This article was published on August 5, 2010

Are Twitter’s recommendations a first look at Twitter Analytics?

Are Twitter’s recommendations a first look at Twitter Analytics?

A few days ago, Twitter started suggesting users for you to follow. The initial suggestions, it seemed, were based upon those who your friends followed. Then Twitter took the service offline prior to launching it site-wide.

Apparently, at least according to Dave Winer, Twitter took that time to implement some new ideas into the system.

Winer notes, on Scripting News, that the recommendations he’s seeing now appear to be taking the context of his tweets into consideration. Since he tends to post about the bit torrent protocol quite a bit, he’s seeing suggestions for others who do the same. Could this be a by-product of Twitter’s push toward better analytics?

Taking a look at my own suggestions, I can’t say that I see the same results as Winer. The vast majority of everyone suggested is the same as what was suggested a few days ago. However, it’s fair to note that almost everything I tweet from my @BradTNW account is related to subjects that these people would tweet about as well.

When Twitter took the service offline, we sent a request for any additional information, and were simply told “See @Support”. Clearly, if Twitter has changed the algorithm behind the suggestions list, we’re not going to get any insight via easy methods.

We’ll keep digging and of course let you know what we find. For now, however, have a look at your own suggestions and see if you can tell any difference. Are there users there who your friends don’t follow? That would be a major sign.