3 ms·
Censorship isn't something I'm in favor of, but I think the idea Berners-Lee has is something different. Any given content has "viral coefficients" that determ
by csense 10y ago
Censorship isn't something I'm in favor of, but I think the idea Berners-Lee has is something different. Any given content has "viral coefficients" that determine how fast it replicates and how likely it is to die out before penetrating.
His idea's to make a social network where toxic content's has lower viral coefficients. Censorship is certainly one path to this goal -- simply kill any objectionable content before it has a chance to spread.
How else could you make a piece of content less likely to spread? Display it less prominently. Warn people that it spreads negativity and they shouldn't propagate it. Automatically identify filter bubbles and occasionally include content with an opposing viewpoint.
The real problem is some/all of these measures may cost you users. How do you lower the viral coefficient of toxic information in your social network without lowering the viral coefficient of your social network itself?
- garysieling 10y agoI've been building a search engine for lectures (https://www.findlectures.com https://www.findlectures.com), so I've spent some time comparing search results vs. youtube. They seem to weight popularity heavily, which makes sense if you're trying to sell ads, but it means a lot of toxic content rises to the top (e.g. conspiracy theories, or tons of short videos denouncing the video of a well known figure, which becomes un-findable). I'm banking on being able to improve this for my niche use case with more quality measures, although time will tell if this pans out. E.g. if a speaker has spoken at prestigious conferences, written a book, what Wikipedia says about that, or that they've invested in better production quality, including closed captions, better audio/video. For the filter bubble problem, I think that being able to see lists of talks without the machine learning what you "want" is potentially a superior experience. I suspect that if you start by filtering to just the higher quality material, you can potentially get a good cross-section of different viewpoints, just by using text text analysis.
- dahdum 10y agoWho gets to define what thought is "toxic" and what is not? I don't see how you can define it algorithmically.
- csense 10y agoDefining it's tricky, and it is hard to see how to do it algorithmically. But that doesn't mean it's impossible. I think Reddit does something interesting with upvote / downvote ratios, I don't remember the specifics. It might also be the case that toxic content spreads differently, or there's some other measurable way to separate it. Data and experiments to measure user behavior will probably be needed to develop an algorithm.