4 ms·

This post was flagged, but I don't think it should have been and I hope it stays up. I have been grateful that HN hasn't had the same level of animosity about t

This post was flagged, but I don't think it should have been and I hope it stays up. I have been grateful that HN hasn't had the same level of animosity about the conflict in Gaza as other places on the internet. But this news article is about a core issue in tech, and should be discussed on HN. The whole world now uses social media to communicate, which means that when there's a war, partisans of both sides will be motivated to fight out the propaganda war on the platforms that tech built. I wonder if there will ever be a way for a global-sized platform which uses algorithmic content feeds to convincingly show moderation neutrality on a topic like this. The HRW report has 1,000 examples of peaceful pro-Palestinian content that was removed by Meta. But only Meta themselves has the information to know whether there is systematic bias. So we get a situation where people sympathetic to both sides both come away feeling that the platform is biased. I think the U.S. culture war has similar dynamics. My engineer brain wants to solve this by giving users more visibility into moderation decisions. Maybe statistics on deleted posts, more legible moderation rules, publicly posted justifications for moderation decisions. Another part of my brain thinks that maybe this is just an unsolvable problem for social reasons in a community that's global-scale. I'm not sure. I do feel like this is still Meta's job, part of their cost of doing business, and if they put something like 5% of their revenue into moderation the situation would be a lot better. In the meantime I find myself personally trying to cut out as many algorithmic feeds as I can from my life. I prefer the curation of a news website or podcast, small group discussions, or 1:1 communication with people I trust. Algorithmic feeds are easy but so far have never done a good job covering issues where there is real conflict between two groups.

