Quantifying the impact of the Twitter fake accounts purge – a technical analysis(elie.net)
elie.net
Quantifying the impact of the Twitter fake accounts purge – a technical analysis
https://elie.net/blog/web/quantifying-the-impact-of-the-twitter-fake-accounts-purge-a-technical-analysis
20 comments
I wish one could get the dataset and see how many bots some specific people, like politicians, might have had as followers.
I find it interesting that the bots preferred Obama, even on a percentage-of-followers basis. What's the explanation? Were Obama's twitter numbers being pumped at some point? Do bots just use Obama's account as a default-follow for legitimacy?
might be the function of the default follow feature and when they were created
We are going to make the dataset public with the paper. I am happy to share it privately in the mean time.
I have to say I rather enjoyed the snide, "Determining how those accounts were able to make such an impressive recovery is left as an exercise to the reader." in relation to users that saw large number of bot followers previously banned, followed by them rapidly gaining huge numbers of new followers... somehow.
I don't imagine the creators of the bot accounts were too upset by Twitter eventually banning millions of their creations. Now they get to sell millions of additional accounts to the same people all over again.
It doesn't take any fancy machine learning to figure out what is happening.
It doesn't take any fancy machine learning to figure out what is happening.
Brilliant. Twitter gets let off the hook by appearing to be "doing something" and people who charge for followers get a revenue boost by this new form of planned obsolescence. Twitter users get screwed by either getting caught up in the purges or having to pay even more money to bot creators to inflate their numbers, likely pricing many out of the market. Seems like a good time to short Twitter stock...
My guess: many humans were caught in the purge. I've seen people explain that Twitter locked their account, asking for a phone number for verification. While they were locked out, they were subtracted from the follow count of the accounts they followed.
My guess, new fake accounts. The best evidence for that is the following snippet from the article:
The most striking example of those massive rebounds is the account of @AdelAliBinAli. Before the purge, this account had 6.2M followers. On July 14, its follower count was down to 2.7M followers (-3.4M / -55%). Between July 14 and July 19, it regained 1.2M followers, and as of July 23rd, this account now has 6.6M followers, which is even more than before the purge and represents 3.9M additional followers gained.
As for the others, having more than a statistically average number of purged followers is evidence that you were buying fake followers in the first place. The obvious explanation for large increases in followers for someone who is already known to buy them is that they bought more.
The most striking example of those massive rebounds is the account of @AdelAliBinAli. Before the purge, this account had 6.2M followers. On July 14, its follower count was down to 2.7M followers (-3.4M / -55%). Between July 14 and July 19, it regained 1.2M followers, and as of July 23rd, this account now has 6.6M followers, which is even more than before the purge and represents 3.9M additional followers gained.
As for the others, having more than a statistically average number of purged followers is evidence that you were buying fake followers in the first place. The obvious explanation for large increases in followers for someone who is already known to buy them is that they bought more.
I don't think we should lose sight of the fact that many human users are being caught up these "purges". I lost my twitter account because they determined I was a "bot" somehow. I didn't want to give them my phone number to prove that I was a human so I ended up having to make another account. Social media has become incredibly toxic in the past few years with hate speech and the self-proclaimed social justice heroes who think banning anyone for wrong-think is necessary to "save Democracy". The whole thing is a mess.
Fortunately for Twitter, the bots lacked the tenacity and capability for repetition that the humans had and did not create new accounts.
I assume you meant to add a /s to that
Exactly. I made an account to post useful links but apparently posting a link flags the account as spammer unless you want to give them your phone number (I don't).
Problem is the electorate isn't educated well enough to use critical thought and reasoning to see through the propaganda.
I was banned from twitter, despite never having tweeted, my appeal was ignored, and I don't even get the phone number option!
I didn't expect the number of followers loss to be over half a billion. Makes me wonder how many accounts were purged. Even if each of the purged account followed 100 of the account measured, it still means millions of accounts were purged.
Twitter accounts were trivially easy to make. Bots could easily churn out tens of thousands or more in just a few hours. The whole service is lousy with bots.
I tried to create multiple accounts (from the same IP mind you) about 3 months ago and many of them required me to verify a phone before registering or got flagged shortly after as potential spam, which was accurate.
I was curious how easy it was and for me it wasn't super simple especially the phone verification step.
I was curious how easy it was and for me it wasn't super simple especially the phone verification step.
The phone verification step appeared when Twitter finally realized (way too late) that they have a bot problem.
A few years ago (before the 2016 election) they didn't give a crap.
They've only started purging bot accounts in the past year or so. There was an uproar from the Fox News crowd about how they were being unfairly targeted by liberal Twitter because their subscriber counts were dropping like flies.
A few years ago (before the 2016 election) they didn't give a crap.
They've only started purging bot accounts in the past year or so. There was an uproar from the Fox News crowd about how they were being unfairly targeted by liberal Twitter because their subscriber counts were dropping like flies.
Did you mean "For example, if an account had 1,000 followers pre-purge and 900 post-purge, it lost 100 followers. Its followers base loss would therefore be: 100/1000=0.1, or 10 percent."