7 ms·

Twitter apologizes for 'Kill all Jews' trending topic

63 points by mudil 8y ago

adamrezich 8y ago
>"At times, we do prevent certain content from trending and we have now done so with this trend." Really? "At times?" I was under the impression that trending topics haven't been allowed to appear without human oversight in quite awhile.
- fweespeech 8y ago
  Yeah, that isn't correct. They react when they see an issue rather than proactively restrict it.
  - adamrezich 8y ago
    Well if it's not what they're doing already they really ought to think about it. Twitter is supposed to be the "safe" micro-blog social network, where one gives up complete free speech in exchange for being shielded from extremist hate speech. The supposed benefit of using Twitter over complete-free-speech alternatives like Gab or certain Mastodon instances is, users are supposed to be shielded from the extreme hate speech that is permitted on those other platforms. If they can't deliver on this to the point where users are seeing "Kill all Jews" trending, then the only edge Twitter has over its competitors is user count, and that is subject to change if they fail to deliver on their other supposed advantage. Basically if you tell me the reason Twitter is better than Gab et al. is that you don't see hate speech on Twitter because that's their policy, yet a user logs in and sees the most blatant stereotypical hate speech possible trending right there at the top of the page, you're not really selling me on what you tell me is supposedly better about your service.
    - izzydata 8y ago
      Where or what is claiming that Twitter is "supposed" to be an internet safe-space?
    - gowld 8y ago
      Who said Twitter was "safe"? Twitter blocks some of the extreme end, but not nearly enough to call it "safe". Twitter hosts credible threats by people who have attempted murder, like the recent pipe-mail-bomber.
      - adamrezich 8y ago
        They've been trying to position themselves as "safe" for years now with their "Trust & Safety Council" and other such nonsense. They repeatedly use "safe" and "safety" in their press releases about their new policies. The replies to my previous comment including yours seem to be missing the point: of course Twitter doesn't remotely begin to deliver on any of these promises of "safety," yet Twitter is still presented as the "safe" alternative to Gab et al. Personally I don't mind seeing "hate speech" because a.) I see hate speech about my own race, sex, and religion on a nearly daily basis at this point and b.) I realize that "deplatforming hate" doesn't do anything but cause it to coalesce elsewhere. By showing that Twitter isn't as "safe" as they claim to be, I'm questioning why we're continuing to use it instead of either a completely-free-speech or a better-curated actually-more-"safe" alternative.
    - captainmuon 8y ago
      > Twitter is supposed to be the "safe" micro-blog social network Strange, I have the opposite perception. On Twitter, I am constantly confronted with "kill the jews" kind of hate speech. If you post something with a remotely controversial hashtag, you'll get frothing responses. Oh, and on some days searching for innocent terms gets you tons of NSFW nudes - if you are outraged by that kind of thing. OTOH, Mastodon is advertized as "safe" alternative - or lets better say "comfy". Whether you feel comfy by being able to express radical free speech and not having to deal with identity politcs, political correctness, "SJWs" and so on. Or whether you are on the other side of that debate and don't want to be bothered by hate speech, misogynism, and other abhorrent things - a loosely federated network like Mastodon offers places for both groups.
  - tweakz 8y ago
    Devil's advocate here. Unless one works at Twitter, how would one know that they haven't proactively restricted some? Both can be true. They can proactive restrict some (which we never see, because they're restricted) and they also restrict when they see an issue.
- kodablah 8y ago
  At times they prevent some. You mention there is always oversight. You can both be right.
rusher81572 8y ago
Typical of the new anti-semitic left
- odonnellryan 8y ago
  Can you give one example of the left being antisemitic? (no, saying "let's try to not hate on Muslims so much" isn't an example of antisemitism).
  - MrZongle2 8y ago
    Here's an interesting article discussing it: http://thefederalist.com/2018/04/06/why-does-the-left-get-a-pass-on-anti-semitism/ http://thefederalist.com/2018/04/06/why-does-the-left-get-a-...
    - honkycat 8y ago
      Great job posting a nut-job shill publication. They have a lot of interesting articles: - The Democratic Party Is Working To Destroy The American Way Of Life - ‘I’m Pro-Life But Wouldn’t Outlaw Abortion’ Parallels Stephen Douglas’s Argument For Slavery - New Evidence Suggests Planned Parenthood Lied To Congress About Aborted Baby Part Profits They sound really "Fair and Balanced"
      - dx87 8y ago
        Not vouching for the linked article, but "fair and balanced" isn't a requirement to post here. Vox and Vice are both openly liberal publications that are regularly posted here. If you have specific disagreements, comment on them. Calling something a "nut-job shill publication" isn't productive.
        honkycat 8y ago
        I was making a joke about how it is a very conservative publication. "Fair and Balanced" was Fox News' motto for a long time. Not that being conservative invalidates everything contained on the blog. But, you can find blogs who claim our government is controlled by shape-shifting lizard people on the internet, so a single article from a biased source isn't a strong argument and having an article titled: "The Democratic Party Is Working To Destroy The American Way Of Life" is pretty telling.
smacktoward 8y ago
The problem is using "trending" as a content discovery algorithm in the first place. "Trending" just means that the topic is being referenced by the largest volume of accounts. Even if you can say with certainty that every one of your accounts represents an actual, legitimate person, you still have the problem that all "trending" does is surface content that's preferred by whichever of your users can mobilize the biggest angry mob. And if you can't say that with certainty (which bot-flooded Twitter absolutely cannot), "trending" becomes a wide open battlefield, trivially subject to exploitation by whoever has the largest budget for sock puppets. "Trending" isn't an editorial strategy. It is the abdication of editorial strategy. It's being asked by Twitter among others to do much more than it is capable of doing, and their users suffer as a result.
- tomp 8y ago
