The Twitter Equation

On the heels of Rachel Clarke's Twitter Frustrations, and a dinner with a friend last night I've gotten to think who I follow. I know I've blogged on some of this before but the way we use these tools changes as time goes on so I thought an update was in order.

When Rachel posted her article, she and I had a back and forth on twitter on the topic of twitter overload (tweeting about twitter is so meta yet so popular). How, if she followed everyone back who was following her, she wouldn't keep up the the flow and how this was actually a hindrance to her using the tool. During this conversation it occurred to me that where I don't care if someone follows me back and I'm pretty liberal on who I follow back (more on that below) that my ratio of followers:followed is generally around 1 to 1 (at this moment I'm following 290 folks and am being followed by 288). This was not the result of a conscious plan the keep those numbers close, it just happened organically. Rachel, being an A-Lister gets followed by most folks then I do.

Where I believe that part of Scoble's secret to twitter is correct, the fact that the value in twitter is in who you follow, not in who's following you, I disagree that sheer numbers is the solution. Driving his arguement to its logical conclusion we should all just follow the public timeline, like a giant IRC channel because that's how we see the most information and hence get the most value out of twitter. Where he once told me that all his followers are quailty twitter users, I don't think everyone out there is quality.

So... now the math kicks in.... It is sloppy math.. more like random thoughts on the topic.

So, what is a quality twitter user? I think that might vary from person to person since there are so many ways to use this tool. I think we can all agree the ratio of a person's tweets you like vs. tweets you don't like would be a first stab at doping out their quality scopre. Similarly, the value of all your followers could be expressed as the ratio of quality tweeters scores vs non-qualtiy tweeter scores (where non-quality is defined as a quality score less then one).

So...For a User's Quality

Q = T_pos / (T_neg + 1)

The value an end user can figure out the value twitter is basicly like this:

V = Sum(Q _pos) / (Sum(Q_neg) + 1)

I'll think more about this in the days to come but that's a first stab.