[book_mentioned]

My latest campaign has been to find a way to add the tiniest bit of accountability for downvotes. I believe at a minimum the total number of downvotes given should be shown publicly on user profiles.

As things function now, downvotes are for all intents and purposes anonymous.

[pvg]

Why do any votes need 'accountability'? Their purpose is to identify comments worth reading. Almost everything people write about voting itself is not worth reading and the less of it, the better.

[yeukhon]

Worth reading is very subjective. I think you are looking at popular opinions. The current system works that way as opposed to some randomized insertion in between. One often problem with popular opinion is that readers might jump on that trail and build a deep thread, which probably reduce the incentive to read any further by either skipping the middle to the end, ir simply ignore the rest.

Would be an interesting research to conduct if randomness can help based on say user karma. For cryto and security we know there are certain users seen as “goto”, so their opinions weigh much more and might end up being the top all the time. I’d like to see some “penalty” by promoting the less-popular users (not comment).

The point is treat tge display not entirely based on number of upvotes the entire time.

[book_mentioned]

My intention is to ensure contributing to publicly censoring/censuring costs something (intentionally tiny, but greater than zero/nothing) each time, beyond the one-time minimum karma requirement. "Downvote to disagree" seems unlikely to scale indefinitely, particularly as mobile means less willingness to contribute beyond clicking up/down (specifically now that there are apps catching on that remove all "fat-fingered web UI" friction!).

It's perfectly fine to disagree with this belief; HN itself does!

[Edited]

[pvg]

I know it's perfectly fine, I'm trying to follow this (very common) line of thought and adherents never seem to explain it in a way I can understand.

For one thing you're not really 'censoring' anything but even if we say, for the sake of argument, that you are why should it have some cost different than the cost of 'promoting' something when you upvote? The mods' argument that meta is bad for actual conversation is also quite compelling. What's the counter to that other than stuff that mostly seems to boil down to some version of 'getting downvotes kind of feels bad' (of which calling downvotes 'censoring' seems like a particularly overwrought variant).

[book_mentioned]

Thanks for taking the time to bring up these counterpoints!

>why should it have some cost different than the cost of 'promoting' something when you upvote

If this is a blocker, show both! My recommendation to add the tiniest bit of friction only to the anonymous negative was an attempt to reflect existing site guidelines. My apologies if terminology is a distraction; I did add "censuring" as a better word.

>mods' argument that meta is bad for actual conversation

Up/down votes are currently only public on the receiver's side, and this meta gamification powers HN. Would revealing the other side of the equation deter abuse of anonymous control (with an acceptable level of side effects)? All I can answer for sure is that intervention for downvote abuse (if any) is behind-the-scenes for now.

[I assume] the pool of eligible downvoters continues to grow; maybe when the time comes the next step will be another moderator rather than crowdsourcing -- the precedent of hiding individual comment scores seems relevant.

[mkobit]

On StackOverflow reputation is subtracted when downvoting answers but not questions [1], which is a neat approach. On your own profile, you can see the proportion of upvotes to downvotes (as well as other activity), but others cannot (maybe moderators?). So, there is at least some personal impact to having a user be accountable for a downvote. There is also generally a culture of "if you downvote, you should explain why", which sometimes seems to be here on HN as well. I remember Disqus removed the downvote count some time back, and Facebook doesn't have a "Dislike" button, so there are different approaches to it. I wonder what some other approaches some communities have taken, and what would be effective? Or even, what is the desired impact?

[1]: https://meta.stackoverflow.com/questions/251610/why-does-dow...

[2]: https://disqus.com/home/discussion/supportqa/why_was_the_dow...

[KajMagnus]

I'm developing Talkyard which has its own voting system: there's the Like vote, and the Disagree vote.

But the Disagree vote isn't a downvote — it doesn't affect karma or sort order. It justs shows how many people disagree about something. That can be good to know, so one can avoid harmful advice. Or just because one is curious about others' opinions.

There is a downvote too, actually, but only available for staff and core members. It works a bit like the downvote here at HN (e.g. removes "karma" and moves the comment downwards, dims it a bit).

Here's how the disagree vote looks: https://www.talkyard.io/please-note-2

[Jtsummers]

Downvotes could remove from users karma. It acts as a constraint. Basically, the more popular your posts the more moderation you can do, but you still lose moderation capabilities over time if you use it excessively.

No need to make it visible why a user's karma is dropping, it just is. Could be they're getting downvoted, or they're using their voting action and reducing it themselves.

The only other option that works well is a meta-moderation feature like /. has, but that's impractical for this site, the way it does things.

[pvg]

It acts as a constraint.

I doubt that really happens, other than for that very brief moment and tiny minority of users who are right around the threshold. The karma losses are capped, it's hard to lose meaningful amounts of karma without actively trying to.

[Jtsummers]

Sorry, I reread my post. What I meant was that if I downvote someone, the mechanism could deduct from my karma (or a fraction of my karma) as well as theirs.

So we're both at 1000 karma, I downvote you. You go to 999, I go to 999.5. Repeat. If I'm only downvoting, I'm forfeiting my karma to that moderation action. Optionally the same for flagging and other things.

This becomes a constraint because a user who just got to 500 karma and downvotes would lose the privilege immediately. A user at 10k karma can downvote pretty much everything. You could also have it cost more to downvote more in a period of time. Like 0 karma to downvote once an hour. 0.5 karma for two or three times an hour. 1 karma for four times an hour. Some scale (linear or faster).

This would prevent or mitigate people going through and downvoting all of a particular commenters posts in a thread just because it's controversial or they wrote one controversial thing. It creates a cost for moderation.

[pvg]

Oh I see what you mean. You can see a system like that in action on Stack Overflow. I don't think it really does very much at all there either, under what seem like much more favourable conditions: harder to accumulate karma, lots and lots of permission thresholds, a community involved in far more moderation than here and piles of caps on all sorts of user actions.

On HN, this would also favour commenting participation over curating participation. It's not obvious this is a win over the way things work now.

[mturmon]

And if you didn't like the notion of 1/2 karma, you could flip a coin to determine whether to subtract 1.

[Y_Y]

I always thought that a user's votes should be normalised by something sensible, like total number of votes or time spent online (with sensible scaling). That way opinionated clickers don't have outsize influence.

[aaronarduino]

If either intended or not intended, talking about the downvoting system seems unnecessarily taboo(gets downvoted). Which is a shame because there is an interesting effect that it has on the community which would be worth looking into.

One of the biggest questions I have: are downvotes an accurate way to moderate low-quality comments? I'm not sure how that can be proved by data but it would be interesting to see someone prove it.

[digi_owl]

I find myself thinking of the Slashdot system (also used by Soylentnews, a Slashdot "spinoff"), where people get to "flag" a certain number of comments on an article they themselves have not participated in. And said flags can never lift the post above +5 or send it below -1, but the most common flag given to a comment is represented next to the score.