  > "Trending" isn't an editorial strategy. It is the abdication of editorial strategy. Not sure I understand your point. “Trending” is exactly what it says on the cover. On the other hand, calling topics picked by the editor trending would be a bit of a perversion of the meaning of the word, no?
  - marmaduke 8y ago
    Patent means that editors should help people find new content, not an ML algorithm under the guise of editorial effort
    - tomp 8y ago
      I partially agree; I see value in both.
  - jakelazaroff 8y ago
    Trending topic is a bit of a misnomer. Twitter isn't identifying "topics" here, per se; they're just extracting the most common substrings verbatim from recent tweets. A human editor might see a topic like this becoming popular and summarize it as "Synagogue Vandalism". They wouldn't just present it as "Kill All Jews" with no other context.
MrZongle2 8y ago
Note that they said nothing about being sorry for being a platform that facilitates hate, just that it showed up as a topic.
- twblalock 8y ago
  Any platform that facilitates people saying what they want to say is going to facilitate hate speech, because some people are racists. The platforms can try to prevent that, and Twitter does try. But it is simply not possible to catch everything unless every single tweet is reviewed by a human censor. That won't scale, but even if it did, the censors would not always be trustworthy. Some people seem to be convinced that we can have large social platforms free of hate speech and fake news. It's not possible, because of the users.
  - gowld 8y ago
    Users have been begging Twitter to try to build abuse controls for years, to deaf hears.
    - camdenlock 8y ago
      I don't understand this argument. Haven't they done just this over the past 4 years or so? Enhanced blocking and muting controls, etc. What more needs to be done? It's up to individuals to curate their own social environments.
    - twblalock 8y ago
      It’s pretty clear that some people on the left would complain about abuse on Twitter unless abuse tools were 100% effective, which is not achievable. And simply implementing such tools would draw the ire of people on the right, including the party that currently controls the presidency and the congress. Twitter is in a no-win situation here.
- icedchai 8y ago
  Should they apologize for human nature? People facilitate hate, no matter what the platform.
msie 8y ago
The Trending Topic is so bad sometimes. The other day I saw that #davies was trending and the article presented by Twitter was some marriage news of some unknown couple. It was really trending because of a soccer phenom playing his last game in Vancouver before heading to the big leagues.
- pkulak 8y ago
  I would contend that MLS is the big leagues. :(
  - moate 8y ago
    I would argue that it's the Major League of Soccer in fact!
captainmuon 8y ago
Another way to deal with this would be to not apologize, but to insist the algorithms are imparital, and to condemn their users. Call a big press conference and say, shit, we found that a big portion of our users are antisemites. Put that trending topic behind a trigger warning and maybe add content disclaimers to antisemite's pages. They could say they are protecting freedom of speech, but also the other side of the medal - freedom to call out positions you find wrong. I understand of course that that ship has sailed. Howver, I also believe that the strategy of "keeping the lid on" positions that are hateful or morally undesirable does not work long term. It was completely unthinkable and taboo to voice nazi positions in public in post WW2-Germany, it is even forbidden by law. But this did not irradicate Nazi thought, it just pushed it underground. And look, suddenly with some historical distance it is coming out again. Same in the US with people running around in KKK hoods and waving the confederate flag. Of course, racism was never gone or solved (no matter what we thought as white kids growing up in the 90s), but now it seems to be OK again to be openly racist in certain circles. To tie it back to the topic, I believe we should not sweep positions we find horrible under the carpet, by saying they are taboo, and pretending we (e.g. Twitter) are impartial and unpolitical. Rather, we have to endure that people hold these positions, and speak out against them. Even if that means that we no longer are "politically neutral".
- dragonwriter 8y ago
  > Another way to deal with this would be to not apologize, but to insist the algorithms are imparital, But since, as they admit, they have historically intervened in the trending rankings, insisting that they are impartial would be lying. They could decide not to intervene this time, but that would raise he question of why this is less deserving of intervention than past issues, if they were being honest and open.
- gowld 8y ago
  You contradict yourselves. Anti-hate-speech laws and culture suppressed the spread of ideas. That's the concept of "chilling effects" (from the other side). If suppressing speech didn't matter, no one would be worried about it being suppressed. With the rise of non-suppressed hate speech on Reddit and Twitter and Fox News and the like, we saw a resurgence of terrorism in the US, terrorism which was very clearly inspired by the hate speech (specifically, it's conspiracy theories) and spread ideas that people wouldn't have thought up on their own.
Endama 8y ago
Twitter needs some kind of negative feedback mechanism to allow for the discovery algorithm to correct for the sentiment of users. A downvote, dislike, un-heart, something along those lines, would be a good signal to allow users to signify what kind of content they deem objectionable.
- tomatotomato37 8y ago
  Downvote-type systems aren't that effective since they usually end up becoming a "I disagree" button and multiply the effect of whatever mob is most active at the moment. It's one of the reasons HN locks them behind a karma wall.
  - Yetanfou 8y ago
    Even that is not an effective way to avoid the negative effects of the downvote button, especially not here on HN where it actually is meant to be used as a "disagree" button: https://news.ycombinator.com/item?id=117171 https://news.ycombinator.com/item?id=117171 The "karma wall" only keeps the downvote option out of the hands of those who follow short-lived "mobs", it does nothing to stave off the balkanisation of the platform.
Wavelets 8y ago
Man, Twitter seems to be eating a lot of crow lately in this domain. They came out swinging against regulation/"censorship" of their users... but this is getting out of hand. Twitter, Facebook, etc. need to start taking some responsibility here or someone is going to do it for them (i.e. the government).

