How can I investigate a twitter user's history of being harassed?
July 11, 2018 3:53 PM Subscribe
If I'm curious about a specific twitter user's history of being harassed, how might I go about finding information?
Ideally this would be using relatively objective metrics that can be gathered in an automated fashion, such as:
- number of accounts blocked by the user during a specific timeframe (under the assumptions that more blocks = more harassment)
- messages to the user from "known bad" accounts, such as MRAs, white supremacists, trolls, and so forth
- messages to the user that fit certain criteria (contain the word "bitch," for example)
- messages to the user from accounts that were then disabled or suspended w/in a certain timeframe
or.... like, whatever else.
I'd like to be able to gather these metrics in an automated way for accounts I don't have privileged access to, i.e., using the twitter API or similar. What metrics would be relevant? How might I gather them? Are there other APIs or resources that could provide additional information?
Ideally this would be using relatively objective metrics that can be gathered in an automated fashion, such as:
- number of accounts blocked by the user during a specific timeframe (under the assumptions that more blocks = more harassment)
- messages to the user from "known bad" accounts, such as MRAs, white supremacists, trolls, and so forth
- messages to the user that fit certain criteria (contain the word "bitch," for example)
- messages to the user from accounts that were then disabled or suspended w/in a certain timeframe
or.... like, whatever else.
I'd like to be able to gather these metrics in an automated way for accounts I don't have privileged access to, i.e., using the twitter API or similar. What metrics would be relevant? How might I gather them? Are there other APIs or resources that could provide additional information?
SocialBlade is one place public Twitter stats are made available.
I'd like to be able to gather these metrics in an automated way for accounts I don't have privileged access to
It strikes me that if the Twitter user is being harassed, going above and beyond the publicly available stats to gather a user's account information and usage without their permission contributes to the problem.
posted by juliplease at 9:12 PM on July 11, 2018 [13 favorites]
I'd like to be able to gather these metrics in an automated way for accounts I don't have privileged access to
It strikes me that if the Twitter user is being harassed, going above and beyond the publicly available stats to gather a user's account information and usage without their permission contributes to the problem.
posted by juliplease at 9:12 PM on July 11, 2018 [13 favorites]
if it's one single specific user, if you're interested strictly in the experience of a particular person and not trying to gather data about trends in a larger group, you just...ask them.
if they don't respond or don't want to tell you, that's an answer that pertains directly to the topic even if it doesn't include the numbers you're after.
posted by queenofbithynia at 12:13 AM on July 12, 2018 [3 favorites]
if they don't respond or don't want to tell you, that's an answer that pertains directly to the topic even if it doesn't include the numbers you're after.
posted by queenofbithynia at 12:13 AM on July 12, 2018 [3 favorites]
« Older Is it possible to move already-existing Notes from... | First Home Purchase: Home for my aging folks? Newer »
This thread is closed to new comments.
- Not everyone blocks users on Twitter. Many people just mute them. So you would need to account for that as well, since muted accounts can still comment/quote-tweet the harassed-person (unlike blocked accounts, which cannot). Also, even blocked accounts can "bypass" the block (I saw this a lot when Trump blocked people), which might make it appear like they haven't been blocked.
- Since troll accounts tend to change their @ handle often to avoid simple detection, you'll need to know their permanent Twitter ID (which you can search here or at a similar site) to 1) ensure continuity of data and 2) avoid duplicate data.
- Building on the above: while not enough of the accounts are taken seriously by Twitter/jack and remain online, a number of them end up suspended and they have backup accounts already in their holster.
- You may want to follow (and make contact with) accounts like (at) bitburner and (at) conspirator0, who are frequently scraping data on known bot/troll networks to determine who they're targeting, how they're targeting, and who is behind the accounts.
- You may be better off trying to scrape troll accounts based on avatars and usernames/visible names, rather than the things they say. You might get a lot of false positives just searching by bad/abusive language. This is because a lot of people use those words in completely neutral, innocuous contexts, and also because a lot of trolls are getting creative and go out of their way to find new ways to say cruel things while still narrowly avoiding Twitter's detection. The Pepe accounts still use Pepe ("rare Pepe", ugh) avatars but the new thing seems to be going by a name that uses the word "Groyper" - usually fitting the word into a celebrity or famous person's name - you know, cutesy shit, except not cute because they're racist. Search "Groyper" on Twitter and you'll see what I mean. There's also a weird obsession among the white supremacists with anime, and a lot of anime avatars, in turn.
I know this doesn't answer your question about automated data scraping, but I hope it gives you a better idea of what data to target for the most accurate results. If I think of anything else, I'll come back to post it.
posted by nightrecordings at 4:35 PM on July 11, 2018