Live data from Hacker News

Why do companies with unbounded resources still have terrible moderation?

moultano.wordpress.com

71–80 of 159 posts

Re: Why do companies with unbounded resources still have terrible moderation?

#71
post #38

Earlier quoted context omitted.

Reddit mods aren’t chosen by the community. Reddit isn’t even mentioned in the article.

Do you mean admins? Because mod teams are self-selecting once a subreddit is up and running and if you don’t like a subreddit you can always start your own. People don’t vote for mods but they participate in subreddits they like. In Albert Hirschman’s typology it’s all exit, voice and loyalty are irrelevant.

Doesn't Reddit plant their own mods into important subs? As soon as a sub starts to regularly hit the frontpage it is forced to only accept advertiser friendly content or get blacklisted/banned.

Re: Why do companies with unbounded resources still have terrible moderation?

#72

The article concludes with: 1. Effective moderation can only come from others steeped in the particular cultural context they're part of. 2. Effort should be given to moderate to give communities the ability to moderate themselves at scale. With that in mind, it may be worth considering Reddit. Each subreddit has it's own moderation team from the community that is charged with enforcing sitewide guidelines like not d…

I do think Reddit is particularly well suited to having good moderation by virtue of its structure, and Twitter is particularly poorly suited. Twitter is all about decontextualizing what people are saying.

I think a useful somewhat-apples-to-apples comparison for moderation is: Quora and StackExchange. Both websites are intended to be Q&A platforms, but their approach is markedly different and results in different moderation issues.

Quora is heavily centralized - all content must be moderated by the company, there is little to no segregation of questions by dedicated community, and all questions are allowed. Individual users must be name-identified and, importantly, are the core ingredient for Quora - people come to read the content of specific people, not for quality of the content in general. Users have stake but no say in the evolution of the system, and are by design incentivised to not care about question or answer quality but instead about the people they engage with. Further, questions are considered to be "owned by everyone" - so users are free to alter questions indefinitely.

In contrast, StackExchange (not StackOverflow) has similar properties to Reddit - partitioned into smaller communities, reliant on user moderation, equipped with a somewhat reasonable appeals process that relies on a "meta" site where users can contribute to improving the site. Users have both stake and say, and importantly are dedicated to improving the average quality of answers rather than who is answering.

The difference in moderation is stark. Quora's issues with moderation have been legendary - anywhere from an inability to stop sockpuppet voting, pay-to-upvote rings, and reversing helpful edits to outright refusing to take action against documented sexual harassment by users against other users. StackExchange has had no such issues, or rather it has had vanishingly fewer issues compared to Quora.

While we obviously can't compare them on equal footing, given that the sites are so different, this is some circumstantial evidence that community moderation, allowing communities to manage their evolution, and removing the focus on individuals is a better experience between two sites that seek to accomplish the same goal.

I'm not sure what this would say about social networks - which are all about emphasis on other people - but it does seem like a great model for online communities that aren't social network focused.

Re: Why do companies with unbounded resources still have terrible moderation?

#73
post #68

There are two classes of users: normal people who want to use your service and people who actively weaponize it for their own ego. Those latter users have all the time in world and can defeat any attempt to stop them. This is because, online, time is the only resource that matters. You have a limited amount of this resource. Your company has a limited amount of this resource. Your normal users have a limited amount o…

They couldn't ruin HN because of the system of voting, flagging, and high karma requirement for downvoting. And this applies in general, the only working way to moderate is distributing the effort between users who are interested in the final result. Ideally big sites would not have one single system for moderation, but multiple lists so that if two groups keep flagging each other they get separated and don't see oth…

I don't know.

HN has plenty of misinformation posted on it constantly. Countless posts that are objectively wrong, and in many cases it is done intentionally. I have seen lies supporting eugenics, lies about how quantum computing is a fake conspiracy, etc etc. And yet the mods only jump in when you're impolite by referring to OP with "you" and "you're" because it is a "personal attack". They are optimizing on the wrong problem, IMO. I would rather discuss with people who are snide rather than with liars who have an agenda.

HN has its own "fake news" problems. And this is the society we are living in; you can say anything as long as it doesn't offend or hurt someone's feelings. Even if it is objectively fake horse shit.

Re: Why do companies with unbounded resources still have terrible moderation?

#74
This is why moderation should stick to the "clear and present danger" standard established by US courts. There has to be a credible threat of violence directed at a specific individual. "Hate speech" is too hard to distinguish from political speech. Don't try to do it in advance. Prior restraint is bad.

Banning people from being excessively annoying over a period of time is perhaps an option. That's after the fact, not before.

Re: Why do companies with unbounded resources still have terrible moderation?

