Ask HN: Any Twitter engs in the house? We need help fighting bots in ZA
I'm from South Africa, and we have bad propaganda and lots of malicious Twitter accounts making Twitter an unsafe place.
Are there any Twitter engineers that can fix Twitter's manual block list upload? It's been broken for months, at least when you upload a CSV with a large enough list.
We're using various techniques to spot 'Gupta' (accounts we believe are being sponsored by one of the families complicit in large-scale looting of our resources) Twitter accounts.
Some accounts are automated, we suspect being run from India (various mistakes by authors of tweets exposing their location), while some seem to be run by South Africans that are being paid by this family.
We've struggled to get Twitter Support to help us, with a friend of mine losing it a bit on the DM [https://twitter.com/Arfness/status/934850071367168000].
We're training a TF model to detect these accounts, and we tweet their activity under #GuptabotKPC, but all of this doesn't help if users are unable to block these accounts.
Please help
5 comments
[ 2.5 ms ] story [ 9.8 ms ] threadhttps://help.twitter.com/articles/20172663?lang=en
You can also create blocks via the API.
https://developer.twitter.com/en/docs/accounts-and-users/mut...
I think it'd be pretty straight forward to build a microservice that makes API calls from a list you keep in a database as well.
To second one: we're avoiding having to get authorisation from users to read/write from their accounts. Twitter also won't hesitate to revoke our API access when we start automatically blocking accounts on behalf of hundreds/thousands of users. Also, imagine if the Russian botfare targets us, they're more sophisticated. We would have to keep users' tokens somewhere, in the wrong hands that's one propaganda machine. We think their time's coming, because our delinquent president is pushing through a nuclear deal that's going to cost our grandchildren dearly.
I use their API a lot, and would love to do that, but at this point it might not be the most practicable option for us. Perhaps @Twitter staff could advise us ...
[1] http://blocklistza.000webhostapp.com/
There's a guy @levelsio on Twitter who made a similar project but for crowdsourcing block words at https://mute.life/. I'm not sure whether he's tried to automate any of it yet.
So you don't want to use the API but you do want to keep the list up-to-date for all users kind of like sources for an ad blocker. Am I understanding your use case right? I've been looking for / to make a Twitter blacklist service for companies that get too spammy.
I saw mute.life, thanks for that.
Yes, don't want to use the API. We can set up a small script as part of our workflow, that updates the latest blocklist, and then people can download it and import it.
The blacklist service would be interesting, although I think Twitter would 'crack down' on it if it becomes too successful.