63 points by mudil 8y ago

>"At times, we do prevent certain content from trending and we have now done so with this trend." Really? "At times?" I was under the impression that trending topics haven't been allowed to appear without human oversight in quite awhile.

fweespeech 8y ago
Yeah, that isn't correct. They react when they see an issue rather than proactively restrict it.
- adamrezich 8y ago
  Well if it's not what they're doing already they really ought to think about it. Twitter is supposed to be the "safe" micro-blog social network, where one gives up complete free speech in exchange for being shielded from extremist hate speech. The supposed benefit of using Twitter over complete-free-speech alternatives like Gab or certain Mastodon instances is, users are supposed to be shielded from the extreme hate speech that is permitted on those other platforms. If they can't deliver on this to the point where users are seeing "Kill all Jews" trending, then the only edge Twitter has over its competitors is user count, and that is subject to change if they fail to deliver on their other supposed advantage. Basically if you tell me the reason Twitter is better than Gab et al. is that you don't see hate speech on Twitter because that's their policy, yet a user logs in and sees the most blatant stereotypical hate speech possible trending right there at the top of the page, you're not really selling me on what you tell me is supposedly better about your service.
  - izzydata 8y ago
    Where or what is claiming that Twitter is "supposed" to be an internet safe-space?
  - gowld 8y ago
    Who said Twitter was "safe"? Twitter blocks some of the extreme end, but not nearly enough to call it "safe". Twitter hosts credible threats by people who have attempted murder, like the recent pipe-mail-bomber.
    - adamrezich 8y ago
      They've been trying to position themselves as "safe" for years now with their "Trust & Safety Council" and other such nonsense. They repeatedly use "safe" and "safety" in their press releases about their new policies. The replies to my previous comment including yours seem to be missing the point: of course Twitter doesn't remotely begin to deliver on any of these promises of "safety," yet Twitter is still presented as the "safe" alternative to Gab et al. Personally I don't mind seeing "hate speech" because a.) I see hate speech about my own race, sex, and religion on a nearly daily basis at this point and b.) I realize that "deplatforming hate" doesn't do anything but cause it to coalesce elsewhere. By showing that Twitter isn't as "safe" as they claim to be, I'm questioning why we're continuing to use it instead of either a completely-free-speech or a better-curated actually-more-"safe" alternative.
  - captainmuon 8y ago
    > Twitter is supposed to be the "safe" micro-blog social network Strange, I have the opposite perception. On Twitter, I am constantly confronted with "kill the jews" kind of hate speech. If you post something with a remotely controversial hashtag, you'll get frothing responses. Oh, and on some days searching for innocent terms gets you tons of NSFW nudes - if you are outraged by that kind of thing. OTOH, Mastodon is advertized as "safe" alternative - or lets better say "comfy". Whether you feel comfy by being able to express radical free speech and not having to deal with identity politcs, political correctness, "SJWs" and so on. Or whether you are on the other side of that debate and don't want to be bothered by hate speech, misogynism, and other abhorrent things - a loosely federated network like Mastodon offers places for both groups.
- tweakz 8y ago
  Devil's advocate here. Unless one works at Twitter, how would one know that they haven't proactively restricted some? Both can be true. They can proactive restrict some (which we never see, because they're restricted) and they also restrict when they see an issue.
kodablah 8y ago
At times they prevent some. You mention there is always oversight. You can both be right.

