It came to my attention that when you receive downvotes for your comments, your karma goes negative and you need to "pay back" to be able to post to Discussion or to Main.

Since new users start with zero karma, having negative karma seems to just encourage those with negative karma to create a new account. We don't want to encourage people to create superfluous accounts, do we? Therefore I think LessWrong codebase should be patched so that karma does not go below zero even with lots of downvotes.

What do you think?

New Comment
31 comments, sorted by Click to highlight new comments since:

The need to create a new account is a trivial inconvenience that makes it slightly less likely that the person in question will stick around. And people with negative karma are probably ones that we, on average, would prefer not to stick around. Your proposed patch would make it more likely that they did stay and remained active.

[-]Emile150

Rather than fixing hypothetical future problems, I would rather have work spent on fixing things that cause problems now, for example:

  • when you click on a permalink to a single comment (like those you find in your mailbox, or in the recent comments), the whole thread it's in is marked as read (and won't be in green any more)

  • There is no way to see the upovote / downvote split on posts and comments, only upvotes minus downvotes (though I'm not sure that would be such an improvement, it's just something a lot of people ask for; there recent "poll" feature may be a better way)

(unless of course you were planning on implementing, testing and submitting the fix yourself!)

  • This came up on #lesswrong IRC channel, it was not a hypothetical future problem at all.
  • Yes, I was planning on implementing, testing, submitting the fix myself.
[-]gwern120

This came up on #lesswrong IRC channel, it was not a hypothetical future problem at all.

Being discussed somewhere does not show it is real. Personally, 'people are cycling from account to account to evade karma limits' doesn't seem like a real problem...

[-]TimS40

Can you be more specific on what you think the problem is?

I agree that the current (downvoting + minimum karma for Discussion) dynamic encourages creating new accounts, holding everything else equal. But I'm not sure what the downside of inflated registered-user numbers is - there's not much sock-puppetry locally, so it isn't clear how additional registrations can be considered relevant to signal-to-noise issues.

And there are tangible benefits to the downvoting system (helps increase quality posts, crowd-sourcing basic moderation).

Ah, cool!

Then, I have much less to complain about, offers of help are much more useful than suggestions from back seat drivers :)

I agree on both, but I think you should have this as a separate discussion thread.

Well, though I also don't really want to start a new discussion on these (this kind of meta stuff doesn't add much value to the site), just say "these seem to be things people complain about even more".

If you have negative karma and can't post, this is because WE DON'T WANT YOU TO MAKE POSTS. This is as it should be.

So far, I very rarely see anyone making posts, getting downvoted, changing accounts, and making more posts on the same topic. As sockpuppetry isn't that hard to detect, I think we would have a good idea if this was rampant. This would have to be an actual problem before I removed something I want from the forums to fix it.

The only case I can think of is that Dmytry/private_messaging guy who went through at least three accounts to keep on ranting and ranting about god knows what. I don't know whether he stopped posted altogether or changed his behavior so as not to be such a nuisance, but in either case, mission accomplished!

[-][anonymous]90

I disagree strongly.

We don't want to encourage people to create superfluous accounts, do we?

Why? Because we will run out of bits?

Therefore I think LessWrong codebase should be patched so that karma does not go below zero even with lots of downvotes.

I disagree. There is a useful distinction to be made between someone starting out at 0 karma or just breaking even and someone running a negative score.

There is a useful distinction to be made between someone starting out at 0 karma or just breaking even and someone running a negative score.

There is, but my point was that you can't make this useful distinction usefully.

I recall there have been a few posters who have cycled accounts to make top-level posts, but I think this declined dramatically after the karma minimum for posting went up. So probably a pretty low-priority, but I agree there is an incentive problem (depending on how attached people are to their accounts).

Why? Because we will run out of bits?

Because we end up with loads of fake accounts which just add noise to the stats of the site for example.

[-][anonymous]40

So far no one has done an analysis on the data set of all registered accounts.

This. And This

Why? Because we will run out of bits?

And This

There is a useful distinction to be made between someone starting out at 0 karma or just breaking even and someone running a negative score.

