Hacker News Reader

Comment by fouc | Hacker News Reader

Does anyone know of or have ideas on how to restrain winner-take-all feedback loops? I'm rather tired of the lack of fat tailed distribution in the tech world

View on HN

I think it would be interesting to use a market based mechanism for voting, where placing an upvote costs a small fee. The fee price increases as more upvotes are placed, and you can always un-place your upvote to get a refund at the current market price.

This means accounts that identify popular content before others do can make a profit by buying upvotes at a low price, and selling them high.

This doesn’t stop content from popular accounts from being upvoted a lot, but it does encourage people to search for otherwise overlooked high quality content - to your point, the fat tail.

https://ethresear.ch/t/better-curation-via-inversely-coupled...

Quadratic voting!

https://en.wikipedia.org/wiki/Quadratic_voting

Very similar! The difference is that Quadratic Voting scales the expense independently for each user, while this system scales it for everyone (and you can subsequently sell your vote)

Would this fee be based on money or "karma"? I feel like if one had to earn upvotes to spend on Reddit, regular users will engage in more karma-farming reposts. At least now it's only the weirdos and people looking to sell accounts.

Has anyone tried a logarithmic or similar scale for voting? Show the 'tier' they are at (A B C D or some rating number) and it gets progressively harder to go up or down a level as you deviate from the starting point.

I'd imagine that fee would be literal money, as this nullifies concerns about people creating fake accounts to get more Karma, and means people are fully incentivised to place votes carefully. It does introduce a "pay to influence" aspect that I don't love, but one would hope that at sufficient scale the Efficient Market Hypothesis would kick in, and trying to pay to influence would just result in you burning your money.

> Has anyone tried a logarithmic or similar scale for voting?

Hmmm good question. I don't think this would make a huge difference, as visibility is generally a function of a post's current rank rather than its actual score, so any monotonic transform isn't going to have a large effect.

I have in mind a post reddit social site (in the save way reddit is post digg), and one of the key issues that I see is that that site has to have distinct karma for each community.

Basically a social network based more upon Slate Star Codex concept of Archipelagoes (communities next to communities) than a traditional communities within communities. I think one of the key original sins of reddit is that it had site wide karma, rather than distinct karma within each community.

That’s a very interesting idea. Is that how the stackexchange/stackoverflow/etc sites operate? They seem siloed but I don’t have much experience with them.

I also saw an interesting point here recently that the reason Instagram seems to be better than FB/Twitter is that there’s no built in way to signal boost with shares or retweets. And really, you can’t even have hyperlinks. Would that be a consideration?

You get 200 basic karma when you go to a new stackoverflow community with an existing account (or you did the last time I did) but other than that you start over from scratch.

However Stackoverflow is also not really a community, so I don't think what they do necessarily applies.

I don't use Instagram much (don't need more addictive stuff in my life), but I seem to remember that you could still like stuff. No hyperlinks would ruin a community in opinion, since it means you can't back a point with facts.

If you ever create this, I would be interested in joining.

Yes. Remove the incentive to grow social capital by making users anonymous and posts temporary. It exists and it’s called 4chan. It turns out that when you remove the incentive to grow social capital, the vast majority of people aren’t interested in participating.

According to their ad page, 4chan has 22 million monthly visitors, which is not as large as Twitter’s 300 million, but I would argue is also not insignificant: https://4chan.org/advertise

But is 4chan an inevitable product of its structure, or of its originating culture? Back when it was the next big thing, I worked with a social bookmarking startup. Weirdly, our site was colonized by a sizable group of Harry Potter fan fiction writers, and they became the dominant group on the site. If 4chan had started out with a different seed group of contributors, it might well be a very, very different place.

> If 4chan had started out with a different seed group of contributors, it might well be a very, very different place.

Hoping that's the case as I'm working on a new chat-based discussion site with no voting and goal of open, rigorous, reverent discussions (sqwok.im). I think there's merit in using signals like activity and others to determine relevance, and agree that part of the problem with 4chan is simply the ethos set in the beginning.

maybe making each like or dislike solve a small problem to remove the incentive unless you have a strong like/dislike eg a really annoying capture

Don't have the web be four sites full of screenshots of each other. I.e. go back to the blog and forums era. Outlaw social media with more than X users. Make content recommendation algorithms illegal. Personalised advertising should be punishable by public flogging for first offenders and death for repeat offenders or particularly large-scale operations.

This is obviously unreasonable - capital punishment should only apply in cases that are both large scale and repeat offences.

Reply on news.ycombinator.com