Hacker News new | ask | show | jobs
by tokyodude 2621 days ago
It won't work without a central authority. See Soundcloud as an example. People tag their music with whatever they think will get them traffic. So, in order to do this you'll need a mass of volunteers which will lead to politics "XYZ should be classified as G! No it should be F!", "classifying ABC as DEF is racists/sexist/..." and other arguments. You'll also get people lobbying to have things removed (right to be forgotten, pornography, drug ad, prostitution ads, disparaging the government -China, -Thailand, etc..., etc...

I'm not saying it shouldn't be done but I think it will be way more work than expected and there will be all kinds of issues.

3 comments

If anything, that sounds like a solid argument to decentralize it. I don't want China's government, white supremacists, churches, soccer moms, Jihadis or grievance-of-the-month activists controlling how information is indexed; I would rather use multiple indexes that balance out controlling interests and biases.
Unfortunately if it's decentralized, then it becomes controlled by spamlords, SEO artists, advertisers, and anyone else who stands to gain from manipulating the index to their advantage. At least if it's centralized, the fights are out in the open and have a chance of converging on something reasonable (like e.g. wikipedia).
Decentralized doesn't mean flat. You can trust to some actors only (and to some they trust to).
I believe the problems are far smaller when building web directories in narrower contexts such as people using the web to learn new topics / acquire new skills. News, politics is where most of controversies lie, compared to, say algorithms or abstract algebra.

Our project LearnAwesome[1] currently relies on volunteers to curate topics, but classfication / ontology engineering is in fact seems to be a hard problem.

[1]: https://github.com/learn-awesome/learn-awesome

No, you wouldn't need a central authority, though various subject indices and/or search interfaces (divorced from the crawl/index) would supply rank and/or reputation scores.