This is an okay example of why I don't use platforms whose conversation norms are controlled by one provider. Consider mirroring your tweets on Mastodon.
"By clicking Delete you acknowledge that your tweet violated the Twitter Rules" is particularly insidious.
Making you delete your own tweet, instead of doing it automatically, is such a sick show of power. Like saying not only can they censor whatever they want, they can even make you do it for them.
I think we humans are bound to experience a uniquely negative reaction any time a non-human entity punishes us for communal rule-breaking. This makes sense, because communal rule-following is a distinctly human activity which has co-evolved to keep us alive. It's embedded deep in our evolutionary firmware, so it is dumbfounding and infuriating to encounter its enforcement by non-humans, especially (but not exclusively) when we feel that it was done in error. You are much more likely to react with anger and perhaps even violence in a situation where your mirror neurons aren't tickled by the presence of a fellow human.
Even with a robust and timely appeal process, this sort of thing is going to have a powerful selection pressure over time, weeding out any remotely controversial, provocative, or innovative expression of meaning, leaving behind a bland, conformist soup of consensus.
Automated mechanisms only really make sense for counteracting automated abuses (e.g. vote farms). Human moderation is still necessary for human abuses (e.g. harassment).
Is there any indication of whether the problem was using the word "kill" or if the problem was sharing a potentially dangerous command (if it was correctly written)?
My suspicion would be that the combination of the words "kill" and "all" in close proximity is the problem. It shouldn't be hard to imagine the sorts of messages this was intended to catch.
What's disappointing here is that human review either didn't occur, or failed to recognize that this wasn't a post that was intended to be caught by this filter.
I've been banned for the exact same thing! I tweeted the classic joke "old people should be killed at birth" and was banned, in 2018. Twitter' bots are very touchy about the word "kill".
Hello Linux users. killall is a UNIX System V command, though the version Linux uses first appeared in FreeBSD 2.1. IOW killall *is not* a Linux command. Would you also refer to your mouse as your "Linux mouse" and your keyboard as your "Linux keyboard," and your monitor as your "Linux monitor?" Please just assume everything in Linux *does not belong* to Linux, as you will be correct far more often than not. Thank you.
Is it a command one can use in Linux? Then it is a Linux command. And frankly I wouldn't put it past some people to have specialized peripherals they specifically use with Linux.
So Linux just annexes ownership of whatever it touches? Where does it end? You should code in Linux C++, using Linux emacs.And you can't get a Linux hardcopy without a Linux printer loaded with Linux ink and Linux paper. Don't forget your Linux ethernet cables and you need to have a handful of Linux thumb drives! And you'll be doing no Linux whatsoever without Linux electricity. ><
Because the things he says are contradictory, often very vague, and often extremely naive, either intentionally or unintentionally, regarding how hard it is to moderate content. As numerous industry leaders have pointed out, people on the Left think the Left is getting overly censored, people on the Right think the right is getting overly censored, in actuality it's an extremely complex system and the site admins just want people to stop threatening each other. Musk, however, has made it very clear that he feels the Right is being censored, and he will fix that, which doesn't bode well for people on the Left.
I'm hoping that if Musk actually does end up buying twitter, it leads to larger uptick for Mastodon.
> Because the things he says are contradictory, often very vague, and often extremely naive, either intentionally or unintentionally, regarding how hard it is to moderate content. As numerous industry leaders have pointed out, people on the Left think the Left is getting overly censored, people on the Right think the right is getting overly censored, in actuality it's an extremely complex system and the site admins just want people to stop threatening each other.
This is a claim that stinks of hypocrisy. If Twitter as it is removes Tweets under the vague and arbitrary notion of pursuing peace (i.e. heckler's veto), how is that any better than what people are accusing Musk of, even when he is not yet at the helm? Keep in mind that people like Yishan admitted to wanting to piss off Moral Majority conservatives during the Bush era. Twitter was ,according to Dorsey, the free speech wing of the free speech party. Why these people seek to compel a false peace now by appeasing today's equivalent is beyond me.
>Musk, however, has made it very clear that he feels the Right is being censored, and he will fix that, which doesn't bode well for people on the Left.
I'm hoping that if Musk actually does end up buying twitter, it leads to larger uptick for Mastodon.
