Comment by altairprime
9 days ago
Social media isn’t always about consuming content. It’s also about getting jolts of momentary joy and reward. You get those in two ways: seeing cool things, and participating in cool things. Especially cool things before they go viral. Clicking like on a post that isn’t viral yet, and gambling to yourself whether it will go viral, has the same dopamine flux when it pays off as winning at the slots. Even my reward-defective brain manages to eke out a moment of reward from that. So if you simply remove the content, what’s left is the gambling market. Gambling on something you upvote going viral isn’t about how much content there is in what you placed your bets on, it’s about being able to have that special knowing look when someone tells you about it because you’ve just won the socio-memetic lottery. And AI isn’t doing anything whatsoever to stop that reward loop.
I proposed once a while back that we should have the HN admins strip all integer counts for a week server-side, to see if the site quality improved or worsened during that time. The mods suggested I ask HN, so I did. HN loathed the idea of it, for every possible reason except this one: removing all those integers would be like quitting gambling cold turkey after years of pulling the vote lever every day. I’m not much less vulnerable to this than everyone else, but I still want to see it happen someday. I remain reasonably confident that our social media site’s quality would skyrocket after a couple days of our posts and comments being disinfected of make-integer-go-up jackpots.
I find this idea of getting in early interesting, because it is completely novel to me. Is it common for people to derive so much pleasure from voting for something before it gathers momentum? You really lean into this idea, likening it to winning a jackpot, so I assume it is at least somewhat widespread.
The ability to make accurate predictions has been rewarded for so long that we now reward it even in the abstract.
By stopping integer counts do you mean not collecting upvotes and downvotes at all or just not displaying them?
If it is the later, I think it can be an interesting experiment, although I doubt it matters that much because you can still gauge the "engagement" based on how your posts rank. But there is absolutely no way HN could work if posts and submissions stop being sorted based on their votes. Community moderation via voting is what allows HN to remain functional despite having only two moderators. If votes stopped mattering for a week then HN will likely be flooded with spam by day two and the experiment will be halted by day three.
Just not transmitting the numbers to clients. I’m aware that a dedicated actor could try to infer their effects through study, code, etc., but I’m interested in the effect on the majority of people rather than those adamantly intent on counting their numbers.
Voting combined with voting on the votes/voters worked quite well in the olde days of yore when Slashdot used such a system. You did not 'vote up' or 'vote down', instead you voted things like 'insightful' or 'overrated'. Some of those categories caused the vote count to go up, others caused it to go down. Users decided for themselves whether they wanted to see all posts or only those above a given threshold. Then there was the meta-moderation system wher a rotating cadre of users could flag abusive votes. If a user got too many sch flags he lost his voting rights. This latter system would be good to have here as well given that I've seen a lot of abuse of the down-vote button where all recent posts for a user who has voiced an opinion outside of some desired narrative get voted down no matter their subject. Such abuse would be caught if there were a meta-moderation system in place. It would help reduce the group-think which is seen on sites like HN and Reddit.
A similar voting system is - or was - in place on lobste.rs where a reason for voting down needs to be given. It does - or did - not have meta-moderation though which takes away the possibility to get rid of vote-abusers.
Is moderation truly accomplished via votes, though, as opposed to flagging?
You say "spam", but unpopular comments are rarely promotional. They lack any one unifying quality. They might be naive, and/or they might be difficult.
Flagging doesn’t report any numbers to clients, so any potential changes to it are out of scope re: my client-accessible integers concerns. There was a big thread about the flagging system in the AI rule-change post a few weeks ago that may be of more interest to you along those lines, though!
https://news.ycombinator.com/item?id=47341705
I especially appreciated dang’s reply, quoted here out of context as a teaser:
> We're going to add that.
The vote counting thing can be interesting.
There's the classic "I wish facebook had a dislike button" or the equivalent for twitter.
But in the thread-based forum context, removing the downvote has interesting effects. For one, it stops people who down-vote-brigade to lower visibility. It also stops the "I don't like that guy" engagement and works on a more positive "I appreciated this comment" mode.
It's not one-size fits all but I've seen positive effects on more marginalized forums.
Oh, I have no objection to the voting buttons — just to us users being able to see the underlying numerical outcomes of us using them.