Comment by berkes
5 days ago
I'm not defending Twitter or their policy in any way (disclaimer: I left Twitter the moment Elmo took over. I despise his hypocrisy and his fascist ideas)
But this could be a "legitimate f-up". Normally, most of these unsafe-url protection and detection is automated in something with the scale of Twitter.
Just like URL-shorteners often are (were?) "seemingly randomly" banned, because a portion of the shared urls are pointing at malware/phishing/otherwise banned content, all urls from this shortener get banned. It may be that signal.me is simply picking up on amount of illegitimate links. Signal is clearly growing strong. Therefore signal.me links' are increasingly seen by Twitter. Most legitimate links, but the amount of illegitimate links will then also increase.
This would trigger an automated ban¹.
The real problem then is that even if it was deliberate (conspiracy theory: Mark messaged Elon: Pls help me curb the growth of the biggest competitor of Whatsapp?) twitter can easily hide behind "overzealous automation, sorry".
¹ Especially if this automation isn't maintained properly, finetuned and kept being tweaked by teams of experts - many of which left or were layd off after the aquisition of Twitter.
I think you buried the lede in your footnote here. Even if it is just a mistake, it's a pretty avoidable one by having a human in the loop to review changes to start blocking URLs to such a commonly linked site. If he thinks that it's "efficient" not to retain enough people to be able to notice that URL fragments and hashtags use the same symbol, he shouldn't be allowed anywhere near an "office of government efficiency", much less in charge of it.
Humans aren’t in the loop for automated bans. That has no relationship to staffing size.
This is likely a problem with the link banning algo not treating signal.me as high volume enough to prevent an automated ban.
That same logic most definitely exists at well-staffed companies and the internet is full of stories of people getting screwed by these systems. Google sinking legit companies with no recourse, locking out Gmail users who had decades of their life there, etc.
> Humans aren’t in the loop for automated bans.
My point is that this _shouldn't_ have been purely automated, but having fewer people to review things forces more things to be automated.
10 replies →
> This is likely a problem with the link banning algo not treating signal.me as high volume enough to prevent an automated ban.
And rightfully so. Despite being discussed a lot here, Signal is not very popular or well-known. Even TFA felt the need to start by explaining what Signal is. TFA then adds:
“This request looks like it might be automated” reads another prompt. “To protect our users from spam and other malicious activity, we can’t complete this action right now. Please try again later.”
And if signal.me is being used for automated spam, automated attacks require automated solutions.
> The real problem then is that even if it was deliberate (conspiracy theory: Mark messaged Elon: Pls help me curb the growth of the biggest competitor of Whatsapp?) twitter can easily hide behind "overzealous automation, sorry".
That would indeed be concerning. (And maybe illegal?) But is it anything more than an unfounded accusation?
1 reply →
[flagged]
You might be onto something, I noticed that all the banned links were in the format:
https://signal.me/#eu/fdy5h1miMifXa...
The URL hash (the part after #) is often not considered by automated systems to be a part of URL that's meaningful, because hash is normally only used for addressing parts of the website that was loaded based on the previous part of the URL. If a particular Signal.me link was flagged for whatever legitimate reason (contained malware or illegal content) it's entirely reasonable that an automated system would strip the hash and block the whole domain (because the path part in this URL is just "/" and nothing else).
It'll be interesting to see whether they address and reverse it. If not, then we can be fairly sure this was intentional.
You can actually post links of the form https://signal.me/asdf. But https://signal.me/#asdf is blocked. That supports you point of view, I guess?
If you can post links in the form of signal.me/asdf then yes, this reinforces my point that they've probably automatically flagged the top-level domain because of how these /#me links are constructed by Signal.
Or they just did a poor job of implementing the block on signal links
The purpose of a system is what it does.
After a review, the program works as coded.
"The purpose of a system is what it does."
This is only true for systems that works perfectly. If the implementation if flawed, the system can do something different from its purpose.
Claiming "The purpose of a system is what it does." is like claiming that software bugs do not exist.
"The purpose of a system is what it does" is a shorthand statement that comes out of the study of complex, human-involved, systems that lack unitary design, often having many actors having a hand in creating over time, and mostly is a statement about the lack of analytical utility of any other concept of the purpose of a system, and it is about analyzing the operation of those systems over a window of time. A longer phrase from the person who coined it on the same topic and explaining it is, “There is after all, no point in claiming that the purpose of a system is to do what it constantly fails to do.”
