Twitter algorithms cannot tell the difference between Neo-Nazis and Republican politicians.

[The Romulan Republic]

That's alright, I can't tell the difference either.

https://www.businessinsider.com/twitter ... ort-2019-4

Twitter is holding back from implementing a possible algorithm that could allow Twitter to more effectively crack down on neo-Nazi and white-supremacist content over concerns it could report accounts of Republican politicians, according to a report from Vice News' tech site, Motherboard.

A Twitter employee told Motherboard that at a recent company-wide meeting, an employee asked why Twitter — which has successfully used a sophisticated algorithm to identify and almost entirely eliminate ISIS-linked content — couldn't do the same for white-supremacist tweets.

According to the employee, another employee that works on artificial-intelligence (AI) issues said that such a sweeping and wide-ranging algorithm could result in some innocent accounts being flagged by accident, which may not be an acceptable trade-off.

In a separate conversation, Motherboard reported, the AI-focused employee said one concern with a white-supremacist algorithm was that it would inadvertently flag the accounts of some Republican politicians, potentially causing a backlash.

"The information cited from the 'sources' in this story has absolutely no basis in fact," a Twitter representative told INSIDER by email in response to Motherboard's reporting.

"The characterization of the exchange at the meeting of March 22nd is also completely factually inaccurate. There are no simple algorithms that find all abusive content on the Internet and we certainly wouldn't avoid turning them on for political reasons," the representative added in the statement.

Twitter has come under intense scrutiny in recent months by critics who say the platform doesn't do enough to crack down on harmful white-supremacist rhetoric. Twitter uses human moderators to go after white-nationalist propaganda.

Read more: Jack Dorsey says Twitter makes it 'super easy' to harass and abuse others, and addressing the problem is his biggest worry

"Most people can agree a beheading video or some kind of ISIS content should be proactively removed, but when we try to talk about the alt-right or white nationalism, we get into dangerous territory, where we're talking about Steve King or maybe even some of Trump's tweets," the extremism expert Amarnath Amarasingam told Motherboard.

Rep. Steve King has been criticized for using Twitter to express ideas with white-nationalist and racist connections. King also retweeted a prominent British neo-Nazi in June 2018.

Read more: Twitter CEO Jack Dorsey and Trump met behind closed doors to discuss social media ahead of the 2020 election

At the same time that Twitter is under a microscope over its handling of white supremacy, a number of prominent Republicans — including President Donald Trump — have publicly accused the company of bias against conservatives and disproportionately cracking down on right-wing accounts. Twitter has denied those accusations.

On April 23, Twitter CEO Jack Dorsey met with Trump in the Oval Office after the president fired off a series of tweets blasting the platform, writing "they don't treat me well as a Republican. Very discriminatory, hard for people to sign on. Constantly taking people off list."

Trump spent a significant amount of time during the meeting complaining about losing followers, according to The Washington Post, which Dorsey said was because of Twitter's routine deletion of fake bot accounts.

They're also getting rightly lambasted for using a different standard to moderate white supremacist content vs. Jihadi content. I mean, they're basically saying "Yeah, we'll use these algorithims to ban Jihadi propaganda, but it would be too dangerous to ban white supremacists because they have more power."

Fucking appeasing cock-suckers. I am so fucking glad I don't use Twitter.

[The Romulan Republic]

More examples of how Twitter enables the far Right:

https://www.ctvnews.ca/canada/b-c-man-s ... -1.4398081

VANCOUVER -- A British Columbia businessman and philanthropist is suing Twitter for publishing what he calls "false and defamatory" tweets that escalated during the 2016 United States election.

Frank Giustra is the founder of Lionsgate Entertainment and CEO of the Fiore Group of Companies, as well as a member of the board of trustees of the Clinton Foundation.

He says in a notice of civil claim filed in B.C. Supreme Court that he faced a targeted attack on Twitter by a group who set out to vilify him for political purposes starting around February 2015.

"Twitter began publishing a number of defamatory and malicious statements regarding the plaintiff," says the notice of civil claim, filed on Tuesday.

"Those publications included tweets stating that the plaintiff is 'corrupt,' a 'murderous thief,' a 'criminal,' and is involved in 'pizzagate,"' it states, referring to an unfounded conspiracy theory that claims Democrats harbour child sex slaves at a Washington, D.C., pizza restaurant.

Twitter also published threatening posts, which included suggestions that Giustra be killed with two "bullets to the back of his head," the claim says.

None of the allegations has been proven in court and Twitter has not yet filed a statement of defence. Twitter declined comment on Saturday, while Giustra could not immediately be reached for comment.

