The recent controversy about celebrities fake Twitter followers brought to attention some cool online tools.
Having large amounts of fake followers is a sign that they’ve probably purchased their Twitter followers, and according to these recent “studies” almost every celebrity has them.
“The averages for the Top 15 Twitter personalities are 30.4% fake, 40.9% inactive and 28.7% real and active.” –Forbes
Fake Follower Check by Statuspeople.com is a web service that checks how many of your Twitter followers are fake. It’s a really great idea, but does it get it right most of the time? Statuspeople tells us the reports are inexact because the tool only measures a sample of your followers and then calculates the average.
This can be called a reverse Turing test. It is similar to the original Turing Test from the 1950s. In Alan Turings famous paper he describes how to interrogate an artificial intelligence to know if it’s really a human behind there. The purpose there is for the human to figure out the computer. In the reverse Turing test the computer tries to figure out the computer or human, to see if it’s a person.
These online tools look for algorithms that signify a bot, or fake follower. Someone who follows a lot of people and barely posts any content themselves would be seen as a fake by Statuspeople’s tool.
Other tools like Twit Cleaner looks for spammy behaviour such as repeat messages or links, to tell if a person is a bot. It gives you a report of all your followers/followings with various dodgy behaviour. Then you can figure out for yourself if they’re bots or not. A bit of a combined Reversed and Standard Turing test.
The best way to find out if a user is fake is still to manually check it, like the regular Turing test, maybe interrogate them =D. These tools can help line up some people with dodgy behavior, a list you’ll then have to moderate. It can still get it wrong as they’re not all bots.
CAPTCHA is another more common form of reverse Turing test you see on the web, in signup forms, etc. The idea is that only a human would be able to figure out the scrambled characters. But like the other tools not fool proof. It works mainly becasue the average user doesn’t have access to the right tools that can get past CAPTCHA, which does exist.
It’s quite common now actually. If you leave a comment here *hint* there will be a reverse Turing test in a way where WordPress tries to figure out if you’re spamming. By looking for certain keywords that spam comments usually have. Although I’m new to WordPress I think it works well against bots.