He's made it clear that people shouldn't be removed from Twitter or have their Tweets deleted for statements that would be protected by the First Amendment. This is the opposite of censorship. The main argument I've seen from figures associated with the left is that they don't want another 4chan, as though somehow 4chan is an aberration of the Web, and not the closest representation to its natural state.
While I'm glad that Mastodon allows people to build their personal Twitters, I wouldn't depend on Mastodon to be censor-free either.
> this kind of thing is exactly what he wants to end
Serious question: Is Musk going to allow threats of violence on Twitter?
The joking tweet in question wasn't a real threat of violence, but it's likely that Twitter's flawed algorithm flagged it as a threat of violence. So what about real threats of violence? Will those not be moderated at all under Musk?
Musk also said he wants to crack down on spam. Which requires automated detection, due to the extremely high volume of spam on Twitter. Everything on Twitter has extremely high volume, making it impossible to moderate manually. But algorithms are very imperfect, so you get a bunch of false positives, or a bunch of false negatives, or both. I wonder how this problem will get solved?
I'd literally trust the current pope to be a better guardian of free speech than Elon Musk.
If you want to judge how much someone values free speech, look at how they treet the speech of people they dislike or whose speech might be to their detriment. Feel free to check how Elon Musk treats free speech within his companies, especially of employees trying to unionize.
The short answer is because his proposed "solution" isn't solving the fundamental issue of fairness in public communication, nor does it guarantee any meaningful improvement thereof. Case in point: Elon is perfect, makes Twitter the best, then dies the next day and we're back to square one because the will of one man is not an institution (explicitly defined in this context as "surviving those who man it").
The long, IMHO, is that eventually the only solution is a truly decentralized system, much like DNS, that no actor alone can enable nor stop as long as other actors refuse to "fork" the namespace for a more restricted one.
That's what "freeing the code" ultimately means, and that comes down to a protocol at a basic level. Anyone (company, individual…) being free to implement whatever interface they want to it.
Then we'd likely move to observing and participating in this unimpeded common "mother / raw / source" stream of messages, through what I imagine would look like more-or-less-biased portals of sorts, culturally flavored aggregators, but ultimately unable to lie about true raw numbers.
Until we all agree about reality, there's no discussing it.
That's why we need dumb pipes, neutral transport, much like the air is dumb in transmitting sight and sound.
That's why we make protocols (as in RFC's), and every single internet-facing machine holds its little piece of HTTP land, of IRC land, tomorrow of <some Twitter-like land> (I know, it's Mastodon, but it's also not in reality as we speak. Ask your sales people about adoption rates and network effects).
Thus follows, IMHO, the answer to your question:
Elon claims a lot of things but none guarantee such a free and fair space for communication that de facto puts him out of the loop as a necessary gatekeeper (notice how Mr. Musk's moral fairness or even good intents, as a "benevolent dictator" is Montesquieu's words IIRC, is actually irrelevant to the core question).
Elon does not suggest that he would "protocolize" Twitter's operations such that the company would act as a central but ultimately _optional_ hub for the sake of a mission —and literally operation— bigger than itself: free and fair human communication (by which point who cares about their code since everyone else can check the raw data stream in countless replicas and redundant streams over the web).
Twitter (Inc.) as the steward and creator of this protocol would actually become valuable over the very long term (think: E/P ratio, employer prestige, etc) because hosting a fair and free human communication is one hell of an important job that few institutions —ever in history, private or public— would ethically be able to manage. But it's now become military-grade or hospital-grade important to a free society. As I was saying in another thread, it's about time we manage some deontological framework for the field of information technology and its practice at large.
[PS: Not that I hold my breath for one second that Twitter as we know it is in any way willing let alone able to undertake such an ethical mission, nor do I dream for one minute that Elon would buy Twitter instead of going from scratch, were the above goals his. The true disruption will likely be menial at first and come from left of field.]
> Open-sourcing code will make Twitter vulnerable and easy to be exploited by hackers.
Is this how people generally think about open source code security? IME, if your code is correctly written, it makes no difference whatsover whether an attacker has access to it. Consider that fact that Linux is open source, and every line change is heavily debated and described in the open. But the huge majority of internet facing machines all run that code. Linux is arguably better to run on an open server than any proprietary offerings.
I think the argument here is that code originated in the open is expected to be correct, or be corrected.