It's proposed to spend coding resources solving a hypothetical "problem" for newbies whose first posts result in negative karma.

Have we determined that this is a problem for any significant number of people?

Moreover, with all the squawk about signal to noise, shouldn't we want to cause problems for such people? The negative karma imposed some cost on the guy. Why is that a bad thing?

And note how this "fix" would mask a troll, making his karma indistinguishable from a complete newbie. That's a more serious theoretical issue than some newbie in a tizzy over how he would wipe out his -2 karma and start again from scratch.

This nonfix "solves" a trivial problem for a set of users we probably want to cause more problems for, and introduces a serious hole for trolls.

Bad idea.

And note how this "fix" would mask a troll, making his karma indistinguishable from a complete newbie.

Total karma scores still don't display below zero, although monthly karma scores do. If you want to know whether a person's karma score is zero or negative, and if so, how negative, you have to look at the scores on their comments. The current system is designed to make negative karma a burden that keeps people from posting beyond a brief reversal in conduct, without publicly displaying just how much the person in question has been downvoted.

Odd. Why should the community be in the dark about a negative karma score?

That's one thing about requirements: what the requirement is let's you build it, why the requirement is let's you maintain it.

As far as I know it's to discourage people from aiming for negative karma, and taking pride in how much they've been downvoted - as proof of their trolling or brave thinking out of the box or willingness to "tell truth to power" or whatever. I think that happened a few times.

(it's a bit like how sometimes punishing a misbehaving child can actually be "rewarding him with attention", and result in more frequent misbehaving)

Interesting. Everything is a tradeoff. When people aren't alerted to a troll, there's a cost, and when they are, there's another cost.

Is there any mechanism for booting a troll?

Is there any mechanism for booting a troll?

Essentially, no. If I recall Alicorn has to delete individual comments rather than ban users... more by oversight than design. As far as I'm aware Eliezer hasn't got a "boot troll" feature available to him either. Matt would... via an SQL UPDATE if nothing else.

Admins have the power to delete posts and comments, and probably users (though I can't think of a time it happened); I wouldn't be surprised if that was used mostly to fight spammers, not trolls.

I don't know why negative karma scores aren't displayed. I vaguely recall the matter being discussed back when the system was implemented, but I don't remember the reasoning behind it, if I was ever privy to it in the first place.

I have seen a few, e.g. this one. And accurate user number statistics are handy for measuring LW growth and the impact of various changes.

[-][anonymous]00

That is the data set of "the accounts that google can find".

Good thing LessWrong cares only about the past and nobody is interested in anything that might happen in the future :p

And anyway this is not the only reason, you can easily argue (I am doing it right now) that it is better for a site to have less fake accounts. It might be just slightly better but it is not very hard to implement a fix anyway so we might as well implement it and make this place slightly better.

A word from a cranky tech professional: unless you are personally capable of making a change to a system, make no assumptions as to how easy or hard it is to implement.

Wise words. I should've said '..if it is not very hard to implement a fix we might as well...'

It's reasonable to conclude "if this is a potential problem, and it's easy to solve, we should implement that solution assuming the solution does not cause other problems." But negative karma was deliberately implemented to keep disruptive individuals from continually posting after being heavily downvoted. To my knowledge, while there are people who've created sockpuppet accounts, the issue of someone creating a sockpuppet to get around negative karma has not yet come up (and if it has, the people doing it managed to create a sufficiently clean break from their earlier records that their taking up new handles was probably for the best.) On the other hand, people who've been regularly disruptive and had their posting throttled by negative karma is an issue that has come up before.

Comments in the welcome thread usually get at least half a dozen upvotes each, which makes it unlikely that a newbie will end up with negative karma unless they are knowingly trolling.

If your worry is that creating new accounts will help trolls, the obvious solution is to create lower limits that are in the positives, so that new accounts won't be able to post either. Nothing onerous - say, +10 lifetime for Discussion, +100 for Main. Certainly removing the barrier to -karma posting is a bad idea.