4 ms·

I would assume Twitter suspended the account due to the rate of posts. Granted, you are allowed 2400 tweets per day, [1] so the account should not have had any

I would assume Twitter suspended the account due to the rate of posts. Granted, you are allowed 2400 tweets per day, [1] so the account should not have had any issues.... [1]: https://help.twitter.com/en/rules-and-policies/twitter-limits https://help.twitter.com/en/rules-and-policies/twitter-limit...

peterwwillis 8y ago
I think the image-post all-caps multi-level text followed by a link with just a number at the end might have been spammy looking to a crap machine learning algorithm
- insensible 8y ago
  Asimov's zeroth Law of Robotics: > A robot may not harm humanity, or, by inaction, allow humanity to come to harm. Causing data loss from a poorly-trained robot's incorrect call is unacceptable. Fix your bits, Twitter.
- edent 8y ago
  That's possible. We get the text from Google's OCR product - and a lot of benches are in all caps. We could try to shift it all to sentence case, but that might lose the meaning of proper nouns. Thanks for the suggestion though, that's the first thing which has made sense about the banning.
Rjevski 8y ago
They should actually put that report button to good use and take that into account when suspending users. A high number of tweeets seen by a lot of active followers and not being reported is actually a very good signal the account is not spamming.
edent 8y ago
We don't get anywhere near that. On a good day we might get 100.
Bukhmanizer 8y ago
My guess would be the source of posts. Looks like they post from "Twitter for Google TV" and "edent's tweeting robot", so they're probably outliers due to the volume of posts and where the posts are coming from. That combined with the characteristics of the text probably triggered the algorithm. That being said, it's unfortunate that this happened to them, the lack of manual review when kinds of things go wrong is shocking.
- duskwuff 8y ago
  "Twitter for Google TV" suggests that they may be using a Twitter API key that wasn't issued to them. Probably not a good idea...
originalsimba 8y ago
We should not assume competency without seeing proof of it. False alerts are the opposite of proof of competency. I am going to assume Twitter suspended the account because their software is broken and they don't care about the sweeping consequences of deploying broken robots to police a social network where people are constantly challenging expectations with creative projects and timelines. It's their laissez-faire attitude about it that really makes them a cancer. "We've blocked your account and we won't tell you why and there's nothing you can do! P.S. Don't bother E-mailing us it all goes into a black hole!" It's everything bad about everything bad on the internet, compacted into a super efficient E-mail. I bet one of these years people wake up and start treating outgoing E-mails without valid return addresses as spam, the way we should be. Or, for accuracy, not as spam but as hostile and societally intolerable.

peterwwillis 8y ago

I think the image-post all-caps multi-level text followed by a link with just a number at the end might have been spammy looking to a crap machine learning algorithm

insensible 8y ago
Asimov's zeroth Law of Robotics: > A robot may not harm humanity, or, by inaction, allow humanity to come to harm. Causing data loss from a poorly-trained robot's incorrect call is unacceptable. Fix your bits, Twitter.
edent 8y ago
That's possible. We get the text from Google's OCR product - and a lot of benches are in all caps. We could try to shift it all to sentence case, but that might lose the meaning of proper nouns. Thanks for the suggestion though, that's the first thing which has made sense about the banning.

Rjevski 8y ago

They should actually put that report button to good use and take that into account when suspending users. A high number of tweeets seen by a lot of active followers and not being reported is actually a very good signal the account is not spamming.

edent 8y ago

We don't get anywhere near that. On a good day we might get 100.

Bukhmanizer 8y ago

My guess would be the source of posts. Looks like they post from "Twitter for Google TV" and "edent's tweeting robot", so they're probably outliers due to the volume of posts and where the posts are coming from. That combined with the characteristics of the text probably triggered the algorithm. That being said, it's unfortunate that this happened to them, the lack of manual review when kinds of things go wrong is shocking.

duskwuff 8y ago
"Twitter for Google TV" suggests that they may be using a Twitter API key that wasn't issued to them. Probably not a good idea...

originalsimba 8y ago

We should not assume competency without seeing proof of it. False alerts are the opposite of proof of competency. I am going to assume Twitter suspended the account because their software is broken and they don't care about the sweeping consequences of deploying broken robots to police a social network where people are constantly challenging expectations with creative projects and timelines. It's their laissez-faire attitude about it that really makes them a cancer. "We've blocked your account and we won't tell you why and there's nothing you can do! P.S. Don't bother E-mailing us it all goes into a black hole!" It's everything bad about everything bad on the internet, compacted into a super efficient E-mail. I bet one of these years people wake up and start treating outgoing E-mails without valid return addresses as spam, the way we should be. Or, for accuracy, not as spam but as hostile and societally intolerable.