Comment

Comment on ‘This shouldn’t be normal’: developers speak out about bigotry on Steam, the world’s biggest PC gaming storefront

sp3ctr4l@lemmy.dbzer0.com ⁨5⁩ ⁨days⁩ ago

Does anybody wanna know the actual mechanics of why Steam is poorly content moderated?

Its because they primarily rely on automated systems, and a very, very small team of inhouse moderstors/admins, as opposed to other comparable platforms (social media networks, basically), that have armies of contracted moderators in low income countries, whose job is to get more and more PTSD every day.

Thats how platforms with comparable amounts of user content do moderation.

Valve absolutely should devote more time and energy to restructuring stages of automated review for user posted comments and content, to improving those review processes, and honestly, should probably just sunset the Steam Forums system, and rethink an entire new approach to it.

But… at the same time, the scale is a significant problem.

Steam has a comparable number of overall daily active users to a major social media platform.

… and the ones that do content moderation, well, they have armies of poor people manually reviewing everything, getting PTSD from that work, and nowadays, likely training an LLM to be a better auto content moderator.

Genuine question for everyone: Do you think that’s an ethically justifiable solution to the problem?

Other genuine question for everyone: What actual technical solution do you think should be implemented?

source

Sort:hotnew top

kilgore_trout@feddit.it ⁨4⁩ ⁨days⁩ ago
I believe the answer is simply to give better moderation tools to the developers on their own games’ Store and Forum pages, since it’s developers who seem to have an issue with current moderation.

source
- pory@lemmy.world ⁨2⁩ ⁨days⁩ ago
  This solves the current problem but reintroduces the one that steam reviews exist to solve: giving the game’s developers control over the most visible discussion channels for the game allows for removal of negative reviews or user backlash. Think about how bad subreddits can be about “removing toxicity” after a GAAS cranks the monetization dial up when the devs are on the mod team.
  
  source
  - kilgore_trout@feddit.it ⁨2⁩ ⁨days⁩ ago
    It could allow to hide the content of the review, but still count it in the total (recommended / not recommended).
    Personally I am not in favour, and I see negative bigoted reviews as legitimate reviews. I wouldn’t hold on the same level professional reviews, but it’s only random players we are discussing here. Let’s not pretend that the positive reviews are always constructive, either.
    
    source
    pory@lemmy.world ⁨2⁩ ⁨days⁩ ago
    Yeah, but if you can remove negative reviews text but not the contribution to “mostly positive” or whatever, the audience has to take it on faith that you “only censored the racists don’t worry. We’re getting brigaded”
    
    source
- Katana314@lemmy.world ⁨2⁩ ⁨days⁩ ago
  Even if they had it, a lot of smaller developers don’t even want to be serving as chaperones for their playerbase. Some have even said they don’t want their game page to create a Steam subforum.
  
  source
  - kilgore_trout@feddit.it ⁨2⁩ ⁨days⁩ ago
    
    Some have even said they don’t want their game page to create a Steam subforum.
    
    Which is allowed, as far as I know?
    
    source
- sp3ctr4l@lemmy.dbzer0.com ⁨4⁩ ⁨days⁩ ago
  Well ok, that sounds reasonable to me!
  
  What kinds of tools do you mean?
  
  Like, I’m not trying to be duplicitous, I genuienly want Steam to not be a cesspool.
  
  partner.steamgames.com/…/community_moderation
  
  There’s an overview of what currently exists.
  
  Yeah, a lot of it is based on having to manually flag things as harassment or bigotry or something like that, especially when it comes to actual game reviews, and it is obviously the case that whatever automated systems Valve currently has in place to auto flag things… are not sufficient.
  
  And just for more context, here is the feed of Steamworks itself, which… more or less, is the sprt of update pipeline for Steam itself, as game devs would interface with it, which is also the system that would be the thing getting updated with new content moderation concepts.
  
  store.steampowered.com/news/group/4145017
  
  source
Aielman15@lemmy.world ⁨4⁩ ⁨days⁩ ago
Your entire comment reeks of “we shouldn’t fight fire because that puts firefighters at risk”.

