A deep dive into Elon Musk's censorship machine: what "The Algorithm" does

[TolstoyDotCom]

Describe the bug
The main bug is that Elon Musk is presented as Mr Free Speech even tho he's strongly pro-censorship.

See this Medium post by someone else for the details: "I Dug Through X’s Open-Source Algorithm to Understand My “Ghost Ban.” It Got Weird Fast."

To Do Something About This
Send the link to the media. Point out Twitter is helping the powerful suppress debate: all they have to do is block their critics to silences them. Repeat the list of reasons from the post why someone might block someone else. Ask the reporter if they think it's fair to silence people just because they state unpopular opinions.

And, any time you see someone saying Musk supports open debate, point out how wrong that is.

[nS-101]

The article itself claims multiple times that it's not trying to present the discoveries as unearthing some anti free speech machine or trying to suppress certain people, it literally says that because of the different visibility classifiers and labels designed to fight against spam bots and such, it could adversely affect real humans which is a real concern. I'm not saying that there might not be censorship issues on X but I think we have to be realistic here, especially considering most of the code is open sourced and viewable for literally everyone.

[TolstoyDotCom]

Do you work for Twitter/X/Musk in some way, Neil? There have been extensive censorship issues on Twitter for over a decade. You can see that today just by looking at all the complaints on Reddit and even LinkedIn.

If you're still confused, have a chat with Grok. For instance, here's Elon's own AI admitting that Elon is technically-incompetent:

https://imgur.com/a/kpn7uV3

It gets even darker, with other AIs comparing Musk's censorship machine to McCarthy, Stalin, Hitler, etc etc.

[nS-101]

i've used twitter since 2014, I'm aware of what's happened to the platform, even when it leaked that elon gave special preference to certain right wing accounts that spread mostly hate by exempting them from being flagged by using certain hateful codewords. all i'm saying is that the algo we see in this repo and the claims made in the article only point out that the algo has a hard time differentiating real blocks against a spammer or bot versus blocks due to personal differences or annoyances among real people, which I can imagine is a hard thing to teach an ml algo. I still think, while there's a lot of work to be done, twitter is still the best social media platform by far to voice your opinion and see other people's ways of thinking

[TolstoyDotCom]

You didn't answer my question, Neil.

Deciding which is the "best" social network is kinda like ranking natural disasters. The Indonesian tsunami killed over 100k people, but I'm sure there have been even higher death tolls.

I was probably on Twitter before you, and back then it was an open service where you could debate people without worrying about using the "wrong" word and being censored.

Then, the diseased, anti-speech, anti-Western minds of Jack/Roth/Vijaya/etc decided - only partially for business reasons - to clamp down on speech. And, they came up with the most disturbed way to do it: based not on what you write, but on what other people think about what you write. That means those who dissent can be easily silenced by their opponents, even if those opponents are objectively wrong.

E.g., if you go into a debate about "banned books" and point out that Barnes & Noble will sell you any of those "banned" books, you'll get blocked by the mob and that will lead to you being silenced.

Instead of changing that and focusing on what someone writes rather than how it's received, Musk has continued their sick concepts, and doubled-down on the McCarthyism and Nazism:

https://imgur.com/a/WMJKJBo

If only the loons who tried to make something out of Musk's arm gesture would focus on his Nazi-like policies, he'd be in deep trouble.

[nS-101]

i think you're getting a bit carried away, and obviously I do not work for elon or any of his companies. i think it comes down to there not being a meaningful way for a ml model to differentiate between blocks against a bot and blocks against a human. i do agree to some extent that everyone should have an equal voice, even if it's controversial and results in them getting blocked or harassed, but there also has to be some meaning in an account receiving a large amount of blocks

[TolstoyDotCom]

I don't think the critics of Musk/Zuck/Jack/etc are getting carried away enough. We've got billionaires deciding who gets to talk back to elected officials and who gets silenced. Those who post trenchant critiques and questions for those officials are more likely to get blocked by their flacks and supporters than those who post childish memes.

And, lots of blocks do say a lot. E.g., I was blocked by a councilwoman for pointing out that thousands of homeless have died while she pushed "housing first" policies designed only to help her cronies. Others aren't calling her on that. I hit a nerve and got blocked. As a result, others viewing her posts might not realize how culpable she is in thousands of deaths. Blocking me says a lot about her, but Musk and (presumably) you would blame me.

And, the "banned books" example I gave above isn't just a hypothetical: when I pointed that out on Reddit, I got a whopping 600 downvotes. I had to build up my "karma" before my comments would appear on several other subs. I was blocked from the sub in question. And, those who brigaded me even followed me to other subs. Once again: Musk would blame me for that.

Somewhere in this endless, too-alt-right-focused jeremiad he correctly points out that Musk uses high-profile out-of-the-mainstream accounts to cover for banning thousands of others no one has ever heard of:

https://arctotherium.substack.com/p/the-closure-of-the-internet

Musk is an enemy of free speech and open debate.