Posts Tagged ‘ranking’

Can we rank twitter streams?

November 13, 2009

‘Interesting article on twitter public. Thanks to you I am there.’ — Paulo Coelho, twitter, 12:19pm 11 November 2009

Can we rank twitter streams? What I mean by ‘rank’ is best illustrated by example.

I carry out a search on google for any of the following: Puerto de la Cruz, Guildford, Washingborough, Harry Parkins, Paulo Coelho. My first examples I am seeking information on place, if google returns information on flights, schools, football clubs, property for sale, then google has failed me. If I desired information on these topics I would add further key words accordingly. In my latter two examples I am searching for specific individuals, a veteran of WWII Bomber Command, a Brazilian writer. If google supplies different individuals, which may be what others desire on their search on these same names, then the onus is on me to refine my search with additional keywords.

How then can this work for twitter? The ranking we use with google has no meaning for twitter. Let us take an example.

A group of a dozen friends on twitter, each has around ten followers, chit-chat flows between them as would flow over a garden fence or down the pub. The group is an isolated network, ie no links external to the group. The group is not closed, ie their tweets appear on the public time line. Unless a third party stumbles across this network, no one would be aware of its existence. But even if they did, of what interest would they be other than to those students of social networks and micro-blogging? On the other hand the information flow within the network is of high value to the participants, that is why they participate.

Large number of followers is an indication of popularity, thus could be used as a ranking metric.

A crude measure at best. Every celebrity would have a high ranking. As does news sources, though I am surprised how relatively low at just over 27,000 is the excellent news source @bbcnews.

With around 50 followers I would fall way way down the ranking. It would be several times this figure, but I check out every follower, and unless they have something worthwhile to contribute, I delete them. If they are con merchants, flogging something, then I report as spam.

Numbers of followers grows exponentially. From 1 to 10, takes as long as from 10 to 100. The reason for this is if you have no followers, no one knows of your existence, with followers, more people know, the effect snowballs. Eventually though it must saturate.

If you wish to influence, to change the way society works and thinks, then it would be of greater value to have fewer followers who are themselves of influence, who in turn have followers of influence, than to be followed by the common herd, especially if followers of influence re-tweet your tweets all the way down the line.

Large amount of traffic, large number of tweets.

Celebrities tend to tweet all day about their meaningless existence. If I wanted this garbage I would subscribe to Hello and OK magazines.

I tend to avoid high traffic as it swamps out everything else. Or, it will have to have high merit. I will look at websites for example BBC News, Indymedia UK, but then I am being selectively in what I read. I look at @ClimateCamp, as they have high value, ie has merit, but I dip my toe in their stream, rather than subscribe as a follower. On the other hand I follow @airportwatch as they have high value but do not swamp me.

According to a Harvard study, around 10% of twitter users account for 90% of the traffic. The average tweet is less than one a month. Many users never tweet at all. I am one of the contributers.

http://news.bbc.co.uk/1/hi/technology/8089508.stm

My tweets are an eclectic mix of environmental information, philosophical musings, links to articles, links to pictures, quotations, re-tweets and chit-chat with friends.

Does a stream have merit? A measure of merit is are the tweets re-tweeted? My tweets are regularly re-tweeted, then often re-tweeted again.

Re-tweet can be seen as a signal-to-noise filter. It picks out what is worthwhile or of value from a twitter stream and thus enhances the signal-to-noise ratio.

Is there engagement, a dialogue, an interchange of views? Our network of friends would score highly, a celebrity merely broadcasts drivel to the mindless masses.

The Guildford Book Festival has a twitter account, but failed to make good use of it during the 2009 book festival. Occasionally during the book festival they would say what was happening that day. Comments to @GfordBookFest were rarely re-tweeted, unless from one of the invited authors (though my comment on this was at least acknowledged). A complete failure of engagement, no attempt to facilitate a discussion on the festival, which would of itself created more interest in the festival.

http://www.indymedia.org.uk/en/2009/10/440470.html

How many others users point to that twitter account? Another measure of the value of that twitter stream, cf a link to a web page or citing a reference.

