An elegant mathematical solution against fluff on News.YC

A few notes:

1) Users with an old account do not necessarily write better comments or post better stories than recent arrivals.

2) Length doesn't equal insightfulness. There might be some correlation, but not a lot. This comment: http://news.ycombinator.com/item?id=110034 is three characters and has a vote of 57.

I like the idea of people being able to write their own formulas and share them. Especially in a forum like this :-)

I like the idea of people being able to write their own formulas and share them.

I don't know how flexible YC News is but it would be great if each user could express their own individual ordering preference relations in their profile.

maybe even a modding system for the modding systems...

I agree, that people with old accounts do not necessarily write better, more insightful comments or post better stories than recent arrivals, but they have been around longer, and are probably more committed to the community.

Case in point, the 3 character post that you mentioned was posted by davidw: http://news.ycombinator.com/user?id=davidw He's been a member for over a year, and his karma is over 2300.

>I like the idea of people being able to write their own formulas and share them. Especially in a forum like this :-)

I think this is a great idea. If all the relevant data were made available (most of it probably is) it could probably be implemented client side in something like greasemonkey.

On the other hand, length plus lots of points is a sure sign of insightfulness, or pg, not that there's a difference.

The MetaForum messageboard software has been doing this for years. Everyone gets to vote, but people's voting power is determined in turn by how valuable to the community they are. It removes the need for moderators entirely, and completely eliminates the ability for spammers to thrive.

Their formula bases a large amount of worth simply on how their posts have been modded in the past, but then also include things like post count and a decaying value on length of time.

News.YC is a community. Our individual behaviour taken as a collective is what constitutes the dynamics that emerge. There are benefits to both maintaining an organic element of letting the community choose for themselves and to having some mathematical/code-based governance.

At the same time, there are a number of assumptions in both the formula above (as well as the title mind you) and its description that are more in favor of eliminating the diversity. This could detract from the whole user experience. While I personally don't like the fluff, the diversity it affords us all more valuable than having our contributions be governed by an equation...

> Of course, the real hacker spirited solution would be to let people make/share their own such formulas to make their own front pages, but I'm not sure if Paul has the time to implement that.

I have this in my roadmap for Vortices, see http://vortices.appjet.net/node/obj-LBlNDy7Ow

A simple weighting equation? Where's the "elegance"?

Elegant in the sense that you apply one general formula instead of hacks (managing things manually is a hack, yes?).

I think ideally, karma should just be there for feedback for the poster and the reader. Paradoxically, the only reason karma is meaningful is that it is useless - there is no incentive to game the system. If more karma means more power, there is an incentive to post to get karma instead of posting to help the community.

If you don't believe that incentivising karma is a bad thing, try this thought experiment: if you gave away $1.00 for each karma point earned, would the quality of the content go up or down? You can add some conditions if you think the experiment is unfair, but I think you would still find that the quality would be reduced with incentives.

Ultimately, 99% of the users are just like you and I and do not want to see news.yc turn into another digg/reddit. Why not take advantage of that, instead of treating it like a technical problem when it is really a community issue.

> If more karma means more power, there is an incentive to post to get karma instead of posting to help the community.

One could argue that in the current community, posting to get karma and posting to help the community are one and the same. Reddit's downfall came more because of a change in the composition of the community than because of karma incentives. At the same time, you're right about the problems of people trying to get karma for karma's sake. I think the best way to reduce this problem is to make it harder to gain karma than just by posting popular links.

> Ultimately, 99% of the users are just like you and I and do not want to see news.yc turn into another digg/reddit. Why not take advantage of that, instead of treating it like a technical problem when it is really a community issue.

You're right. It is a community issue. But as soon as the "hordes of 14-year-olds" invade, the 99% you cite will drop. The current technical solution doesn't prevent that from happening. We're trying to find one that does.

Man, I'd kill to get $1 per karma point... I wonder how much my account would go for on eBay.

No. But, weigh comment karma a lot more than submit karma and it might be alright.

Is comment karma tracked differently than submit karma?

How about forcing new users who want to contribute to go through a bit of a logic test? Perhaps answer some simple questions before signing up, to ensure they're at least a little alive where it counts?

It'd be interesting if the weighting algorithm was somehow exposed to your account so you could mess with it.

Another thought: I think the karma score should also represent rate, instead of just weight. Otherwise the first people here become entrenched in the leader board. A static hierarchy seems pretty unhackerish.

PG responded to this thought previously, saying that karma inflation takes care of it.

What is karma inflation?

Over time karma is worth less, because a popular story will get 20 or 50 or 100 points when it might have only gotten 7 points when this site was just getting started

Yep. I lurked for about a year and then in the last ~2 months since I registered and started interacting, I've gotten almost 600.

The problem with this is that Hacker News isn't supposed to grow very big.

Bingo. I think that's the key. And therein lies the problem. How does one keep a community small and high-quality without losing diversity.

Good idea. There is a 'conflict of interests' when you give the length of the comment a value. It will give people incentive to write longer comments for unnecessary reasons, which is the definition of fluff.

c.disagree : concise > ramble

Where's the MathML? =P

The everybody will put: date of birth = 1800