There are no 100% ethical solutions to every problem, real life is a compromise. You can get better ethical results by allowing those workers to get adequate monetary compensation for their work and seek medical help if they need it. Otherwise what’s the solution, allow everyone to read the same stuff? Why is that more ethical? Is it more ethical for the random user (who may also be a suggestible kid, or a person belonging to a persecuted minority) who reads that stuff? Is it more ethical for the developers who get their game review bombed by fascists and bigots, and see their source of revenue diminish or fizzle out because of it?

As for the legal responsibility, it becomes so when the platform is complicit with the users writing hateful stuff. You are not responsible for the random shithead declaring his love for Mein Kampf. You are responsible for the hundreds of users who do while repeatedly ignoring the reports of their misconduct, thus implicitly accepting and normalizing their behavior.

source
- sp3ctr4l@lemmy.dbzer0.com ⁨4⁩ ⁨days⁩ ago
  I think thats quite an unfair characterization.
  
  Primarily because firefighters, firefighting, tends to be a fairly exclusive field, that requires a lot of training, that tends to pay pretty darned well.
  
  Whereas the armies of content moderators tend to be incredibly poorly paid. The entire way this kind of work is done is that it nearly always either entirely or largely is done by the lowest bidder, in the poorest places possible.
  
  As compared to firefighters, who… at least in terms of municipal firefighters, well that tends to be fairly local.
  
  (* * * With the massive glaring exception of using prisoner labor to fill in gaps in often extremely dangerous firefighting conditions, which is more comparable to exploiting those who don’t really have better options * * *)
  
  I am pointing out that yes, the problem exactly is that none of the potential solutions here are ethically wonderful, that this is not a kind of ‘oh well obviously they could just do this simple and easy fix and everyone would be happy’ kind of situation.
  
  So… your ethical calculus seems to conclude that stopping the spread of bigotry and fascist rhetoric in richer countries is worth the cost of the sanity of workers in poorer countries.
  
  Your ethical calculus seems to be that if 100s of users of a website/platform don’t get banned rapidly for violating TOS, then the website/platform should be held legally liable for that, which would mean that you believe that basically every website platform with over half a million DAU, that doesn’t use a complex layered system of LLMs with absurd economic and environmental costs, that they should all be sued or fined into non existence.
  
  … Unless you maybe want to clarify more exactly what you mean here.
  
  You also don’t directly address at all the idea of using an LLM for these tasks… which is what all of the megaplatforms with much more active consistent, rapid, and often overzealous or erroneous moderation do.
  
  I’m just trying to present the actual totality of the moral ramifications of the involved systems and practices relevant to this topic.
  
  If confronting the actual ugliness of them challenges you, makes you defensive and accusatory, good.
  
  That means you likely never thought about the totality of the situation here that deeply.
  
  source
  - Aielman15@lemmy.world ⁨4⁩ ⁨days⁩ ago
    I already answered your questions, but you seem more intent at discussing abstract ethics like an armchair philosopher rather than the real problem at hand.
    
    Whereas the armies of content moderators tend to be incredibly poorly paid. The entire way this kind of work is done is that it nearly always either entirely or largely is done by the lowest bidder, in the poorest places possible.
    […]
    So… your ethical calculus seems to conclude that stopping the spread of bigotry and fascist rhetoric in richer countries is worth the cost of the sanity of workers in poorer countries.
    
    Why is the assumption that those workers must be poorly paid? If Valve, the multi billion dollar company whose owner owns multiple yachts as well as the company producing them, doesn’t pay its workers adequately, then Valve is at fault. The solution shouldn’t be to throw up hands and go home. There is a solution but they aren’t willing to take it because it would require them to spend money, which is what I said in my first comment.
    
    Your ethical calculus seems to be that if 100s of users of a website/platform don’t get banned rapidly for violating TOS, then the website/platform should be held legally liable for that […]
    
    You know damn well what I meant but you keep this enlightened bullshit going on.
    
    Valve literally got reports about those reviews and ignored them. They are at fault. Full stop.
    
    If confronting the actual ugliness of them challenges you, makes you defensive and accusatory, good. That means you likely never thought about the totality of the situation here that deeply.
    
