn't very important information, esp if you are building a MVP. It's more important to timebox the MVP and say we'll deliver something in a few weeks, certianly not more than three months. How can you get feedback more quickly?

	Individual productivity: how much has an individual contribute? Previously used poor metrics like LoC (liability not an asset) and hour worked.
	"The cost centre pattern fills the vacuum of our inability to define, model and measure the value most workers create for their organisation" - Ken Judy
	Produly said we have 10m lines of C++. high cost on maintainability. Some people use LoC as an asset on a balance sheet via SVN hook.
	These things create bad behaviour. In my experience it's the same people working heroic hours that fucked it up in the first place. Working long hours makes you stupid.
	Measuring the value of people: Toyota: how do you create great people? Coursees - nope, people train you - nope. how do you do it? I don't know.
	Primary role of leaders in Toyota is to teach people. What do they teach? How to do experiments, fast, and always trying new things. Measure the value delivered.

	Jack Andraka (15 yo): "Make sure to be passionate about whatever it is you get into, because otherwise you won't put the right amount of work into it". Insightful about building anything. This is key to everything we do.

	Value is what gives life meaning. Find out what you care about. If you could do anythng and not fail what would you do?


	Q&A: Problem with estimates if that they are wrong. Esitmates have negaitve value: they are wrong and will give the wrong expectations. Biggest value is shared understanding of what to build. You get an idea what/how you will build it.
	Managers still ask: You don't know what you are going to build.

	"I'd say: start out _every_ new project with a team of 6-10 people, and set them a time limit of 2-3 months to build a POC, release it to production, and incorporate feedback from real customers. If you decide to continue, you can start up more teams. Make sure your architecture supports teams working on vertical slices as described in "Steve Yegge's Platforms Rant" (Google it)."

	Pricing model: http://www.youtube.com/watch?v=czYDshjmys4#t=6m20s