Paulo Coelho is a Brazilian writer who has sold over 100 million books worldwide. First published over twenty years ago, The Alchemist has sold 20 million copies and recently celebrated two years in the New York Times best-seller list. Not surprisingly @paulocoelho has approaching 200,000 followers on twitter. He is regularly re-tweeted by his followers, some like parrots re-tweet everything they receive, others add value by adding variants. There is a dialogue and discussion, the traffic is two way. His twitter stream drives traffic to his blog, the blog drives traffic to his twitter stream. The blog is often more of a discussion forum. Paulo Coelho poses a thought or a topic, there is then a discussion between all participants. He has also devised a novel use of twitter, for interviews. Followers are invited at an alloted time to post questions, the first five in English and Portuguese he answers.

I recently heard a puerile discussion on twitter on the Moral Maze on BBC Radio 4. The participants did not use twitter, had no idea how twitter worked.

http://www.bbc.co.uk/programmes/b00nkcfk

Twitter and social networking came of age during the flawed Iranian elections this summer. Twitter, and to a lesser extent facebook, were used to organise street protests, to avoid trouble. The face of the revolution was Neda. The first I knew of Neda and her senseless slaughter, was when Paulo Coelho posted a video clip on his blog. This was a day or so before it was picked up by the mainstream media. Those who saw it on his blog, reposted, tweeted, what he had shown. It went viral and spread around the world. Following the Iranian example, activists are now making use of twitter, both to provide information and to coordinate actions. For weeks Neda and Iran were trending topics on twitter.

http://www.indymedia.org.uk/en/2009/07/434151.html?c=on
http://paulocoelhoblog.com/2009/06/23/iran-by-neda/
http://paulocoelhoblog.com/2009/06/26/the-doctor/

I was recently at a meeting with a surviver of the Tiananmen Square Massacre of twenty years ago. He spoke of how it had been erased from the Chinese collective consciousness and how important it was to keep that memory alive. He said the Chinese feared the internet, that there were more police patrolling the internet than patrolling the streets.

http://keithpp.wordpress.com/2009/10/23/eyewitness-account-of-tiananmen-square/

How therefore to rank twitter streams or twitter users? There are those who are trying. On the metrics I have discussed, our isolated network of family and friends would rank very high. They would agree with that ranking as they would value their network, but would it be of any interest to those outside of that network? Paulo Coelho would also rank very high, and for those who follow him they clearly place value on what he has to say, but would those outside his followers express any interest? Celebrities would rank low, even though they have a large following and often generate a large amount of traffic, for the simple reason they have nothing worthwhile to say and there is no engagement.

What we are trying to measure is influence and the numbers influenced, which may be via re-tweets. Influence does not directly correlate with number of followers or the number of tweets, indeed the higher the number of tweets the likelihood is the lower the signal-to-noise ratio, but these are factors to be taken into account. The ratio is likely to be logarithmic, that is going from one to ten followers has the same impact as going from 10 to 100 or 100 to 1,000. With the number of tweets and lowering signal-to-noise ratio, there must be some ceiling or saturation point that is approached.

What is your TweetLevel? Is it a meaningful measure?

For what it is worth the day before this article was posted tweetlevel gave me 34 and for comparison Paulo Coelho 73. I am advised to follow more people, to increase my followers, engage more, my trust level is pretty good. Maybe more interesting is how these scores vary over time. The danger of course is that people will tweet to maximise their metrics rather than because they have anything useful to say (always assuming of course they had anything interesting to say in the first place).

In reality not something to get hung up about. A bit of fun, nothing more. Rather like those machines you find on the pier where you grip a handle or place your hand on a metal plate and out pops a little printed ticket with your future mapped out. The value of any particular twitter stream is going to be what value you put on it, unless you follow like sheep.

Special thanks to Paul Coelho, whose tweet of a couple of days ago gave me the germ of an idea.

Also see

how to measure ‘importance’ on Twitter

He’s also got a cat called Retweet …

The Winner Stands Alone


Follow

Get every new post delivered to your Inbox.

Join 561 other followers