Over the last day or 2 I’ve been dusting the cobwebs off my PHP toolbox to help André out with neat little idea he came up with. The idea is that on Twitter, the results are more current, and therefore in some cases, more relevant.

While debating the merits of Lightroom versus Aperture and unfortunate cropping in landscape composition over a coffee, the discussion eventually turned to the project at hand and the topic of authority when ranking search results. When searching on Twitter, you get the most recent results first, but as the noise level increases, so does the uselessness of the results. Thanks to the fine minds at Google, their results are ordered based on some kind of authority, so what would it take to do the same with Twitter?

Our initial thoughts were around followers – the more followers I have, the more authority I should have. But that determines popularity. And after sitting through a year with Ms Popular in my Std 8 maths class, I can say with a high level of certainty that popularity does not automatically mean authority. I suspect though, that she was somewhat more authoritative after her second go at Std 8 maths.

Hunting about I found that there are already 2 sites that try to tackle it in this way: twitority and twithority. Since I can spout any nonsense on Twitter that springs to mind (and I often do), we need a system that assigns authority based on what is said, not by who says it. There are already 2 ways to monitor this: favorites and re-tweeting. Both of these provide a quantitative way of gauging the popularity of an individual statement without being based on the popularity of the person making the statement. Of course popularity will always come into it – the more followers you have, the more opportunity you have for being favorited and re-tweeted. But hopefully it would create a level of relevance that would otherwise be missing.

All discussions between technically minded people eventually end up at implementation, and I don’t believe it is possible to determine these kinds of results by relying on the available API alone. This smells like something that needs to mirror the public timeline, dynamically applying authority values to the tweet as it is indexed. Sounds like a job for Google really – they already have the algorithms for determining authority on web posts, plus they would be able to include both incoming and outgoing links from twitter into their existing algorithms. I’m pretty sure that once twitter search results become relevant, there’s a revenue model to be had around that as well.