rusher81572 8y ago

Typical of the new anti-semitic left

odonnellryan 8y ago
Can you give one example of the left being antisemitic? (no, saying "let's try to not hate on Muslims so much" isn't an example of antisemitism).
- MrZongle2 8y ago
  Here's an interesting article discussing it: http://thefederalist.com/2018/04/06/why-does-the-left-get-a-pass-on-anti-semitism/ http://thefederalist.com/2018/04/06/why-does-the-left-get-a-...
  - honkycat 8y ago
    Great job posting a nut-job shill publication. They have a lot of interesting articles: - The Democratic Party Is Working To Destroy The American Way Of Life - ‘I’m Pro-Life But Wouldn’t Outlaw Abortion’ Parallels Stephen Douglas’s Argument For Slavery - New Evidence Suggests Planned Parenthood Lied To Congress About Aborted Baby Part Profits They sound really "Fair and Balanced"
    - dx87 8y ago
      Not vouching for the linked article, but "fair and balanced" isn't a requirement to post here. Vox and Vice are both openly liberal publications that are regularly posted here. If you have specific disagreements, comment on them. Calling something a "nut-job shill publication" isn't productive.
      - honkycat 8y ago
        I was making a joke about how it is a very conservative publication. "Fair and Balanced" was Fox News' motto for a long time. Not that being conservative invalidates everything contained on the blog. But, you can find blogs who claim our government is controlled by shape-shifting lizard people on the internet, so a single article from a biased source isn't a strong argument and having an article titled: "The Democratic Party Is Working To Destroy The American Way Of Life" is pretty telling.

smacktoward 8y ago

The problem is using "trending" as a content discovery algorithm in the first place. "Trending" just means that the topic is being referenced by the largest volume of accounts. Even if you can say with certainty that every one of your accounts represents an actual, legitimate person, you still have the problem that all "trending" does is surface content that's preferred by whichever of your users can mobilize the biggest angry mob. And if you can't say that with certainty (which bot-flooded Twitter absolutely cannot), "trending" becomes a wide open battlefield, trivially subject to exploitation by whoever has the largest budget for sock puppets. "Trending" isn't an editorial strategy. It is the abdication of editorial strategy. It's being asked by Twitter among others to do much more than it is capable of doing, and their users suffer as a result.

tomp 8y ago
> "Trending" isn't an editorial strategy. It is the abdication of editorial strategy. Not sure I understand your point. “Trending” is exactly what it says on the cover. On the other hand, calling topics picked by the editor trending would be a bit of a perversion of the meaning of the word, no?
- marmaduke 8y ago
  Patent means that editors should help people find new content, not an ML algorithm under the guise of editorial effort
  - tomp 8y ago
    I partially agree; I see value in both.
- jakelazaroff 8y ago
  Trending topic is a bit of a misnomer. Twitter isn't identifying "topics" here, per se; they're just extracting the most common substrings verbatim from recent tweets. A human editor might see a topic like this becoming popular and summarize it as "Synagogue Vandalism". They wouldn't just present it as "Kill All Jews" with no other context.

