Representative democracies were invented to solve the "junk food policy" (i.e., passing out all of the money in the treasury) problem of direct democracy. Maybe there could be a representative link aggregator where annual elections determined who was allowed to vote on links.
A variation on representative democracy would be to have a recommender system identify demographics, and then to weight each demographic equally. So, for example, the votes from the "blogspam fan" demographic would be normalized to one, to fairly compete for representation with the much smaller "long thinkpiece fan" demographic.
Another solution would be quadratic voting, where you can vote once for free, but the nth vote would cost (n-1)^2 reputation points. That would allow established community members to express their intense dislike of certain content, to balance out the larger population's mild preferences.
It would be pretty cool if there was Reddit, but each subreddit could implement a different voting system. We might see a lot of progress and experimentation.
One of my instructors way back at Cornell was a grad student named Kevin Walsh, now an associate professor at Holy Cross, who had solved this problem in a bit of a radical way. Context was that back then you had Gnutella, LimeWire etc.—peer-to-peer types of networks—and you wanted to be able to enable people to rank media a certain number of "stars" for its quality because otherwise people post not-safe-for-life content with nice endearing names and others mass-download that content and then get grossed out. But you publish this and then people immediately use the anonymity afforded by the Internet to spam the upvote button as they distribute their garbage.
My understanding from Walsh was that the problem was essentially an economics problem—you want to incentivize good behavior and deincentivize bad behavior—and once you understood this you could use the network to correct itself, essentially saying “if you use your upvotes like the typical user does then the typical user will trust you, if you use your upvotes like the typical spammer does then you’ll instead end up in a clique with typical spammers.”
Some googling reveals that the page is still alive on cornell.edu [1].
It seems like they distinguish between "recommendation making" vs "verifying correctness of content" and Credence is meant to solve the latter:
"Since Credence is not a recommendation system, your thumbs-up and thumbs-down decisions should be based on an objective evaluation of whether a file's description matches its contents, not on matters of taste." https://www.cs.cornell.edu/people/egs/credence/faq.html
Oh geez, this metaphor between representative democracies and social media sites clicked for me really well, and I very much don't like that because I'm interpreting this as a strong argument in favor of user-sourced moderators (who I think cause more problems than they solve).
You've made me uncomfortable, so... thanks, I guess?
