I tried to build something like this in 2007, together with a small band of nerds and geeks and Linux enthusiasts. It was called Beeseek. [0]
I knew close to nothing about building a company or a project, or how a proper business model would have helped it. I was the leader (SABDFL) of the group, and unfortunately I didn't lead it well enough to succeed. We had some good ideas, but ultimately we failed at building more than the initial prototype.
The idea behind it was simple: WorkerBee nodes (users' computers) would crawl the web, and provide the computational power to run Beeseek. Users could upvote pages (using "trackers" that anonymously "spy" the user in order to find new pages - repeat: anonymously). The entire DB would be hosted across multiple nodes. Auth and other functionalities would be provided by "higher level" nodes (QueenBee nodes).
Everything was going to be open source.
Well, it didn't work.
Thankfully, because of Beeseek, I met a few very smart people that I am in touch with to this day.
Life is strange and beautiful in its own way.
Weird, though, that today I still believe that Beeseek could have been the right thing to build. Who knows?
In what ways does what OP describes remind you of your project? Just that it was an open source web search?
One difference from what you describe is that the OP is specifically recommending against decentralization/federation, where it seems to have been the core differentiator of your effort. I don't think what OP is describing is quite what you are describing.
I think that DDV was arguing against decentralization/federation for searching the index. Not necessarily related in any way to building the index (if the distributed nodes all just forward back results to central hub).
I knew close to nothing about building a company or a project, or how a proper business model would have helped it. I was the leader (SABDFL) of the group, and unfortunately I didn't lead it well enough to succeed. We had some good ideas, but ultimately we failed at building more than the initial prototype.
The idea behind it was simple: WorkerBee nodes (users' computers) would crawl the web, and provide the computational power to run Beeseek. Users could upvote pages (using "trackers" that anonymously "spy" the user in order to find new pages - repeat: anonymously). The entire DB would be hosted across multiple nodes. Auth and other functionalities would be provided by "higher level" nodes (QueenBee nodes). Everything was going to be open source.
Well, it didn't work.
Thankfully, because of Beeseek, I met a few very smart people that I am in touch with to this day.
Life is strange and beautiful in its own way.
Weird, though, that today I still believe that Beeseek could have been the right thing to build. Who knows?
[0]: https://launchpad.net/beeseek