Lately I find that there is an overwhelming amount of people who seem to get a kick out of collecting followers just to expand their audience. There are also a number of theories on how to increase the amount of followers.

A Twitter Bot Contest

I am not too interested in the quantity of followers myself, but I do find this interesting from a strategical point of view. I think it would be very interesting to test different theories on maximizing the amount of followers by setting up a number of twitter bots with different strategies to see how well they perform.

The idea for a twitter bot contest comes from reading Robert Axelrod‘s book “The Evolution of Cooperation” many years ago. The book described different algorithms competing to perform best at Prisoner’s Dilemma (among many other things). Competing to get twitter followers is very different from the tit-for-tat competition, and many times more entertaining since (mostly) real people are involved on the other end.

Twitter ranking

There are a lot of different ways to measure twitter success – from the most basic following / follower ratio to more advanced twitter ranks that takes the amount of people referencing your account into consideration. A ratio is simple and nice, but it does not favour a user following 10.000 people that he follows back over a user with 10 people that follow each other.  It is obviously a lot harder to get 10.000 people into your network than to just invite your friends, so quantity should be rewarded.

A success criteria for such a bot could be the number of followers with a punishment for the number of people the bot is following. After all, following someone is very easy, but getting them to follow you back requires more effort. (For a real person it is probably more ethical to follow the people following you back, but getting a lot of followers without following them back is a lot harder to do).

I came up with the following equation to achieve the score distribution that I was after:

Just following someone gives 0 points if nobody follows back, getting lots of followers without following anyone back gives you a score equal to the number of followers, and the same number of following and followers gives a score equal to half the followers. I don’t think the number of tweets or retweets are impotant in determining the score because they both affect the number of followers. So making the number of followers most important parameter should cover it.

Plotting this equation gives the following graph:

I have made my own test bot. It has been running for five days, and it is now following 1.737 people with 706 people following it back. This gives it a score of  just 152 according to the above formula. The bot is still running, so I will not reveal its name yet because I don’t want to influence its performance by promoting it.

What do you think?

The score could of course also be used to rank normal twitter accounts. Let me know if you think this is a useful way of ranking twitter users.

Now it’s just a matter of someone else building a bots to compete with this score. Let me know if you are half as geeky as I am these days and join in. I’ll gladly share some rules and the code to get this going.

Oh, and you can follow my updates on twitter if you want to.

]]> http://herrklein.com/2009/03/twitter-ranking/feed/ 2