throwaway55479 3y ago
That is "the beauty" of this conflict (if something this heinous could ever have a beautiful side): It poses many interesting technical challenges. But, quite understandably, people everywhere (including HN) are fast to move the "warring" right onto our screens (hence the flagging). So, as you mentioned, algorithmic feeds is one of them. There are many others like how to create temporary infrastructure for such a besieged, war-torn city [0]. I suppose many would say, "This goes for all wars, what is special about this one?" Well, I would argue this is a "special conflict". It is one of (if not THE) longest-running military occupation in the world [1], where, coincidentally, the US spends its largest tax-payers-funded foreign aid [2]. Edit: And because many are now commemorating the birth of Jesus Christ, it is worth recalling that it is also special for being that birth's place (spiritually, if not literally). [0] https://news.ycombinator.com/item?id=38673300 https://news.ycombinator.com/item?id=38673300 [1] https://en.wikipedia.org/wiki/List_of_military_occupations https://en.wikipedia.org/wiki/List_of_military_occupations [2] https://www.visualcapitalist.com/cp/which-countries-receive-the-most-foreign-aid-from-the-u-s/ https://www.visualcapitalist.com/cp/which-countries-receive-...
- nomdep 3y ago
  PS: For many (myself included), Christmas is about loving your family and not “the birth of Jesus Christ”, even if their mothers see it that way
bitwize 3y ago
> My engineer brain wants to solve this by giving users more visibility into moderation decisions. Visibility into moderation decisions makes them more gameable/exploitable/circumventable. The whole purpose of moderation is to remove unwanted speech; justifying the fairness of such removal is only secondary, and can be elided if it interferes with the primary goal. Hence "You have been banned because your posts violate our Terms of Service" messages with no further explanation, and things like shadowbanning. It sucks to be on the receiving end of these things but they make online communities much more pleasant.
- infotainment 3y ago
  Exactly; consider the "censorship" that everyone generally agrees with, spam filtering. If the precise nature of the spam filter is known (keyword lists, etc), then it becomes much easier to design emails that avoid being filtered.
  - fulafel 3y ago
    Spam filtering is the opposite problem of people trying to communicate their views on mass killings of civilians in an important respect: In the case of spam, you don't want the spammers to reformulate their message to get past the filters. In the case of mass killings, you do want to allow the social media users to tone down their writing.
    - infotainment 3y ago
      Sounds good in theory, but in practice I think you’d ultimately just incentive the kind of behaviors you already see on TikTok: users would just add asterisks to filtered words, instead of changing anything of substance.
- anigbrowl 3y ago
  This is basically an argument for getting rid of laws and ruling by fiat, with a judicial process that amounts to repeating 'you know what you did' whether or not that is honest.
gumballindie 3y ago
> My engineer brain wants to solve this by giving users more visibility into moderation decisions. Politics are not an engineer’s job, nor is content moderation.
heresie-dabord 3y ago
> this news article is about a core issue in tech I agree. Let's assume that a Social-Media Business (SMBiz) gains sufficient value from hosting topical discussion to want to continue. But there is cost, reputational risk, and lawyers are expensive. How can SMBiz effectively moderate a conversation/debate/flamewar about a controversial event or subject, especially when the SMbiz wants to maximise user engagement while respecting laws and government requests? Like it or not, that's the business reality. People who want a truly neutral discussion platform that operates with transnational freedom and total personal freedom will find that their ideals alone are not enough to sustain the business. Prior Art: People often say that HN's Dang and Slashdot's moderation system are the best examples of "effective moderation" that we know. If the moderation is effective, it means SMBiz is attentive to topics and comments and discussion is insightful and interesting; it isn't overwhelmed by frivolity, falsehoods, defamation, and death threats (FFDDs). If the job is done well, SMBiz may even receive accolades. See Reddit's recent experience with its moderators for a negative example. Here are some of the rules that will need to be applied, in stochastic FIFO order. Items are lettered for reference. a= At all times, keep a clear eye on the business equation. Effective moderation will have a significant cost. At any scale! b= Seed the discussion with verified accounts/users. Reputation must be earned and maintained. See Prior Art. c= Rate limiting, especially for controversial topics. No short, stupid Reddit posts. People must weigh their words. Sometimes, the microphone must be disabled. d= No flamewars, no propaganda, no falsehoods. See Prior Art. Lock any discussion that achieves high FFDDs. Throw away accounts, ban offenders. Make it sting. e= No anonymity unless a verified account/user has sufficient reputation to post in anonymous mode. But it costs in ReputationCoin. f= No bots, no bridages. See Prior Art to estimate workload and infrastructure. g= Enable and reward trusted moderators. See Prior Art. = = = Final Notes > My engineer brain wants to solve this I get that you are very interested in the Moderation Problem, but I think this wording wins no support. Backers will say that you are way out of your social depth, and people in the debate will dismiss your intent as naivety or bike-shedding (frivolous or misplaced optimisation) or a combination of the two. One important reason why HN succeeds because Dan G (who is indeed attentive, and principled) rejects the most ungovernable type of conversation: flamewars. > convincingly show moderation neutrality on a topic like this There is no neutrality, of course. SMBiz can state its bias and interests clearly. But it will be hard to make a profit from achieving neutrality because strident factions in polarised debate do not want neutrality.

throwaway55479 3y ago

That is "the beauty" of this conflict (if something this heinous could ever have a beautiful side): It poses many interesting technical challenges. But, quite understandably, people everywhere (including HN) are fast to move the "warring" right onto our screens (hence the flagging). So, as you mentioned, algorithmic feeds is one of them. There are many others like how to create temporary infrastructure for such a besieged, war-torn city [0]. I suppose many would say, "This goes for all wars, what is special about this one?" Well, I would argue this is a "special conflict". It is one of (if not THE) longest-running military occupation in the world [1], where, coincidentally, the US spends its largest tax-payers-funded foreign aid [2]. Edit: And because many are now commemorating the birth of Jesus Christ, it is worth recalling that it is also special for being that birth's place (spiritually, if not literally). [0] https://news.ycombinator.com/item?id=38673300 https://news.ycombinator.com/item?id=38673300 [1] https://en.wikipedia.org/wiki/List_of_military_occupations https://en.wikipedia.org/wiki/List_of_military_occupations [2] https://www.visualcapitalist.com/cp/which-countries-receive-the-most-foreign-aid-from-the-u-s/ https://www.visualcapitalist.com/cp/which-countries-receive-...