    Please stop this enlightened philosopher bullshit.
    
    source
    sp3ctr4l@lemmy.dbzer0.com ⁨3⁩ ⁨days⁩ ago
    
    Why is the assumption that those workers must be poorly paid?
    
    Because that is how this always works out, in practice, under the currently existing economic reality.
    
    Human content moderators basically work on the same paradigm as Amazon MTurk’ers, as DoorDash drivers. Independent subcontactors of contractors.
    
    Its not a job that needs highly trained people. Its a job that needs a massive volume of people to handle a massive volume of content, then you have a tiny number of data scientists running analytics on associations or trends and patterns, to fine tune and condense an aggregate of many human decisions into systems of heuristics for an auto content moderating system.
    
    This just is how this industry works.
    
    Having some other kind of paradigm with say, less, more highly paid human content moderators is not scalable, it does not work to handle the sheer volume of shit being squeezed through the pipes.
    
    That is just the practical reality as it currently is.
    
    If Valve, the multi billion dollar company whose owner owns multiple yachts as well as the company producing them, doesn’t pay its workers adequately, then Valve is at fault.
    
    I mean, to an extent, I agree with you, see my original post that you responded to you where I said Valve should absolutely place more focus and effort on developing better automated systems and changing up the paradigm of how all of it works, so that its more effective.
    
    The solution shouldn’t be to throw up hands and go home. There is a solution but they aren’t willing to take it because it would require them to spend money, which is what I said in my first comment.
    
    Mhm. Yep. Anyway, your now seemingly proposed solution of… I dunno, paying Valves existing human moderator team more money?
    
    Won’t solve the problem.
    
    Or… do you want them to just hire thousands to tens to hundreds of thousands of human moderators?
    
    Because that is the scale of the # of humans Valve would need to hire.
    
    What should their wages be?
    
    Should they just burn through all of the money that they have set aside to throw at developing innovations in the industry, should they stop inventing/manufacturing Steam Machines, Steam Frames, Proton, etc?
    
    Valve is estimated to have a net worth of roughly $10 billion. Estimated yearly revenue in 2025 of $17 billion. No estimates exist of their yearly costs, as far as I can tell.
    
    So ok, lets say they have… $1 billion a year, they can devote to paying an army of moderators.
    
    If you pay them $25 an hour, thats $52k for a full time year. Thats 19,230 people.
    
    That also makes ~98% of their employees those people.
    
    Thats a massive shift to what the company even is.
    
    … this is another huge reason why companies use contractors for this.
    
    Also, 20k people might kinda help, but you’d probably need more like 5x to 20x that.
    
    You know damn well what I meant but you keep this enlightened bullshit going on.Valve literally got reports about those reviews and ignored them. They are at fault. Full stop.
    
    Did they intentionally ignore them?
    
    Or is their moderator team and automated systems maybe stretched beyond capacity, and needs a comprehensive overhaul?
    
    If you are talking about this:
    
    bsky.app/profile/…/3mc2eaesex22u
    
    Yeah that’s a bot response, an automated response, with a random first name attached to it.
    
    I’ve had several encounters with Steam Support before.
    
    This is how it works, you have to keep poking and pleading pretty damn persistently to get any kind of response that does not read like a standard boilerplate canned response to a set of basically most prominent tags as to the type of thing you are contacting them for.
    
    Now, is this a shitty thing to do, have a bot pretend to be a person?
    
    Yes.
    
    … its also extremely common, across all kinds of industries, to do this.
    
    Hence, why my initial comment, again, called for overhauling this system.
    
    Please stop this enlightened philosopher bullshit.
    
    You may note that I’ve made nearly 0 attempt at actual philosophy, an actual prescriptive system to define and evaluate morals and moral decisions.
    
    I am just describing the complexity and details of the situation.
    
    Meanwhile, to you, this is ‘rich company bad, can fix if wanted to’.
    
    … While I’m giving you the specifics of … well it matters what exactly the situation is, how you solve the problem is the important and also complicated part.
    
    Just identifying that a thing is bad and should be fixed is facile, obvious.
    
    Trying to come up with a potentially feasible solution, muchless an optimal one, is complicated, involves nuance and detail that you seem to be offended by.
    
    source
    -> View More Comments