I cannot convince Twitter that I am human(swyx.io)
swyx.io
I cannot convince Twitter that I am human
https://www.swyx.io/proving-our-humanity/
5 comments
The problem here is that proving yourself to be human is just a proxy for the real goal which is to prevent malicious activity.
There are plenty of things which bots could do that sites would find undesirable, however all of that behavior falls into categories: things that a human can't reasonably do and things a human can reasonably do. If an account is doing something a human can't, it's easy to detect them as a bot when they do those things. If the problematic behavior can be done by anyone, then you need a way to detect humans doing it too. If a bot behaves indistinguishably from a human and does nothing that a human wouldn't be allowed to do, then why should the service care if it's a bot or a human? If you are catching all the actual abuse, then any further filtering out of bots is at best redundant.
The argument against profiling is generally that it is unfair, and that argument does not extend to bots, but another argument which does is that profiling is simply inefficient. Treating every american in cuba as a potential spy will yield an enormous number of false positives while those actually trying to spy will simply disguise that they are american. A much more robust and efficient system is to look for actual signs of espionage.
The author does give several reasons that they think they might be an edge case, but none of them seem to be too exceptional. Posting a few times per day and using a vpn are not too unusual, but more importantly they are not harmful. Someone at twitter is going to have to spend a non-zero amount of time dealing with this service ticket, and twitter will get some small but non-zero bad pr. Even when the system works, every one of those verification messages costs money on twitters end and causes annoyance for the user. If every twitter user spends 1 minute per year on average dealing with verification, that's 570 man-years wasted per year, and for what?
There are plenty of things which bots could do that sites would find undesirable, however all of that behavior falls into categories: things that a human can't reasonably do and things a human can reasonably do. If an account is doing something a human can't, it's easy to detect them as a bot when they do those things. If the problematic behavior can be done by anyone, then you need a way to detect humans doing it too. If a bot behaves indistinguishably from a human and does nothing that a human wouldn't be allowed to do, then why should the service care if it's a bot or a human? If you are catching all the actual abuse, then any further filtering out of bots is at best redundant.
The argument against profiling is generally that it is unfair, and that argument does not extend to bots, but another argument which does is that profiling is simply inefficient. Treating every american in cuba as a potential spy will yield an enormous number of false positives while those actually trying to spy will simply disguise that they are american. A much more robust and efficient system is to look for actual signs of espionage.
The author does give several reasons that they think they might be an edge case, but none of them seem to be too exceptional. Posting a few times per day and using a vpn are not too unusual, but more importantly they are not harmful. Someone at twitter is going to have to spend a non-zero amount of time dealing with this service ticket, and twitter will get some small but non-zero bad pr. Even when the system works, every one of those verification messages costs money on twitters end and causes annoyance for the user. If every twitter user spends 1 minute per year on average dealing with verification, that's 570 man-years wasted per year, and for what?
https://en.wikipedia.org/wiki/No_Longer_Human
I always look for opportunities to recommend Osamu Dazai.
I always look for opportunities to recommend Osamu Dazai.
interesting, thank you! i wish the author had shared what exactly they sent via paper.. but i guess i can write a simple cover letter. lets see how this goes...
You're not going to make things change by continuing to use the platform - right now you're actually doing the opposite - you're sending a message that no matter how badly they treat you, you're happy to keep going and expend significant amounts of effort working around their tech team's incompetence to try and get back in.