influence |ˈinfloŏəns|
noun: A person or thing with such a capacity or power
Much has been made this past week of Shifted Pixels' Top 100 most influential Australian Twitterers. It's a praiseworthy exercise undertaken at a time when Twitter is exploding in popularity. Prominent politicians are now using it in droves, even Britney Spears is getting in on the act, and these Twitter-focused stories are making the front pages of newspapers and news sites across the world.
The problem is that this list is fatally flawed.
Twitter itself is becoming a societal pulse for people who spend a vast amount of time plugged into the web (and the phrase "online" is also becoming less relevant as people are continuously connected). Given that the web is becoming the defacto platform for practically anything and everything, you could argue that Twitter is becoming a societal pulse - full stop.
Subsequently in terms of influence, yes, Twitter itself is becoming massively influential, allowing people to virtually flashmob news and events that are then covered in detail and by dozens or hundreds of people, and all manner of conversations and exchanges are taking place at any one time. These and other ways in which Twitter is being used are absolutely fascinating to observe and participate in. The ebb and flow of conversation is epic (there's a book begging to be written sometime soon).
And yet, as clear as Twitter's rise as a service is becoming, determining an individual's influence on Twitter is not so easy to do, and certainly not something you can do with a simple and seemingly ham-fisted calculation: TwitterRank = (Following / 10) + (Followers) + (Updates/100).
Any possible way of measuring a person's influence would, with the information immediately available, have to start with the number of people following that person.
Agreed.
People follow other people because there's a stab of common interest, of information that's useful or of value between for those two people. Based on my own experience the levels or relevancy are numerous, complex, and can change on a daily basis.
But, the number of updates a person has posted and the number of people a person is following themselves? These two pieces of data are irrelevant in the context of determining influence, meaning that 2/3 of the factors used in the calculation that are irrelevant, which in turn makes the calculation fundamentally flawed.
Let me say that again slightly differently: the number of updates someone has posted and the number of people they're following bears no significance on their influence whatsoever.
Perhaps as a measure of an individual's own activity, this calculation would be valid, but I can still see no reason to slash certain aspects of the data by 90%.
The problem with using personal activity data is simple: It's blindingly obvious that you could game this system for the number of people you follow and the number of posts you create.
At best, the calculation in question produces mildly indicative results of influence, but only early on before people start messing with it. (This is why lots of recognisably influential people are in the list to some degree).
To make headway in truly determining who influential Twitterers are (and truth be told I'm not ultimately sure it's going to work in itself), we have to look at who is being influential. That is, who is generating and leading discussion.
Again, the best starting representation here is the number of people following you. A further indicator would be how many @replies you have against your name from other people (not how many you write yourself). This data, while not immediately visible, is available. What this would indicate is how much conversation your activity produces - surely a far more worthy indicator of influence when considering a social network.
Harder but very useful data to have standalone or as part of a calculation would also be the number of links posted by someone, multiplied by the number of hits on that link (including photos etc). Twitpic offers hit stats on posted images, for example.
You could further build on this by seeing when and where the growth in followers and conversation occurred for particular people. For example, was it a particular event where someone grabbed the attention, or did they generate a whole new discussion. Ross Dawson's Future of Media event springs to mind. I wouldn't mind betting that people like Chris Saad got a whole host of new followers after that event.
All of this data represents and allows you to measure the activity generated by the person - "the ripples" they create - not their own personal activity. That's a vital distinction to make.
Anecdotally, the most influential local person (so that's discounting @BarackObama!) I've seen and followed on Twitter is Mark Pesce (@mpesce). Here's a guy who seemingly increased the viewing figures and visibility of his TV show (New Inventors) through Twitter.
Another influential guy in all those that I personally follow is Stephen Collins (@trib), who led me to people like Pesce, @Kcarruthers, @Stilgherrian and many more. These are people who lots of other people visibly follow.
A proper social network analysis using incremental data from Twitter would be amazing in determining true influence but that data would be hard to come by and, I expect, impossibly dense (it would probably require tracking "@newfollower is now following you" emails sent out by Twitter's servers).
Also notice the fact that the above is in my sphere, meaning influence on Twitter, while broadly measurable, is actually pretty subjective and will become increasingly subjective as use of the service grows beyond the web/tech/media areas of online discussion. Twitter Influence cannot be periodically conclusive in the way the "Top 100 rich list" is.
This discussion on influence seems to be an important one given Twitter's happening mainstream breakthrough, so it's great to see the discussion get going. But determining influence via the method used so far is, to put it simply, a waste of time.
