They use a number of different criteria to identify fake Twitter accounts, including:
The account is following less than 50 people and has less than one follower
More than 30% of all tweets use spam phrases, such as “diet,” “make money,” and “work from home”
The account is more than two months old and still has a default profile image
[/quote]
I don't know exactly how these things were weighted to decide that an account is fake... but apart from the second point, it sounds like "Fake" account could be more like "I signed up just to follow this guy and then forgot about it" account.
You can't apply this methodology to accounts belonging to such large celebrities for one important reason: when registering for a Twitter account you're prompted to follow "recommended" people. Someone could be told about Twitter, register an account, click "follow all" on the recommended page and then never sign in again... Justin Bieber would have gained a legitimate follower but they wouldn't seem real.
This methodology can only really be applied to smaller accounts, ones that have not had exposure through the Twitter recommended users system.
Here is an example, after creating an account you go through an introductory process that requires you to follow at least 5 people to continue: http://i.imgur.com/xT0tc5c.png and then on the next page you have to follow another 5.
Not to split hairs, but what's the difference between a bot logging in one time and following a bunch of people never to log in again and a breathing human being doing the same?
I built twitteraudit.com for fun to find out if people purchased followers.
Like the article says, these tools work best for people with less than about 500k followers... because once you cross that threshold, you get closer to the actual ratio of real/fake users on all of Twitter.
From the article:
[quote]
[/quote]
I don't know exactly how these things were weighted to decide that an account is fake... but apart from the second point, it sounds like "Fake" account could be more like "I signed up just to follow this guy and then forgot about it" account.
I would assume tweeting spam is weighed heavily. I doubt they care about simply inactive users.
It's actually very hard (at least or even) for humans to categorize twitter users as spammer/fake or real.
Ironic considering the fake HN users ITWorld has been running for years.
https://news.ycombinator.com/submitted?id=joewooglin
https://news.ycombinator.com/submitted?id=tech77
https://news.ycombinator.com/submitted?id=jfruhlinger
https://news.ycombinator.com/submitted?id=abennett
https://news.ycombinator.com/submitted?id=bdking
and they want to run an app against your twitter account that can post and change your profile
You can't apply this methodology to accounts belonging to such large celebrities for one important reason: when registering for a Twitter account you're prompted to follow "recommended" people. Someone could be told about Twitter, register an account, click "follow all" on the recommended page and then never sign in again... Justin Bieber would have gained a legitimate follower but they wouldn't seem real.
This methodology can only really be applied to smaller accounts, ones that have not had exposure through the Twitter recommended users system.
Here is an example, after creating an account you go through an introductory process that requires you to follow at least 5 people to continue: http://i.imgur.com/xT0tc5c.png and then on the next page you have to follow another 5.
Not to split hairs, but what's the difference between a bot logging in one time and following a bunch of people never to log in again and a breathing human being doing the same?
I built twitteraudit.com for fun to find out if people purchased followers.
Like the article says, these tools work best for people with less than about 500k followers... because once you cross that threshold, you get closer to the actual ratio of real/fake users on all of Twitter.
It's not hard to see who is likeley to have bought a ton of fake followers: http://www.twitteraudit.com/michaelgonline http://www.twitteraudit.com/FidelHerrera http://www.twitteraudit.com/hackmurphy
So ... has no followers then?
It sounds like my twitter account is fake. I don't produce. I only consume.