[chansiky]

My opinion is downvotes should require a reason for downvote. Upvotes imply aggreement of the original post therefore do not need further explanation.

Getting in the weeds here, but then if others agree with the downvote rationale, they can upvote the downvote which can be used as some metric. This already happens on HN and reddit - people downvote all the time then provide a reason, but not all people provide a reason for their disagreement, allowing for useless downvotes that dont help the community in any way - for example an "ad hominem" downvote. What this means is that disagreeing opinions are allowed attacks without any risk of retaliation, and in what game is that ever fair?

[iaml]

Yeah, I agree it would be interesting. Very often I'd like to be able to reply to everyone who brigade-downvotes things on reddit when for example they don't have some context or something counterintuitive is actually true according to reputable source.

[randomdata]

> Upvotes imply aggreement of the original post therefore do not need further explanation.

Agreement isn't useful feedback. An upvote implies that the comment is funny, a virtual laugh if you will, which is useful feedback to allow you to hone the entertainment value of your comments.

I suspect the next person will have a different take, though. Which ultimately means that the votes mean nothing at all beyond that someone, assuming no bots are at play, pressed a button.

Therefore, votes are just a poor man's analytics system. They give some vague feedback that someone was near your comment and nothing more. Which button was pressed makes no difference.

[ddingus]

The downside of that is increased meta.

And the reason is simple:

They did not agree and or lack shared perspective before the down vote and that being true after the downvote starts another disagreeable discussion.

It all gets worse when people know who downvoted them...

[chansiky]

Only the initial downvoting commenter would have to reveal his name. Then others can just bolster the downvote which harms the rating of the original comment, sort of like a "let him without sin be the first to throw the stone" situation. Then others can join in if they agree. And that might hurt if they're throwing stones at us but at least we get to see where the stones are coming from. Currently we dont, we just get stoned and dont know why.

[ddingus]

Secondly, being told why doesn't help as much as you would think. That itself is another meta discussion. People won't take the why as given. They will want to argue it, challenge the qualifications of others, and you can go down the list. It all happens.

People I see taking that feedback to contemplate is rare.

The number one reason why people want to be told why so they can argue their case. And that's what's going to trigger The Meta.

On Quora this problem happened and a bunch of us created a court where someone could appeal downvotes, and sort of argue their case. That discussion would involve feedback to the person who got downvoted, and if it all didn't warrant being downvoted, a whole bunch of people would up vote, basically correcting a wrong.

It was a super interesting exercise in the genuine ambiguity text communication has. The fact is we really can't determine intent from text. People can, and we'll take things all sorts of interesting and crazy ways.

And whether they are correct in doing that or not, the discussion to sort it out is laborious and time-consuming.

Now I will tell you, a bunch of people working hard on this problem and sort out a disagreement or downvote that shouldn't have happened. And maybe a third of those were worthy exercises in that the person who initiated the process took away something that genuinely help them to improve.

Another big percentage, were just fixing bad down votes. But it cost a lot to do that.

The rest were fairly painful meta discussions. Unproductive.

No the reason that all made some sense on Quora was the downvotes had Fairly severe consequences and fairly rapidly.

Here, it's not such a big deal. Anyone who makes more good decisions about what they say and not will gain karma. That's all that's required.

So rather than Stones, it's more like little Pebbles.

Frankly, the easiest way to get past this here is to just not worry about it, and try to be a good human.

In a lot of cases, people who get some downvotes here probably weren't trying their best to be good human. If they try harder they get less downvotes. Good As It Gets.

[ddingus]

I'll end with this:

The reason why moderation works here as well as it does, is because dang and others actually care, actually spend the human time interacting with people, and spend time cultivating norms and culture that leads to more decisions that are good than not.

This is a human problem, not something we fix with a rule or algorithm or clever metric.

Someday, maybe when machines can derive real meaning from text, we can revisit this discussion and be productive.

And I even have an indicator for you.