> Claiming "The purpose of a system is what it does." is like claiming that software bugs do not exist.
It's more like claiming that there is no meaningful difference to any outside observer between a bug that is not eradicated over an extended time window and an intentional feature.
I disagree, because it's the opposite: the purpose of a system only aligns with what we want the purpose to be if the system works perfectly. Whatever we consider the purpose to be doesn't matter for the effect the system has on the world, because if we choose to keep the system alive as-is, it will keep doing the thing that we apparently don't want it to be doing. It's the same for a program - if you choose not to fix a bug, that bug is part of the functionality. It doesn't really matter to the user whether it's supposed to work differently, unless you actually fix the bug.
I do agree with you in cases where the system is being continually refined, but I don't think the quote talks about changing systems, only about constant ones.
Seems reductive.
But in the absence of credible information to the contrary, it's not wrong.
https://en.wikipedia.org/wiki/The_purpose_of_a_system_is_wha...
when a system's side effects or unintended consequences reveal that its behavior is poorly understood, then the POSIWID perspective can balance political understandings of system behavior with a more straightforwardly descriptive view
11 replies →
It's a thought terminating cliche.
Systems thinking is much more nuanced and productive IMO. Drift Into Failure by Sidney Dekker is a good introduction.
1 reply →
It's a good "razor" to strip away rationalizations and sophistry used to defend perverse or failed systems and ideas.
E.g. the "purpose" of fascism is nihilistic violence and the "purpose" of communism is rule by a bureaucratic elite that is somehow more equal than everyone else. All the sophistry is bullshit.
1 reply →
Three issues with believing this: 1. How long does it take to undo this though even if it might be an automated screwup?
2. Why it isn’t getting banned on other social networks and only on X?
3. Didn’t X previously block Substack and Mastodon URLs?
The thing is that, in a platform based on link sharing, it should be known which domains point to URL shorteners.
Even if you automate their handling, the algorithm should know that, if it bumps into a say signal.me, bit.ly or goo.gl URL, it should first do a GET and then apply the algorithm to whatever is provided in the Location header.
Not doing this for a widely used URL shortener like signal.me is just a show of technical incompetence.
As you point out, "honest mistake" can be used by sophisticated intentional aggressors to get away with their attacks.
For a long time, the advice was "Never attribute to malice that which is adequately explained by stupidity." but aggressors evolve to fit their surroundings. When the population largely follows this rule, it becomes a competitive advantage to fake incompetence.
Perhaps both malice and incompetence should be treated the same, especially regarding punishment, until proven otherwise. After all, robust systems are designed in such a way that a single mistake can't cause harm. If somebody fails to design a system so that multiple mistakes (how many depends on cost and severity) have to stack up, then he should be held responsible.
This ties in with something that took me far too long to recognize: Trust has two pillars.
One pillar is alignment of values, and therefore intent. The other pillar is competence.
These are the same issues faced by AI development, as well as representative government, or anything regulating a dynamic with competing elements or agents.
Yet our plurality voting system would be insufficient even to keep a car on the road and driving within the speed requirements. If only the founding fathers had recognized the need to have more information included in ballots so that negative campaigning wasn't as effective if not more effective than positive.
If we voted with {+1, +0.5, -0.5, 0, 0, 0...} weights, without duplication of non-zero values, the smartest, most constructive candidates would have a better chance. Each district would have its own blend of 3-4 viable parties, and the nation would be all the healthier for it. (Side note: Yes, this is still one person one vote--you could imagine voting with a single checkbox for a single permutation of all possible assignments of the scores, as an intermediate form.)
Back to your point, though: Yes, incompetence and malice can have the same effect in the short term. The long term is what determines the difference, both in effect and our responses to it.
I have realized something quite related in my growing years of experience both interviewing and observing contributors to a technical/engineering organization
Q: Given two engineers, one incompetent and one malicious, how can you tell the difference between the incompetent engineer and the malicious engineer?
A: It doesn't matter.
> One pillar is alignment of values, and therefore intent. The other pillar is competence.
That's a good point.
> If we voted with {+1, +0.5, -0.5, 0, 0, 0...} weights, without duplication of non-zero values
Are you describing range/score voting?