Giustra says in the civil claim that he asked Twitter on multiple occasions to remove the posts and the social-media company deleted some of them, but it has neglected or refused to remove others and continues to publish a "large number" of false, defamatory, abusive and threatening tweets.

He says the tweets have damaged his professional and personal reputation and caused "significant emotional distress and anxiety" for him and his family.

"The false allegations and criticism about the plaintiff engaging in fraud brings the plaintiff's honesty and integrity as a businessman into disrepute," the civil claim says.

"Further, the tweets falsely characterize the plaintiff as a pedophile which impairs his ongoing involvement with the important and socially beneficial children's charities that the plaintiff operates or assists."

Giustra wants a mandatory permanent injunction requiring Twitter to delete, remove, cease or prevent publication of the tweets, as well as a mandatory permanent injunction requiring the company to prohibit publication by its users of materials that are defamatory of the plaintiff on an indefinite basis. He's also seeking general damages and costs.

[Broomstick]

Twitter algorithms cannot tell the difference between Neo-Nazis and Republican politicians.

Well, duh - if it walks like a duck, talks like a duck....

[Solauren]

For Fucks sake twitter, does anyone have a brain over there?

Letting the algorithms lose and monitoring them would be simple
Just set up a 'Vetted' Table for accounts they shouldn't block *(see below), and have people manually verify the algorithim's choices.


Step 1 - Algorithm identifies problematic accounts, and puts them into a central database. (Into the 'Import Table')
Step 2 - Cross check the 'Import table' to the 'Previously Vetted Table'. Anyone on 'Previously vetted', gets removed from the algorithm results.
Step 3 - Manually verify the rest of the identified accounts as legitimate political accounts or not *(see below)
If the list is large enough, block viewing of accounts, and their owners from logging in, during verification.

*Accounts that should not be blocked
Elected Politicians, Registered Politicians/Party Members within their home country (this would require co-operation from those parties, so if they don't co-operate, Twitter can say 'we asked, they didn't provide us the information to verify, so we suspended assuming it was a fraudulent claim), and Uncover 'Bait' accounts (i.e F.B.I using them to atrach Neo-Nazi's so they can monitor them), and the like.

Realistically, blocking a legitimate politicians account could be considered interference with the political process, and could open them to lawsuits.
Yeah, NeoNazis could do the same, but most probably don't have the money for a lawsuit, or want the negative publicity/public attention.

[The Romulan Republic]

Is Twitter actually under any legal obligation to allow all political figures to access their platform, regardless of their conduct? Because while we know that they're not realistically going to ban, say, Donald Trump because he has a huge following and his Twitter account makes them money, he must have flagrantly violated their terms of use on all sorts of occasions.

Keep in mind that contrary to frequent misconceptions, the First Amendment prohibits only censorship by the government- there is absolutely no prohibition on a private company saying "we're going to ban you for saying this on our platform", because its their private property (there's that free enterprise the Right loves so much).

But in any case, the reason that they don't solve a perfectly solvable problem is pretty clearly because they don't want to. Because they're okay with their company being a platform for Neo-Nazi propaganda. Nazis have a disproportionate presence on social media, and tolerating them makes a social media company money.

[AniThyng]

The issue is not that republicans are indistinguishable from white supremacists, it's that they don't think the risks of a few *false positives* is worth it, yet. Emphasis on the term *false positive*, as opposed to "*true positive* but on a "legitimate" republican politician.". But I suppose it's debatable if the false positive is that or not.

Catch 22 points for this from the article:

"At the same time that Twitter is under a microscope over its handling of white supremacy, a number of prominent Republicans — including President Donald Trump — have publicly accused the company of bias against conservatives and disproportionately cracking down on right-wing accounts. Twitter has denied those accusations"

Frankly I wouldn't be so sure that it's in any way easy to block Islamist content - as the article notes, obvious stuff like easy, but after that, a lot of Islamist talking points of the sort that can be attributed to Islamist politicians are really not very different at all from white supremacist talking points, you just don't notice because you're focused on white supremacists and truly extreme Islamist stuff, not the everyday stuff put out by regular Islamist politicians outside the US.

[Solauren]

Is Twitter actually under any legal obligation to allow all political figures to access their platform, regardless of their conduct? Because while we know that they're not realistically going to ban, say, Donald Trump because he has a huge following and his Twitter account makes them money, he must have flagrantly violated their terms of use on all sorts of occasions.

I believe there are laws, depending on the area/region/district, about giving 'equal time' for political advertising. Depending on money of course.
(i.e if you are charging 50 million for a 30 second commerical spot, and one party pays it, but the other does not, to bad for them). Since Twitter is free, and open to the public, a good lawyer MIGHT be able to argue either way in court.