A while back the decision was made to include one space between the period at the end of a sentence, and the capital letter at the beginning of another one. The result of that is also a capital letter after the period required for an abbreviated word.

This ambiguity is why those of us who prefer two spaces at the end of a sentence do so. It is so software can understand when a sentence actually begins. As things stand now, there's no real distinction between the abbreviated word, and the legitimate end of a sentence, meaning we get autocapitalization wrong.

When machines can understand meaning well enough to sort this out, is also the time that we might revisit moderation. Cheers

( going back to two spaces would be really nice, but this discussion just gave me a reason to prefer one space now for the indicator purpose mentioned above.)

[chansiky]

Thanks for the write up, just wanted to note that I read through all of this. I get what you're saying about the meta now, clearly I didn't fully understand why that was bad thing.

Speaking of meta though, I have to say, I still stand by my initial thought that an alternate rating mechanic is at least worth exploring, as the upvote/downvote feels like a system from an earlier era before anyone realized how large its social impact could be. I'd be interested to know what the ideal system that promotes a healthy convergence to the center (rather than one that increases polarization) would look like.

I can't deny the effects of great moderation/cultivation, that they are the most important part, but being a technologist I still want to see what happens when the variables are tweaked.

[ddingus]

Well, for example here, just how large is the social impact?

And where it is large, and here it just isn't, why?

As for tweaking...

What center? Along how many axis?

[ddingus]

I have similar thoughts and have had the pleasure of participating and or moderating in a few very different scenarios.

Frankly, the product of that in part validates your desire better means and methods. There probably are some, and I believe analysis of higher order effects can filter many out.

I shared the other product, which is our ownership of our conversations. Shockingly low numbers of people understand the options available to them, and shockingly high numbers desire control of others as a potential solution to conversations they find disagreeable.

Norms are quite possibly the most powerful tools available to us because those speak more to what people can control.

The various systems we can invent tend to put an illusion of control over others, and or where they do actually control others they tend to be very expensive.

Consider the block.

Block someone and suddenly they just are not a part of your conversations, or your club, if say the block is to a group.

The blocked person, when individually blocked can basically carry on talking with everyone else. Often, that conversation is both invisible to the blocker as well as about them. The meta resulting from that is amazing! See Twitter.

Frankly, the block is more for the blockers benefit than it is for the person so blocked.

Where group blocks happen, they overlap with bans. This does a far better job of controlling others, but it then can involve others who had no part otherwise. Can force people to pick their friends, maintain confidences, lose touch with family, all sorts of higher order effects in play. All generally expensive.

Worse, is the reality of being offended, or breaking a rule, perhaps unclear, like "be nice."

We are each as offended as we think we are. There is no objective measure beyond coarse boundaries we find the hard way and those tend to propagate as norms. Interestingly, people will form clubs to avoid norms. See Reddit. Discord.

As I wrote earlier, and as a proponent of taking ownership of my conversations, I weigh speech and am difficult to offend and or angered by what basically rando people my tell me online.

When a clown calls you out as an ass, that is as laughable as it is low value. Who are they and what do they really know? Here is the insight hard won:

A good chuckle, coupled with a meaningful response that gives the clown an out to up their game is powerful and resonates in a community positive way.

Cries of righteous indignation also have power, but resonate in a community negative way.

The former meta has value and can yield insight and set strong norms community wide. Entertaining too.

The latter meta is low value, is a who is the bigger asshole type chat, and will set strong norms that offer future low value. Can be entertaining.

Negative entertainment is super easy because of how those play out.

The common thread here is meaning and how we come to know our minds and those of others through ambiguity. Text is pretty terrible. Understanding intent is difficult.

Norms are used here to great effect. There are a list of rules and I confess to not reading them. I do not have to. The norms here are very clear, and consequences generally scaled to feedback, but not harm, or be expensive.

So people can explore a little and find their way with few worries.

Here is an observation:

Moderate on value.

A troll, for example, can obtain high value for a very low investment in many places.

One post can get thousands involved and the troll is entertained for a song.

To the community, that one post was expensive!

