Individuals and companies have the right to do what they want with their own property, and I don’t have any great ideas on how to solve this dilemma. One solution would be to have many platforms but we live in a winner take all world and platforms like YT get more powerful because of network effects.
The content hosting with youtube is nice, but the bigger problem solved by YT is search. So a video up on a server somewhere might as well be dark unless something can link to it and someone looking for it or something similar can be pointed to it. That search feature itself can't be centralized or else we have the same censoring problem.
It would be nice if a type of searching system could be bundled with IPFS, though I don't know what that would look like. The sooner we have a system where no one can close the gates, the better.
Maybe with something like opentracker there could be an endpoint like /search that translates search phrases into a sparse column vector and looks up cosine similarity against that instances low dimensional space representation of the tf-idf against torrent hashes, those low dimensional space representation files can also be distributed in torrent file and as well as way to constantly update/compute new low dimensional space representation files as hashes are added.
For example it would probably look something like what I do in [0] but as a compiled module linked into and accessed from [1] as well constant update/rebuild of ldsr files like [2] linked and kicked off from flags passed in [3] as new hashes come in.
[0]: https://github.com/cinquemb/hackedteam-email-index-mining/bl...
[1]: https://erdgeist.org/gitweb/opentracker/tree/ot_http.c
[2]: https://github.com/cinquemb/hackedteam-email-index-mining/bl...
[3]: https://erdgeist.org/gitweb/opentracker/tree/opentracker.c
At the same time, the author gets much more views saying his video was censored than if he was just one more video made on the platform. Just like singers would look for the +18 black mark on their LPs back in the 80s/90s.
That's quite an assumption that human moderation is worse than an algorithm.
> "and the amount of people needed would be impractical."
What is up-voting and down-voting and reporting? It's not like there's a lack of human eyeballs on the issue already. If no one watched, then no impact.