Keep in mind that contrary to frequent misconceptions, the First Amendment prohibits only censorship by the government- there is absolutely no prohibition on a private company saying "we're going to ban you for saying this on our platform", because its their private property (there's that free
enterprise the Right loves so much).

True. However, by the same token, it is illegal to deny service to people based on stuff like religion, race, sexual orientation, and the like.
So long as people cloak their garbage in 'political view', or 'religious view', it could be very hard to keep them out/off.

But in any case, the reason that they don't solve a perfectly solvable problem is pretty clearly because they don't want to. Because they're okay with their company being a platform for Neo-Nazi propaganda. Nazis have a disproportionate presence on social media, and tolerating them makes a social media company money.

Now, that's not ENTIRELY fair. I agree, it certainly does look that way. However, speaking as someone that has worked in IT, and with non-techie managers....

It's entirely possible there are levels of management at twitter that don't understand the problem and possible solutions.

I can't tell you how many times I mentioned the word 'database' to a manager or higher at work, and they go 'oh, MICROSOFT EXCEL can do that?'
And when I say 'No', they accuse me of lying or wasting their time. I have literally had to go and load up Microsoft Access in front of them, followed by wikipedia for definitions, to explain to them what a database is. Then I have to explain it comes with Office and is installed on everyone's computer already, and doesn't cost any more to use.....

It's also entirely possible the programmers have said 'we can't teach the code to do what a human can do', and the news article is just running with that.

I admit, I'd be very interested in seeing a break down of how much money twitter makes off various political views, however.

[Elfdart]

This reminds me of a joke I came up with many years ago:

Q: "Do you what the difference is between David Duke and Jesse Helms?"

A: "Me neither" OR "About 200,000 votes"

[Tsyroc]

Is Twitter actually under any legal obligation to allow all political figures to access their platform, regardless of their conduct? Because while we know that they're not realistically going to ban, say, Donald Trump because he has a huge following and his Twitter account makes them money, he must have flagrantly violated their terms of use on all sorts of occasions.

Keep in mind that contrary to frequent misconceptions, the First Amendment prohibits only censorship by the government- there is absolutely no prohibition on a private company saying "we're going to ban you for saying this on our platform", because its their private property (there's that free enterprise the Right loves so much).

But in any case, the reason that they don't solve a perfectly solvable problem is pretty clearly because they don't want to. Because they're okay with their company being a platform for Neo-Nazi propaganda. Nazis have a disproportionate presence on social media, and tolerating them makes a social media company money.

I'm pretty sure that legally Twitter is in the clear but there are some openings for challenges.

Donald Trump not being able to block people because he's the President might suggest that Twitter isn't just Twitter the company but "the public square". If that gets ruled on strongly enough they might have to let all but a very restricted minority on. I think Jack Dorsey kind of reiterated Twitter as the public square, or something to that effect when he was on one Joe Rogan's podcasts. At least I think it's what he'd like it to be. Unfortunately there appears to be lots of garbage in our public square and nobody really wants to clean it up.

I've seen several videos talking about whether social media (Twitter, YouTube, Facebook etc...) are platforms or a publishers. Being a platform should make them less legally responsible for what's on their service. Being a publisher makes them more legally responsible but would give them more power in determining who and what is on their services. I think a lot of them are walking a fine line at the moment. It's got to be massive pain in the ass.

I'm not on Twitter so I don't really understand what the issue with Trump blocking people was. As far as I know that meant he wouldn't see their posts but they could still see his. That doesn't seem much different than getting his comments via TV or the newspaper. I wouldn't be able to respond instantly to him there either.

I'm curious what would happen if Trump ditched Twitter and started doing his schtick on Gab?

[bilateralrope]

I'm not on Twitter so I don't really understand what the issue with Trump blocking people was. As far as I know that meant he wouldn't see their posts but they could still see his. That doesn't seem much different than getting his comments via TV or the newspaper. I wouldn't be able to respond instantly to him there either.

Going from memory here, Twitter has two ways of filtering someone's tweets:
- Muting. You don't see their tweets. They get to see yours.
- Blocking. You don't see their tweets. They don't get to see yours.

The problem with blocking is that, by hiding your tweets from someone, you also prevent them from replying. Limiting the ability to reply is an unacceptable restriction on speech when it's the government blocking you. Thus, the government isn't allowed to block people on Twitter, only mute them.

[Tsyroc]

I'm not on Twitter so I don't really understand what the issue with Trump blocking people was. As far as I know that meant he wouldn't see their posts but they could still see his. That doesn't seem much different than getting his comments via TV or the newspaper. I wouldn't be able to respond instantly to him there either.