nomdep 3y ago
PS: For many (myself included), Christmas is about loving your family and not “the birth of Jesus Christ”, even if their mothers see it that way

bitwize 3y ago

> My engineer brain wants to solve this by giving users more visibility into moderation decisions. Visibility into moderation decisions makes them more gameable/exploitable/circumventable. The whole purpose of moderation is to remove unwanted speech; justifying the fairness of such removal is only secondary, and can be elided if it interferes with the primary goal. Hence "You have been banned because your posts violate our Terms of Service" messages with no further explanation, and things like shadowbanning. It sucks to be on the receiving end of these things but they make online communities much more pleasant.

infotainment 3y ago
Exactly; consider the "censorship" that everyone generally agrees with, spam filtering. If the precise nature of the spam filter is known (keyword lists, etc), then it becomes much easier to design emails that avoid being filtered.
- fulafel 3y ago
  Spam filtering is the opposite problem of people trying to communicate their views on mass killings of civilians in an important respect: In the case of spam, you don't want the spammers to reformulate their message to get past the filters. In the case of mass killings, you do want to allow the social media users to tone down their writing.
  - infotainment 3y ago
    Sounds good in theory, but in practice I think you’d ultimately just incentive the kind of behaviors you already see on TikTok: users would just add asterisks to filtered words, instead of changing anything of substance.
anigbrowl 3y ago
This is basically an argument for getting rid of laws and ruling by fiat, with a judicial process that amounts to repeating 'you know what you did' whether or not that is honest.

gumballindie 3y ago

> My engineer brain wants to solve this by giving users more visibility into moderation decisions. Politics are not an engineer’s job, nor is content moderation.

heresie-dabord 3y ago

> this news article is about a core issue in tech I agree. Let's assume that a Social-Media Business (SMBiz) gains sufficient value from hosting topical discussion to want to continue. But there is cost, reputational risk, and lawyers are expensive. How can SMBiz effectively moderate a conversation/debate/flamewar about a controversial event or subject, especially when the SMbiz wants to maximise user engagement while respecting laws and government requests? Like it or not, that's the business reality. People who want a truly neutral discussion platform that operates with transnational freedom and total personal freedom will find that their ideals alone are not enough to sustain the business. Prior Art: People often say that HN's Dang and Slashdot's moderation system are the best examples of "effective moderation" that we know. If the moderation is effective, it means SMBiz is attentive to topics and comments and discussion is insightful and interesting; it isn't overwhelmed by frivolity, falsehoods, defamation, and death threats (FFDDs). If the job is done well, SMBiz may even receive accolades. See Reddit's recent experience with its moderators for a negative example. Here are some of the rules that will need to be applied, in stochastic FIFO order. Items are lettered for reference. a= At all times, keep a clear eye on the business equation. Effective moderation will have a significant cost. At any scale! b= Seed the discussion with verified accounts/users. Reputation must be earned and maintained. See Prior Art. c= Rate limiting, especially for controversial topics. No short, stupid Reddit posts. People must weigh their words. Sometimes, the microphone must be disabled. d= No flamewars, no propaganda, no falsehoods. See Prior Art. Lock any discussion that achieves high FFDDs. Throw away accounts, ban offenders. Make it sting. e= No anonymity unless a verified account/user has sufficient reputation to post in anonymous mode. But it costs in ReputationCoin. f= No bots, no bridages. See Prior Art to estimate workload and infrastructure. g= Enable and reward trusted moderators. See Prior Art. = = = Final Notes > My engineer brain wants to solve this I get that you are very interested in the Moderation Problem, but I think this wording wins no support. Backers will say that you are way out of your social depth, and people in the debate will dismiss your intent as naivety or bike-shedding (frivolous or misplaced optimisation) or a combination of the two. One important reason why HN succeeds because Dan G (who is indeed attentive, and principled) rejects the most ungovernable type of conversation: flamewars. > convincingly show moderation neutrality on a topic like this There is no neutrality, of course. SMBiz can state its bias and interests clearly. But it will be hard to make a profit from achieving neutrality because strident factions in polarised debate do not want neutrality.