MrZongle2 8y ago

Note that they said nothing about being sorry for being a platform that facilitates hate, just that it showed up as a topic.

twblalock 8y ago
Any platform that facilitates people saying what they want to say is going to facilitate hate speech, because some people are racists. The platforms can try to prevent that, and Twitter does try. But it is simply not possible to catch everything unless every single tweet is reviewed by a human censor. That won't scale, but even if it did, the censors would not always be trustworthy. Some people seem to be convinced that we can have large social platforms free of hate speech and fake news. It's not possible, because of the users.
- gowld 8y ago
  Users have been begging Twitter to try to build abuse controls for years, to deaf hears.
  - camdenlock 8y ago
    I don't understand this argument. Haven't they done just this over the past 4 years or so? Enhanced blocking and muting controls, etc. What more needs to be done? It's up to individuals to curate their own social environments.
  - twblalock 8y ago
    It’s pretty clear that some people on the left would complain about abuse on Twitter unless abuse tools were 100% effective, which is not achievable. And simply implementing such tools would draw the ire of people on the right, including the party that currently controls the presidency and the congress. Twitter is in a no-win situation here.
icedchai 8y ago
Should they apologize for human nature? People facilitate hate, no matter what the platform.

msie 8y ago

The Trending Topic is so bad sometimes. The other day I saw that #davies was trending and the article presented by Twitter was some marriage news of some unknown couple. It was really trending because of a soccer phenom playing his last game in Vancouver before heading to the big leagues.

pkulak 8y ago
I would contend that MLS is the big leagues. :(
- moate 8y ago
  I would argue that it's the Major League of Soccer in fact!

captainmuon 8y ago

Another way to deal with this would be to not apologize, but to insist the algorithms are imparital, and to condemn their users. Call a big press conference and say, shit, we found that a big portion of our users are antisemites. Put that trending topic behind a trigger warning and maybe add content disclaimers to antisemite's pages. They could say they are protecting freedom of speech, but also the other side of the medal - freedom to call out positions you find wrong. I understand of course that that ship has sailed. Howver, I also believe that the strategy of "keeping the lid on" positions that are hateful or morally undesirable does not work long term. It was completely unthinkable and taboo to voice nazi positions in public in post WW2-Germany, it is even forbidden by law. But this did not irradicate Nazi thought, it just pushed it underground. And look, suddenly with some historical distance it is coming out again. Same in the US with people running around in KKK hoods and waving the confederate flag. Of course, racism was never gone or solved (no matter what we thought as white kids growing up in the 90s), but now it seems to be OK again to be openly racist in certain circles. To tie it back to the topic, I believe we should not sweep positions we find horrible under the carpet, by saying they are taboo, and pretending we (e.g. Twitter) are impartial and unpolitical. Rather, we have to endure that people hold these positions, and speak out against them. Even if that means that we no longer are "politically neutral".

dragonwriter 8y ago
> Another way to deal with this would be to not apologize, but to insist the algorithms are imparital, But since, as they admit, they have historically intervened in the trending rankings, insisting that they are impartial would be lying. They could decide not to intervene this time, but that would raise he question of why this is less deserving of intervention than past issues, if they were being honest and open.
gowld 8y ago
You contradict yourselves. Anti-hate-speech laws and culture suppressed the spread of ideas. That's the concept of "chilling effects" (from the other side). If suppressing speech didn't matter, no one would be worried about it being suppressed. With the rise of non-suppressed hate speech on Reddit and Twitter and Fox News and the like, we saw a resurgence of terrorism in the US, terrorism which was very clearly inspired by the hate speech (specifically, it's conspiracy theories) and spread ideas that people wouldn't have thought up on their own.

Endama 8y ago

Twitter needs some kind of negative feedback mechanism to allow for the discovery algorithm to correct for the sentiment of users. A downvote, dislike, un-heart, something along those lines, would be a good signal to allow users to signify what kind of content they deem objectionable.

tomatotomato37 8y ago
Downvote-type systems aren't that effective since they usually end up becoming a "I disagree" button and multiply the effect of whatever mob is most active at the moment. It's one of the reasons HN locks them behind a karma wall.
- Yetanfou 8y ago
  Even that is not an effective way to avoid the negative effects of the downvote button, especially not here on HN where it actually is meant to be used as a "disagree" button: https://news.ycombinator.com/item?id=117171 https://news.ycombinator.com/item?id=117171 The "karma wall" only keeps the downvote option out of the hands of those who follow short-lived "mobs", it does nothing to stave off the balkanisation of the platform.

Wavelets 8y ago

Man, Twitter seems to be eating a lot of crow lately in this domain. They came out swinging against regulation/"censorship" of their users... but this is getting out of hand. Twitter, Facebook, etc. need to start taking some responsibility here or someone is going to do it for them (i.e. the government).