In the recent past we have been made believe that the sheer number of tweets mentioning a link is the measure of popularity on Twitter. I don’t speak about popular users here. I mean popular items or tweets.
Twitter is neither Digg nor Google so measuring tweet popularity based on links is just half of the story.
Lately three services contest this assumption with three alternative ways of measuring tweet popularity:
- Tweeply counts the number of @replies as a metric to determine the most popular tweets
- Favstar counts how often a tweet has been bookmarked or rather added to favorites (clicking the star icon on Twitter.com)
- Finally Topsy has found a way to find out which links within tweets are really popular by taking only influential Twitter users into account
The first thing you’ll notice when comparing these three services with each other and other simple link counting interfaces like
is: completely different tweets are popular depending which site you use.
There are a few reasons for these differing results:
- There are bots of all kinds on Twitter. Legit bots or RSS collecting Twitter accounts and spam bots creating new users by the thousands. How do you filter them? Topsy seems to have found a solution by focusing on a white list of influential users, those Twitter users who are highly connected and active on Twitter.
- Replies are real conversations and only those having a huge following can get many replies. Retweets in contrast can spread from one user to another with no obstacles. Replies are one to one conversations. One speaker speaks, and many reply.
- Adding something to favorites on Twitter only happens on a rare occasion. The most favorited tweets of mine after more than a year have been only bookmarked twice! So counting favorites has the potential to identify the true gems as of now.
Some at the end of the day measuring popularity on Twitter is not as simple as you might have suspected. Of course it’s most probably a combination of all three above. The number of retweets, replies and bookmarks should be combined and analyzed in a more complex algorithm to determine true popularity. Of course the influence of the user contributing the tweets must be measured as well. Let’s see who figures that out first…