Federate search

What I'd like

self-hosted search engine
meta search, checking several APIs, e.g. bing, google (are there others?)
search through an index of my own web browsing history (I may be looking for sth I read last week)
search using search API offered by sites, eg Wikipedia, github, archive.org, media.ccc.de, peertube, youtube even (do they even offer APIs? They totally should!)
search documentation sites I visited before (so not just the page I read, index the whole thing or use their search)

as much of that in local index, have that be very efficient (in power, storage is probably not an issue)

just some stuff that came to mind, no order, no implied usefulness

Notes (taken from his slides):

The past

we used to have catalogs of stuff, e.g. Yahoo
Syndication (Meta Content Framework, RDF Site Summary, Really Simple Syndication)
Parallel Search (WAIS), gathering, indexing, brokering (e.g. Harvest)

Re-imaging the present

List of generally trustworthy sites (e.g. Wikipedia, it's better than is generally credited)
Auto-discovered RSS + XML Sitemap feeds (* Hello Fediverse)
Personalised search (ElasticSearch, Apache Solr etc)
Parallel search (e.g. SearXNG, LibreX) of noteworthy sites (via regular search engines) + personalised

So basically:

Search history quickly loses its relevance if the sites that are found go away.

So maybe it needs to be combined with archivebox?

Maybe the archive could be kept in a form like https://alexwlchan.net/2024/static-websites/ .