The Perfect Social Voting System | Hacker News Reader

The Perfect Social Voting System

judstephenson.com

5 pointsJudson16 comments

You're essentially treating each article as if it's an asset. Users are allowed to go long or short on the article, depending on their prediction about its final karma value.

Are people then allowed to exit their positions? Can I place a stop loss? Can I upvote this article, but when it gets 5 down votes decide that I should exit it and go short?

If I can't exit my position, then what is the incentive for me to ever be the first person to vote on an article?

I think it's intuitively obvious to everyone that an article with 20 up votes will not start getting down voted all of a sudden. If I were to guess, I'd say that after 5 or so up votes, you're 90% or more to see a higher close; after 20, it's probably more than 99%. However, at 0 votes, you're probably close to 50/50 (again, just guessing based on experience).

If you can't exit, then your risk of ruin is way too high to make initial voting worth it. If you can exit, then it becomes easy to game the system by switching sides quickly.

Also, what about the ability of people to submit really crappy articles on a dummy account and then down vote them on a main account in order to build karma?

I did toy around with weighted entrance rewards, but it makes the model more complicated.

After thinking about reversals a bit more, it would make sense to allow users to "exit" their vote. Bringing the total vote count (positive or negative) closer to zero and allowing more unbiased voting to continue (as people exit their votes, the percentages fall back to 50/50).

You might be interested to know that this method has been employed by Cloudmark for many years in their collaborative spam filtering network: http://portal.acm.org/citation.cfm?id=1105677

However, it works so well for them because the consensus on what is "ham" and what is "spam" is nearly universal enough that it can be relied upon to be agreed with by the majority of agents (people) the vast majority of the time. For news links, this will unlikely be the case unless you are actively attempting to create a news site where only a very few have power over what is shown and what is not.

Also, the Cloudmark technique chooses very few people to reward and this is on purpose. They are sparing with the reputation points and quick to downgrade on controversy in order to prevent a high-reputation individual gaming the system. Your news reputation system would likely have to incorporate a similar technique in order to be similarly robust. However, in doing so, I would imagine that most people would not see their reputation ever rise and would likely stop using the system. (due to lack of the "incentives" of which you speak) Cloudmark gets around this because people are already marking emails as spam or not in their email clients already, so there's no marginal cost associated with training Cloudmark's network or building their own reputation.

Its an interesting direction you propose, but I believe you'd be exposing your user base to scale-free network effects in which only a small few would gain sufficient reputation to "move the needle" and the rest would wallow in relative obscurity. This would then essentially replicate the Slashdot of 2000 AD but with the marketing message saying that it was "fairer" because of all the "incentives" people have.

You might want to think of how to structure a social news site as a kind of online auction, choosing a structure in which truth-telling is a dominant strategy, instead, such as http://en.wikipedia.org/wiki/Vickrey%E2%80%93Clarke%E2%80%93...

It doesn't matter if few people vote, so long as enough votes are registered to rank the comments/articles.

The only thing I can see happening is incentivizing karma whores like me to support popular viewpoints even if I disagree (I don't downvote, but I don't have to upvote). Again, I don't think participation is a problem as long as there is a consistent group of active participants.

Also I don't vote for a warm fuzzy feeling of making the site better. If someone says something I think is right, then I upvote to show my support and hope his viewpoint is propagated enough so that further misunderstandings on the issue is avoided. This way I will decrease the likelihood of some ignorant person making his/her way back to me and forcing me to waste time re-espousing said viewpoint. My motivation for contributing is purely in self-interest.

That is a valid point, but I would counter that, while a small group of active participants does allow "the process" to work, more people would express their opinion if voting was incentivized.

Also, the "warm fuzzies" was a simple metaphor for upping the incentives for voting, as "warm fuzzies" could be considered themselves a net marginal gain. If your going to play the karma game (up and down voting) why not reward the users in the same nominal terms used to describe the content they voted on.

My commenting, and most peoples' commenting I would assume, is done to get a weight off their chest, not for karma. Unless we can redeem karma for cash, even karma itself has little incentive to be gained by users. I'm not sure how voting making more people vote would make more people comment in this way.

Interesting and very insightful points you make in your blog entry. However, I think there is one invalid assumption your social voting is based on - the fact that the first and last user will vote.

In the point in time at which they vote, the first and last user experience a hope for a future incentive - but no actual incentive. Why should the first user vote on a comment about some piece of content?

This dilemma makes this system the same as the current, pre-dominant social voting system - it provides no initial incentive for someone to create activity, only the hope for a future incentive.

There's no need to have absolute karma points or additional incentives.

We just need an intelligent collaborative filter that weighs articles relative to you. That will give users an incentive to vote as they wish (rather than vote for what one thinks others think good).

Pandora News anyone?

The point is to give users a reason to vote on stories other than for "that warm feeling of communal good".

If you aren't convicted about the destination of an article, then you won't vote. Karma makes HN game-like.

Its about making content that other people like, not content that a few people have filtered for.

This system would only promote the common denominator. For example, divisive (but good) articles would get less voting attention because people would be afraid of voting the wrong way. No?

Since we don't know exactly how people would react to given incentive/risk profiles that would undoubtable differ story to story we can assume two scenarios:

- Stories that have perceived "deadlocks" would be considered volatile. They have a great chance of "breaking out" with large support, so they would be voted on early by supporters to reap the reward.

- Divisive stories/comments would scare away any voters and eventually only stories that were beneficial to the entire community would be voted on. So in this scenario, this system helps to define the community.

edit: Also, the problem that you describe isn't isolated to my system. In the current HN system, people don't comment on stories because they fear that their comment will be "divisive" and attract negative votes. My system encourages voting, which would reward someone for voting on a "-3" comment (that makes it back to 0 or 1) that is beneficial to the conversation.

Pretty interesting. I'd like to see it in practice.

Speculative market based on interest, relevance and attention is an interesting idea. Does it need to be tied to a social news website?

Articles and comments that are upvoted provide more marginal utility? Or are they simply more pleasing to the reader?

If I tell you that your business model sucks and that you should try selling shoes, you (and others) may be immensely displeased. But then you may try selling shoes and make a fortune.

Nope. Marginal utility and immediate gratification (or ability to predict what a crowd likes) is not the same thing.

Sorry Daniel, but this is in no way what I was talking about. I am not saying that marginal utility = what the crowd likes. Neither did I say that HN isn't "working". Only that providing a more quantifiable reward for correctly voting on comments could prove beneficial.

My fault for misunderstanding.

So perhaps people could bet parts of the karma that they have already accumulated that the story would get more votes than the running average?

Reply on news.ycombinator.com