There are plenty of different tools out there to measure your influence on Twitter such as Klout, Twinfluence, Twitter ratio, and there are many more. I’m not sure all of these tools have the same definition of “influence on Twitter”. This is the same question we ask when we say “What is an authoritative website?”. PageRank and HITS came along with a solution to that problem, and then the cracks in that were identified and addressed by tweaking, changing things and trying new things out. I believe the same is true for Twitter influence, or Tweeter authority.
Right, anyway…(btw I just really liked that picture from konterfai)
Daniel Tunkelang from The Noisy Channel blog came up with a solution too, and I like it. He’s Chief Scientist of Endeca, and an experienced computer scientist, so you can trust him to come up with something that makes sense and this is pretty interesting. It’s “A Twitter analogue to PageRank”.
“If X says something, what will be the expected impact?” (Daniel Tunkelang)
Why meausre influence?
- I’d like to be able to measure my own influence, since one of my goals is to increase the leverage associated with my ideas. If I were a company, the same would apply to measuring brand capital. In my own case, I’d like to be able to check my balance of reputation capital.
- I’d like to know who the influencers are so I can monitor them and in some cases court them. Of course I’ll have other criteria about the people and their areas of expertise. But the ability to explore and the ability to sort by influence are complementary. For example, I’d love to know who are the most influential people tweeting about information seeking.
It is as he himself says, but a starting point for such a ranking algorithm, but this is a good start. Daniel does say “it’s sloppy for me to use “PageRank” as shorthand for stationary distribution on the link graph.”
How it works:
His solution allows for someones influence to be calculated recursively, taking into consideration the probability of a retweet. There’s a pretty equation for you to see on his post and he explains it well also, take a look.
He says: “It also strikes me as hard to game, since it isn’t counting retweets, and it’s hard to add much influence through followers who don’t have any influence themselves.” In this respect, as he says, it is similar to PageRank because as he points out it uses a stationary distribution of the link graph rather than the click stream.
The detailed post is available for you to read and I urge you too if this kind of stuff tickles you.
He observed “a follower who follows many other people adds almost nothing to a person’s rank”.
The basic breakdown is (from Jason):
1 – The amount of attention you can give is spread out among all those you follow. The more you follow, the less attention you can give each one.
2 – Your influence depends on the amount of attention your followers can give you.
3 – It also strikes me as hard to game, since it isn’t counting retweets, and it’s hard to add much influence through followers who don’t have any influence themselves.
Jeremy (tell me who you are and I’ll add a link!) comments on Daniel’s post and says “Wouldn’t it be more interesting to do exploratory Twitter search, and come across less influential, less connected voices, but perhaps who have something more interesting to say?” – I think this is a really good point.
I see Tweets that read “I’m 1 follower away from 1000!” and “Finally 2000 followers!” and things along those lines. I responded to one and asked “What do you do with all those followers?” The response was and I quote “Uhm….Idk just for people to see all my crap aka things i say(type)”.
What is the probability of all 1000 (or whatever) followers reading a given Tweet? How on earth does anyone read the immense flow of data coming from 1000 odd followers?
Different people have different reasons for using Twitter, and this is my short breakdown:
- Those using it to play “collect followers”
- Those genuinely looking to share useful information
- Those, looking to help customers
- Those who want to meet and interact with like-minded people
- Those who just read and rarely tweet
- Those who broadcast and never interact
A way of making more sense of the huge amount of data concerning not just tweets but also users and their characteristics can only help!
Be a part of the TunkRank idea:
If you want to join the discussion, do so. There are lots of good ideas floating around and it’s really interesting stuff. Follow @TunkRank on…you know where.