Going from memory here, Twitter has two ways of filtering someone's tweets:
- Muting. You don't see their tweets. They get to see yours.
- Blocking. You don't see their tweets. They don't get to see yours.

The problem with blocking is that, by hiding your tweets from someone, you also prevent them from replying. Limiting the ability to reply is an unacceptable restriction on speech when it's the government blocking you. Thus, the government isn't allowed to block people on Twitter, only mute them.

Okay, cool. Thanks.

If figures though. Trump should have just muted people but instead blocked them because he's Trump. Swinging dick power move (in his mind), or tantrum? Who can tell?

[Ziggy Stardust]

I also feel obliged to make the point that both the article and the posts in this thread have so far been taking it for granted that the algorithm is actually effective at identifying hate-speech in general, notwithstanding the Neo-Nazi/Republican thing. What is its false positive rate? What is its false negative rate? Or, perhaps more importantly, what is its positive predictive value?

Speaking as someone who actually implements machine learning algorithms (admittedly in a different context, though I've done a little NLP), as advanced as they may seem they are still EXTREMELY limited for anything this complex. For all the fanfare being made about the Neo-Nazi/Republican thing, it almost certainly has a dozen other classification problems. The fact is, it's going to be incredibly difficult to develop an algorithm that can do this because we have no "gold standard"; it's not even easy for people in 90% of cases to unambiguously identify whether something is hate speech. That's why "dog-whistle" is such a thing.

[Solauren]

If figures though. Trump should have just muted people but instead blocked them because he's Trump. Swinging dick power move (in his mind), or tantrum? Who can tell?

It's Trump. Likely, he doesn't know the difference.

[The Romulan Republic]

Is Twitter actually under any legal obligation to allow all political figures to access their platform, regardless of their conduct? Because while we know that they're not realistically going to ban, say, Donald Trump because he has a huge following and his Twitter account makes them money, he must have flagrantly violated their terms of use on all sorts of occasions.

Keep in mind that contrary to frequent misconceptions, the First Amendment prohibits only censorship by the government- there is absolutely no prohibition on a private company saying "we're going to ban you for saying this on our platform", because its their private property (there's that free enterprise the Right loves so much).

But in any case, the reason that they don't solve a perfectly solvable problem is pretty clearly because they don't want to. Because they're okay with their company being a platform for Neo-Nazi propaganda. Nazis have a disproportionate presence on social media, and tolerating them makes a social media company money.

I'm pretty sure that legally Twitter is in the clear but there are some openings for challenges.

Donald Trump not being able to block people because he's the President might suggest that Twitter isn't just Twitter the company but "the public square". If that gets ruled on strongly enough they might have to let all but a very restricted minority on. I think Jack Dorsey kind of reiterated Twitter as the public square, or something to that effect when he was on one Joe Rogan's podcasts. At least I think it's what he'd like it to be. Unfortunately there appears to be lots of garbage in our public square and nobody really wants to clean it up.

I've seen several videos talking about whether social media (Twitter, YouTube, Facebook etc...) are platforms or a publishers. Being a platform should make them less legally responsible for what's on their service. Being a publisher makes them more legally responsible but would give them more power in determining who and what is on their services. I think a lot of them are walking a fine line at the moment. It's got to be massive pain in the ass.

I'm not on Twitter so I don't really understand what the issue with Trump blocking people was. As far as I know that meant he wouldn't see their posts but they could still see his. That doesn't seem much different than getting his comments via TV or the newspaper. I wouldn't be able to respond instantly to him there either.

I'm curious what would happen if Trump ditched Twitter and started doing his schtick on Gab?

Yeah, now that I think about it, the precedent about Trump not being allowed to block people could definitely cut both ways.

[Zwinmar]

So, I got this comment on Facebook because I support the idea of Unions:

[my name] , nice selfie. As a racist lazy Lib Dem Russian SOCIALIST Party Communist Marxist Union Supporter.
Tell us when you first realized you HATED Supporting Yourself Isreal CAPITALISM FREEDOM AND AMERICA ?
Light up another Bong. Loser.
No profits, No lazy ignorant Union Jobs.
Go Kill ADDITIONAL Union POLICE OFFICERS of all race gender and sexual persuasion. WITH all Lib Dem socialist Communists beloved Black Gang Members LIVES DON'T FAULKING MATTER.
Unions drove 90% of American business off shore.(misspellings are his own)

[Ziggy Stardust]

For some reason I am inordinately fascinated by his decision to emphasize "SOCIALIST" out of all the words in that first vocabulary soup of his. It's almost like he felt like it wouldn't register unless he really nailed the SOCIALIST part home. I mean, it's one thing to call someone a lazy, racist communist, but that really doesn't have enough punch to it without the SOCIALIST, I guess?