So what to do?

I, with some others, employed the concept of value and norms to pretty great effect.

We required toxic people to include a benign phrase in their posts, which were otherwise allowed.

They hated that, but also talked about it. The norms were to help people add value, not be toxic.

It also inverted the entertainment. Suddenly the troll was not entertained and many members were!

If it escalates, then deny them 4 letter words. Five, three, less? More? Ok. Just no 4 letter ones.

And so on.

With the right norms, a group very quickly becomes inoculated against the worst, yet can still converse and ideally gain a member in good standing.

The concept was simple: they paid a tax when their contributions cost the community more than they were worth.

The moment they end that practice, no more tax!

This kind of thing works best when a significant body of members knows how to own their conversation. They know to laugh, or advise one to reconsider and know when to avoid pages of righteous indignation and or who is the bigger asshole.

Sidebar: all members of a conversation about who is the bigger asshole deserve that conversation. (Very strong norm to establish there)

Sorry, but my ramble does get back to your tweaks:

Rather than upvote/downvote on agreement, do so on value.

Any system that can collect value may also have resonant higher order effects and I will leave you with the idea of healthy resonance, that is appropriately damped as "the center" you seek.

Undamped resonance is an echo chamber.

No resonance is a support forum.

Here we see things resonate, but not as one thing, more like chords, somewhat harmonious, not discordant.

Simple up and down, coupled with norms can do that. And one secret here is the non obvious dampening.

High value resonant speech happens and is encouraged. Discordant things are not denied, just nudged away.

I personally never downvote. It is not needed. I like flag to get at speech with toxicity potential, but my own bar is high.

I also like vouch. Same reasons.

These hint at ways to communicate value and I very strongly suggest value is where the better systems exist, if they do exist apart from skilled humans who, unlike machines, can deal in meaning and come to know minds.

[ddingus]

And third, discussions are at least two way affairs.

Often, moderation is seen as the objective parent that keeps everything in bounds.

The reality is, our individual boundaries very considerably.

We all have a shared responsibility to not allow discussions to go bad, and how we respond to text we don't like determines whether they go bad.

Fact is, very few of my discussions go bad, because I don't allow it. Importers not allowing it is not worrying about the downvotes, and instead focus on what I can control. And I control me, not anyone else in the discussion.

Someone tells me to fuck off here for example, I'm going to ask him why. I'm also going to advise him to edit that away before they get down voted, because we've got more productive things to talk about.

There are many similar ways to handle these things, in very few people actually employ them.

I feel spending time on that is as productive, if not more than time on more effective moderation.

[ddingus]

First, As soon as we introduce who, then we introduce meta about them. And doing that ends up a discussion that all of us will be impacted by.

There aren't easy answers. Pretty much everything in this whole discussion has been done, impacts easy for anyone who wants to look to see. And I have. This is a topic of great interest to me. I have moderated in the past, and found it difficult and challenging. I've learned more here than anywhere else.

( I did three, now four, replies here, partially because I'm using voice dictation, and partially because there are three major, potential points of discussion.)

[spiffytech]

Tildes.net tries something interesting here: there is no downvote, but users can flag comments with one of the provided negative attributes (offtopic, noise, malice)[0]. This discourages "downvote to disagree", and provides enough signal to e.g., sink an off-topic comment, but summon a moderator for malicious comments.

[0] https://docs.tildes.net/instructions/commenting-on-tildes#la...

[chansiky]

Thanks for sharing, this is a really interesting thought and its very close to one I had based on this thread, with the one difference being that instead of letting others decide what kind of comment one makes, we should really lean into thinking of commenting as a sort of game and let the user decides what kind of comment they are making, by letting them choose to attack/defend/support a position.

We've seen how the upvote/downvote w/ranking systems play, I'm not a big fan of what reddit has become lately, and I wonder where HN would be without great moderators. At the very least it would be refreshing to see how alternative mechanics lead to different communities.

Sort of related is this thought, Wikipedia is an MMORPG: https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_is_an_MMOR...