Meanwhile, code written behind closed doors may rely on security by obscurity more than it should - especially in the cases of violation detection/spam/similar cases.
I don't know that I agree with this argument, but it's plausible.
Anti-abuse definitely has by far the strongest argument there.
My position (on anti-abuse specifically) is that the pros may still outweigh the cons, but that there are very real and sizable cons to everyone if you open source it (as opposed to most software, where the cons are far more likely to be born by the company).
Maybe there's a best of both worlds where companies settle on the same open core but do their own modifications and fine tuning.
Moderation is not the same thing as security. Is there such a thing as "correctly written" code for moderation? Twitter automated moderation seems to be just pattern matching that needs to be later evaluated by a human... though it's not clear that Twitter actually bothers to have a human look at it.
This is especially true for spam bot detection. Bots don't need a "bug" in the code in order to "exploit" Twitter. A bot just needs to be not labeled by Twitter as a bot. So the goal wouldn't be to find a bug in Twitter's open source code but rather just see how Twitter decides what is a bot and what is a human, and then make the bot slightly more human-like. But nobody in the world has a "correct" test to decide what's a human.
I don't even understand the joke, it doesn't seem funny. The original comic isn't funny either.
>haha women have to do what we say.
I'd flag that too tbh. Clearly posting a linux command in a vacuum is not going to get your account locked (nice clickbait title though), there is the whole context of a shitty joke.
If you don't at least acknowledge that the joke could be interpreted wrong, but instead mald on your blog, and ramble on HN about muh free speech elon mask blah blah censorship, you should probably have your account banned.
The joke was "interpreted wrong" by an algorithm, probably just some pattern matching with the words "kill" and "all".
"The Twitter violation algorithm took 3 minutes to detect, flag, lock my account, and send me an email. There is no way a human can read tweets so fast, so it’s a violation detection bot army doing at work. Doing this at scale means all tweets are following some fast regex, and likely, an AI is getting trained and making decisions."
And they appealed and then Twitter decided to uphold the decision.
What is wrong with an AI doing some flagging?
I've seen plenty of people say "killed" "kill" etc on Twitter in the context of news stories. Those accounts don't get banned. Sample size N=1 for an AI flagging your weird tweet, even if it was okay, out of thousands of tweets doesn't seem a bad hit rate.
Let's also not ignore the tweet and the thread is a really shitty joke. Then you go on to make a huge deal out of it? Just delete your dumb joke and move on.
> And they appealed and then Twitter decided to uphold the decision.
Does Twitter ever not uphold its own decisions? I have serious doubts whether an intelligent human ever looks at these cases.
> Let's also not ignore the tweet and the thread is a really shitty joke. Then you go on to make a huge deal out of it? Just delete your dumb joke and move on.
There are several problems. One: "By clicking Delete, you acknowledge that your Tweet violated the Twitter Rules." This is a forced confession of an innocent person. You don't see a problem with that?
For any joke you can possibly say, you will find someone that will be offended. Flagging jokes you don't like or don't understand is part of the cancel culture?
I don't think twitter's system for handling the appeal is good (especially because it seems to be horrendously broken - if you have a user base numbering in the millions and need to have a ban appeal process, make sure it actually works so people know what they did wrong).
That said, I also think that the joke the author made was in bad taste and I can easily see why twitter flagged it and temporarily banned his account.
There was no judgment of "taste" at all. This was all automated.
"The Twitter violation algorithm took 3 minutes to detect, flag, lock my account, and send me an email. There is no way a human can read tweets so fast, so it’s a violation detection bot army doing at work. Doing this at scale means all tweets are following some fast regex, and likely, an AI is getting trained and making decisions."
I understand that it was all automated in nature. I was saying that personally I thought it was in bad taste, sorry if I wasn't communicating that correctly :)
If you acknowledge that there was a mindless bot behind this, not a human, and you also admit that you wouldn't ban an account over it, then why say "I can easily see why twitter flagged it and temporarily banned his account"? What kind of work is that statement doing?
> IMO, the upside of open-sourcing code may be moderate, but the downsides are huge.
>Every piece of code has vulnerabilities and bugs. Open-sourcing code will make Twitter vulnerable and easy to be exploited by hackers. So I am against open-sourcing the entire code-base.
"By clicking Delete you acknowledge that your tweet violated the Twitter Rules" is particularly insidious.