#76
post #62

Earlier quoted context omitted.

>They going to abuse your platform to the extreme to promote their views. Facebook and twitter _encourage_ people to abuse their platform, that's why you can write 100 tweets from one account on a single day and they don't try to stop you at all, they know all that content will show up in their statistics, it will also make other users reply to those comments and so on; so in the end they have all incentives to keep…

IP banning is not effective in an era of disposable VPNs and Tor. I mean you could try to keep up with Tor exit nodes and VPN gateways but pretty soon you'd be banning all of Azerbaijan or something.

Ip banning for new accounts together with creation time exceptions seems pretty good thought: almost all existing users would be unaffected. Only costs you growth untill you find some extras to put into the account creation process.

Re: Why do companies with unbounded resources still have terrible moderation?

#77

Earlier quoted context omitted.

You might not even know it if you see it! It might be dog whistles that you don't even understand until it's pointed out to you.

Check out old Mae West movies where she adheres to the letter of the Hayes Code yet always finds a saucy way to get her point across. A couple classics: Reporter: "What do you think of foreign affairs?" West: "I'm all for 'em!" "Is that a gun in your pocket or are you just happy to see me?" Language is infinitely malleable. Algorithms don't stand a chance.

> Language is infinitely malleable

I mentioned this before but my GF worked on an mmo's for tweens. Said one experiment with filters ended with a 12 year old boy writing, I'm going to stick my giraffe up your pink bunny.

Re: Why do companies with unbounded resources still have terrible moderation?

#78
post #69

because Free. If we hadn't invented free, we'd have moderation because being repeatedly abusive and troll live for money, is in the end self defeating. If you want moderation enough, get rid of free.

If it is not free, it is a walled garden.

Maybe the optimism people had of the internet being an all inclusive community was ill fated.

Exclusivity is may be the next fad, driven by the trolling of few in an open community.

Re: Why do companies with unbounded resources still have terrible moderation?

#79
post #68

Earlier quoted context omitted.

They couldn't ruin HN because of the system of voting, flagging, and high karma requirement for downvoting. And this applies in general, the only working way to moderate is distributing the effort between users who are interested in the final result. Ideally big sites would not have one single system for moderation, but multiple lists so that if two groups keep flagging each other they get separated and don't see oth…

I don't know. HN has plenty of misinformation posted on it constantly. Countless posts that are objectively wrong, and in many cases it is done intentionally. I have seen lies supporting eugenics, lies about how quantum computing is a fake conspiracy, etc etc. And yet the mods only jump in when you're impolite by referring to OP with "you" and "you're" because it is a "personal attack". They are optimizing on the wro…

In your example if someone pointed out that quantum computers required some equations to hold with far greater precision than anything measured, so there is small but non-zero probability that it will turn out quantum computers can not work, would it be a lie that moderators should delete?

I believe it's not moderators job to decide what is a lie and what is not, if you see a lie, write a truthful comment refuting it, that way you can convince people instead of trying to silence.

Re: Why do companies with unbounded resources still have terrible moderation?

#80
There are really two issues here, both of which make it difficult for these sites to moderate their 'communities':

1. Moderation doesn't scale. It can't be automated with an algorithm (well), it can't be done well by a bunch of low paid contractors with a handbook and it can't be done across a million user platform by about three people.

Sites like Twitter, Facebook, YouTube etc tried to get round this by using the above algorithms and contractors, and that's partly why they're so terrible at it.

2. Moderation is community specific, with each community having different standards for what's acceptable. For instance, a site like 4chan or 8chan may find nearly anything acceptable, whereas the official Disney or Lego site might only want stuff that's appropriate for a young audience.

That's why forums work so well, because the userbase and owners decide what's acceptable there, and those outside of said grouo can join/not join based on their preferences. Same with Reddit to some degree, though that is complicated by people being able to move between subreddits so easily and them being hosted in the same place.

Unfortunately, many of these larges don't have this. They have one free for all open space where hundreds of different communities are forced to interact with each other. They stick millions of people into a newsfeed/search/whatever that despise each other's views, have completely ideas about what's acceptable and hate every ounce of the other side.

That's a recipe for disaster. Putting those who fundamentally hate each other in the same area for hours per day leads to a toxic atmosphere, and creates a situation that's almost impossible to police well. It's why school and prison are so bad; the kids and prisoners just have too many fundamentally different, conflicting worldviews to get along.

The only way to fix this is to:

1. Accept that ommunities should be moderated by their members, not robots or contractors.

2. Split the community into smaller ones where people can set their own boundaries for acceptable behaviour.

3. Try and stop intercommunity trolling as best as possible

Post reply on HN