RSS feed
Twitter rankings – sense and sensibility
Everybody loves ratings, especially if you’re in there. Most of the time you know it’s crap, but it’s a vanity thing I guess 
Twitterholic is doing an automated hitlist, and you can filter by location. Here is the top 100 for Belgium. It’s bogus to start with because it only filters out these people who have “Belgium” as a location in twitter. People using like a city are left out.
What always bother me is that it’s just a rating by followers. Sure, this means something, big time. But it’s not the only thing that matters. If Stephen Fry is followed by almost 250k people, that’s big! Same goes for a lot of other people as you can see on the overall top 100 on Twitterholic.
What I think has some value as well is the following/followers ratio. And there’s 2 ways to look at. 1/ you can start following back people who start following you (polite) or 2/ just start following lots and lots of people and hope they’ll start following you back (or at least a decent chunk).
Don’t get me wrong, you obviously can follow as many people as you like, and there is absolutely no judgment. And I know there are people out there we live by it, and want the biggest possible gap between followers and following. Live and let live, but I don’t play these games. If somebody follows be I usually check that person shortly. If it’s somewhat interesting (for me) I might follow back. But I’m kind of picky because I simply can’t focus following too many people at once. Other than that, if I meet people, or I see them speech or something, and they have a twitter account, big chance I’ll follow them. I guess this is more or less how most people do it.
Because of this, I do think the ratio has some kind of value. I did some kind of experiment with it, and used the twitterholic numbers to calculate a different list.
1/ followers – following
Just a simple subtraction. Theory is to take only the number of people into account that are following you on top of the people you’re following yourselves. Here’s the new top 10 (Belgium)
|
# |
who | site | followers | following | difference |
|
1 |
Veerle Pieters | Blog | 7313 | 192 | 7121 |
|
2 |
Paddy Donnelly | Blog | 1231 | 182 | 1049 |
|
3 |
Bart De Waele | Blog | 1447 | 403 | 1044 |
|
4 |
Robin Wauters | Plugg | 2717 | 1748 | 969 |
|
5 |
Serge Jespers | Blog | 1040 | 166 | 874 |
|
6 |
Kris Hoet | Blog | 1009 | 206 | 803 |
|
7 |
Clo Willaerts | Blog | 1198 | 722 | 476 |
|
8 |
Steffest | Stef.be | 765 | 373 | 392 |
|
9 |
Imke Dielen | Blog | 648 | 285 | 363 |
|
10 |
Security4all | Blog | 615 | 253 | 362 |
Since this is dynamic I pasted the numbers in here as well. So this is a snapshot, and it will change for sure.
2/ followers/following ratio
Divide the numbers of followers by people that person is following, and you get a number indicating that for every people he/she follows, he’s followed back by x people. I left out the accounts that aren’t following anyone (this is just push, no conversation) and as well the radioo accounts (again, push).
|
# |
who | site | followers | following | ratio |
|
1 |
Veerle Pieters | Blog | 7313 | 192 | 38,09 |
|
2 |
Jan Van Boghout | Macrabbit | 160 | 9 | 17,78 |
|
3 |
Paddy Donnelly | Blog | 1231 | 182 | 6,76 |
|
4 |
Serge Jespers | Blog | 1040 | 166 | 6,27 |
|
5 |
De Standaard Online | Standaard (newspaper) | 180 | 35 | 5,14 |
|
6 |
Steven Pauwels | Blog | 394 | 79 | 4,99 |
|
7 |
Kris Hoet | Blog | 1009 | 206 | 4,9 |
|
8 |
Bart De Waele | Blog | 1447 | 403 | 3,59 |
|
8 |
Appelogen | Blog | 242 | 70 | 3,46 |
|
10 |
Juan Pablo | Domain reactivator | 353 | 121 | 2,92 |
(I left out Christophe Logiste as well. Although in the Belgian list he seems to be French)
I think this worked out quite well. It’s not perfect though. How do you think this “algorithm” can be improved. Must be something with standard deviation I guess, because there are a few results that are kind of twisted.
It’s hard to determine, but there is like a number we have to take into acount. For example if you go over 500 followers this must mean something. On the other hand, following over 500 (even less) people can’t be efficient. Again: no judgment! But it is my believe that if you are following over 500 people it’s more to get your message out, over communication. Sure you can use tweetdeck or something else to structure the people you’re following, but this kind of just proves the point.
It’s just an open question: how can we optimize?
In any case, we can be pretty sure the most influential Belgian twitterer is Veerle Pieters. Totally deserved, she made herself count in the global design world.
I’m also pretty pleased I know more than a few of the people listed above quite good and in person. They’re people that definitely move things forward, and I praise them for that!
