Warning: do not read this unless you are obsessively interested in twitter and have nothing better to do with your time.
I’ve been stimulated into thinking about this by various sites I’ve seen recently which measure your online connectedness – plus some tweets and blog postings by Tony Hirst. Some of these measures are pretty crude – for instance measuring how connected you are on twitter by adding the number of people you follow to the number of people who follow you. What if you hardly ever go into Twitter? Or if most of your followers just registered for Twitter once and then gave up? In that case you’re not as connected as you might seem.
Taking your followers first, can we find a way to measure the “quality” of your connectedness to those people ie to know if they’re actually viewing your tweets? The more often they go into twitter the more likely they are to be reading what you’re writing. The easiest way to measure how often they’re going into twitter is to see how often they’re tweeting themselves. That person has a twit value which could be based on their average daily number of tweets sent, perhaps calculated over the last month.
Say you have 3 followers: Martin who tweets on average 5 times per day, Gráinne 7.5 and Mary 0. Adding your followers’ twit values together gives 12.5 which we can call your tweet impact. If your average number of tweets sent is 2 then you have a daily tweet impact of 25. So that’s one part of our measurement for connectedness.
If you only send tweets and never read others then you’re only connected in one direction which is not very connected… So we also need to measure the number of tweets you read. We can’t know how many you bother reading of course but we can calculate how many you receive. Of course if you never go into twitter then receiving 1000 a day is pointless. But the more you tweet the more likely you are to be reading your tweets so we can base your own twit value on your average daily number of tweets sent.
So if you follow 4 people, Martin, Gráinne, Mary and Tony (who tweets 11 times per day) you can add up their scores to get 23.5. As your average daily tweets are 2 you can multiple 23.5 by that to get 47. This is a measure of the impact on you from the tweets you receive.
That would be quite a good place to stop but unfortunately my brain keeps ticking over here and there are other factors which can be considered. Direct tweets I propose should be discounted because they’re equivalent to emails (and it’s impossible to find out automatically who those direct tweets are to).
What about @tweets then? These are likely to have a much higher impact on the person you’re sending them to than one sent to all your followers. @tweets should contribute to your interconnectedness score in a different way to normal tweets. They should get a score higher than just the twit value of the person they’re aimed at, say five times the twit value. And for everyone who’s following you and that person, they are also likely to be impacted more than just with a general tweet because they’ll be interested in what that says about the relationship between the two of you. Perhaps your score should be double their normal twit values to indicate the higher level of tweet impact.
Thus if you send an @tweet to Martin, who’s twit value is 5, the tweet impact on him is actually 25. But Gráinne also sees that tweet because she’s following both of you. Her twit value is 7.5 but the tweet has greater than usual impact on her because she takes more interest in it so it doubles to 15. The total tweet impact then is 25 + 15 = 40.
@tweets received by you could be scored similarly.
One final factor which occurs to me is the relative value of tweets you read from people who are also following you. There’s a good chance you’ll know that they’re your followers therefore you’re more likely to take note of the tweets they’re sending you. The tweet impact could therefore be increased, say by a half.
Now that I’ve got this rubbish out of my head, I should now be able to get to sleep.
[…] As to why am I picking on Niall…? Well, apart from making the point that by engaging in ‘public’ social networks, other people can look at what you’re doing, it’s partly because thinking about this post on ‘Twitter impact factors’ kept me up all night: Twitter – how interconnected are you?. […]
So when are you going to write the formula to work all of that out …
And, does it make a difference, if you & your group of twitterers are a closed group & none of you have any other followers/ are following anyone out side the group. …
Emma
Tony, sorry I kept you awake all night – but I blame you for me not being able to get to sleep till 2am in the first place.
Emma, I don’t see that it would make any difference to do this analysis over Twitter as a whole or over just a sub group (as Tony’s been trying to do with Open University staff). In fact it would be interesting to see to what extent people who are well connected in an institution are also well connected across Twitter as a whole.
Here’s a quick attempt at the formula:
[twit value] = Average tweets per day of a person; n = number of people following you (followers) or followed by you (followed)
so the simple version is:
[twitter interconnectedness] = [your twit value] * ([twit value of follower 1] + … [twit value of follower n] + [twit value of followed 1] + … [twit value of followed n])
Taking into account @tweets it becomes:
[twitter interconnectedness] = [your twit value] * ([twit value of follower 1] + … [twit value of follower n] + [twit value of followed 1] + … [twit value of followed n]) + [your twit value] * 5 * ([@tweets sent by you to @tweet receiver 1] * [twit value of named @tweet receiver 1] + … [@tweets sent by you to passive receiver n] * [twit value of named @tweet receiver n]) + [your twit value] * 2 ([twit value of passive @tweet receiver 1] + … [twit value of passive @tweet receiver n]) + [your twit value] * 5 * ([@tweets sent to you by @tweet sender 1] * [twit value of @tweet sender 1] + … [@tweets sent to you by @tweet sender n] * [twit value of @tweet sender n]) + [your twit value] * 2 ([twit value of passive @tweet sender 1] + … [twit value of passive @tweet sender n])
I didn’t take into account the final factor and I haven’t checked this properly. But you did ask and I bet you wish you’d hadn’t now…