I don't think avoiding duplicated values is necessary but it's pretty well known that score voting is the best system: https://rangevoting.org/
See the diagram at the bottom of the page describing voter regrets. Everybody at least somewhat interested in voting systems would prefer this, especially over plurality/FPTP which is the stupidest system possible. But a lot of people are clueless or willingly supporting a broken system.
There's also an explorable explanation by Nicky Case: https://ncase.me/ballot/
Sidenote: the fact you need to explain that expressing more information in one vote is still one vote shows how clueless people are. Obviously every vote has the same power to influence the result but some people will try to wear you down through misunderstood technicalities.
I like to add the statement, "Sufficiently advanced negligence is indistinguishable from malice."
"Sufficiently advanced malice is indistinguishable from negligence"
FTFY :)
1 reply →
"Mark messaged Elon: Pls help me curb the growth of the biggest competitor of Whatsapp?"
Didn't they wanted to beat each other up in the public?
I would have prefered that concept and not shady deals. (and while it is of course possible, I really doubt it in this case)
> Didn't they wanted to beat each other up in the public?
Isn't that like ultimate bro code for "I love you man"?
I will go with "legitimate f-up" too. Elon Musk has been pretty vocal about Signal in the past, mostly positively. If blocking Signal URLs has been intentional, he would have probably have mentioned it somewhere.
Shortened URLs are dubious by default. It is also possible that there really is a lot of spam/scam happening on Signal right now with signal.me URLs as an entry point. I mean, why not? Every messaging platform can be used for that, even more so if end-to-end encrypted as it makes spam detection harder. In fact, one of the first messages I received on Signal was an obvious scam from a user pretending to be Amazon.
Are links to mastodon still banned on twitter? Because that was a thing after Musk took over. So much for being a free speech absolutist.
You're making the mistake of taking a (communal + antagonistic) narcissist at face value. They are known to lie to suit their current goals and when those goals are achieved, they will lie to suit their new goals, whether the lies are congruent with each other or not.
This is a guy who:
- publicly called a rescuer "pedo guy", then falsely claimed it's a common insult from South Africa
- in a private email called him a "child rapist" and made up allegations of a 12 yo bride
- hired a PI to dig up dirt on him (which failed to corroborate any of his allegations)
Western society really needs to destigmatize discussion of mental illness, including diagnosing public personalities based on their behavior. Give them an opportunity to defend themselves, sure, but at some point, they become a danger to others (usually not to themselves) and should be required to seek treatment or be committed to a mental institution.
It allows links to bsky and Mastodon posts, but it doesn't auto-summarize bsky posts at the moment.
1 reply →
[flagged]
2 replies →
If Elon Musk really as narcissistic as you paint him, he would definitely take credit for these bans, or at least comment on them.
A narcissist wants to be the center of attention, by definition, and you are actually playing his game here. A "legitimate f-up" is the worst for a narcissist, as it would show he is nothing special, and incompetent.
That's why I go with the "legitimate f-up" explanation. There is no clear intent, it lacks flair, and it can just cause a mild controversy at best. Though I am still pissed that it turned out to be something about Elon Musk again and not Twitter/X. As much as a "nanomanager" he likes to call himself, he is not the only one working at Twitter/X, there are still people with some responsibilities there. It is not all a masterplan by Elon Musk.
As for banning links to other platforms like Mastodon, it was mentioned explicitly on the Twitter Support account that "we will no longer allow free promotion of certain social media platforms on Twitter", followed by a list of platforms that included Mastodon, but not Signal. I don't think it is still the case, and I am not aware of anything similar regarding a ban on signal.me links.
1 reply →
I mean, this has been going on for years at this point (for instance see https://fortune.com/2022/12/18/twitter-suspends-paul-graham-...). At some point you have to stop extending the benefit of the doubt.
does the reason matter? regardless of whether it is a mistake or or censorship, the end result is the same.
The reason absolutely matter : a mistake can happen to anyone, and be fixed within a short time, while censorship is deliberate and will probably not be fixed
If a 'mistake' keeps happening for two years, then it is realistically not a mistake. This is not Musk!twitter's first 'mistake' of this nature, by a long shot.
1 reply →
twitter is not anyone. you would expect a company of this size with millions of users to have checks and tests.
6 replies →