How am I doing?
In the original twitterholic list I’m number 44
Using the first tweak brings me to place 39
The second tweak puts me at 35
Going in the right direction, but still some work to do
You can follow me on twitter at http://twitter.com/minorissues
I think you’re making a crucial mistake in taking the published number of ‘following’ at face value.
Since I started working with Tweetdeck, I’m following everyone that follows me. This does not mean that I read every tweet of all of them – I’m selecting a few relevant people for me, and put them into a ‘must read’ selection in Tweetdeck. All others I just read incidently. Of course, I’m reading intently every @netlash tweet.
This means that the ‘following’ number that is published on my public profile (662) is no way the number I actually follow (which is more like a 250 some).
Same comment as Bart.
I follow at least 500 people who are no longer active on Twitter, but it’s more of a hassle to find those and remove them since they don’t bother me anyway.
Ah, the numbers. Fascinating. I need to find some way to get public statistics in the next thing I build, since everyone always seems wildly fascinated by ratios and lists. All of this can be brought back to human nature. Just something to make us feel good about ourselves. However, that’s not necessarily a bad thing. Just as long as enough of those “connections” are real.
What do you guys suggest? You’re influencers yourselves, that’s what we know (apart from the numbers). You have to admit this followers/following discrepancy does play a difference, right? I guess my attempts aren’t right after all, but it’s a start
For me this feels natural. If someone starts to follow me, and I see he’s following 1000ths of other people, I loose interest quite often. I know I’m generalizing, but I feel a lot of these people are just doing it for the numbers. Not too interested in what you have to say.
If we look at Veerle for example. Her numbers are amazing. But would you consider them as impressive if she was following 7000 people herself.
Influence tells us something about people wanting to follow you, to hear what you say, no matter if you follow them back.
But the followers number itself is saying something of course. So my formula isn’t right anyway. Would mean that if Veerle was following 7k peeps, she would have no influence whatsoever .. that’s wrong obviously.
Perhaps this is more close: followers + (followers-following). This keeps the influence of the high number of followers, and as well as the discrepancy.
On the above list this would mean this top 10:
1 Veerle Pieters (vpieters)
2 Robin Wauters (robinwauters)
3 Bart (netlash)
4 Technology Geek (TechnologyGeek)
5 Paddy Donnelly (paddydonnelly)
6 Stefan Fields (cashtrix)
7 Serge Jespers (sjespers)
8 Kris Hoet (crossthebreeze)
9 dirkdupon (dirkdupon)
10 Clo Willaerts (bnox)
And Bart: I get you want to follow people back, it’s polite and everything. And tweetdeck is great for this. But don’t you think it’s bogus as well. Aren’t you telling these people: I’m following you, but yet I’m not?
Just a question, no judgment whatsoever! Just want to understand and optimize.
Cheers!
It is, and it is not.
Especially when people keep their accounts closed (as in: only viewable by followers).
There are moments when I read everything in my timeline – of all people I follow. And there are moments when I skim.
But I *always* read the things that are aimed directly at me. If this comes from a closed account, I can not see the context. If I follow them (and they allow me), I can read the context.
Steven,
@tferriss has a very small ratio (157/ 21,655) and @ChrisBrogan follows almost as many people that follow him.
So who’s the better person to follow?
Tim Ferris doesn’t engage his followers that much, whereas Chris provides a lot of value to many folks Twitter.
One thing I look at is the number of @’s or the “conversation quotient” on Twitter-Friends. Read more here: http://www.twitip.com/measuring-your-twitter-networks-health/
In the end, I guess it depends upon your goals for using Twitter.
John
Thanks all for your comments, they all make sense!
It seems there are too many exceptions to define a model or algorithm.
However, I still have this gut feeling there is something about it. With all these hit lists, if you’re really on to it, you can force yourself in there by just starting to follow lot’s of people and without offering too many value. In a country as Belgium (which is rather small) it even isn’t difficult I guess. Not sure what would be the way to take this into account evaluating value (which is indeed something else than having lot’s of followers).
What about this observation: apart from the amount of people following, I hardly see any great twitterati following more people than people are following him/her. Is this something that stands: if a person is following more people than people are following him/her it’s probably a bad indicator.
I see an exception when someone is new to twitter, and finding the way .. but in this case, I guess, you can’t yet speak of a certain status anyway.
And of course there are other exception, like yourself john .. As I see your twitter I consider it valuable, but actually I hardly see it elsewhere.
Steven,
“Is this something that stands: if a person is following more people than people are following him/her it’s probably a bad indicator.”
Yes – that’s a great indicator, although it’s more of a ratio thing to be considered as only one data point among many when assessing a Twitter user.
Great discussion!
John