I was recently alerted by Fred­erik Berlaen – our res­i­dent type designer and script­ing cham­pion – of Type­tweets, a neat web­site by Kyle Meyer, a user expe­ri­ence designer for Minneapolis-​based Clock­work Active Media Systems.

The con­cept for Type­tweets is decep­tively simple – it basi­cally is an up to the minute look at the typo­graphic dis­cus­sions hap­pen­ing across Twit­ter. The web­site dis­plays the last 100 tweets from Twit­ter that men­tioned a typo­graphic related term. These tweets range from humor­ous, to use­less, to pro­vid­ing hours of fun with links to other sites. It’s really up to the com­mu­nity at large to deter­mine the qual­ity of its con­tents at any given time.

So how does Type­Tweets work? It scours the Twit­ter feeds look­ing for spe­cific key­words. As Twit­ter seems to have some limit to their number, right now there are eleven of them. How­ever Kyle is work­ing on cre­at­ing a way to include more terms in the near future. Ini­tially the terms were:

• Typog­ra­phy
• Type­face
• Serif
• Typo­graphic
• Hel­vetica
• Ver­dana
• Amper­sand
• Let­ter­press
• Caslon
• Palatino
• Futura

Obvi­ously this is hardly an exten­sive list, and Kyle is happy to wel­come sug­ges­tions. But there is a caveat that com­pli­cates mat­ters. If the term has any other uses Type­tweets ends up pulling in tweets that don’t have any­thing to do with typog­ra­phy. For exam­ple, there is appar­ently a hotel or some­thing named Claren­don that people like to dis­cuss, a lot. This is also the reason why Kyle is hes­i­tant to include “font” as it is used so fre­quently, and by so many people, that it’s a bit of a double-​edged sword. But that won’t stop him from turn­ing it on and off inter­mit­tently at times.

What exactly hap­pens under the hood is a rel­a­tively simple thing in theory, but play­ing nice with Twitter’s servers com­pli­cates things a little. Essen­tially, as Kyle explains, this is what happens:

1. A cron job runs a PHP script every five minutes.
2. The PHP script sends a query out to Twit­ter, and gets the tweets back as data.
3. The data gets parsed through a decent bit of Regex to make it more useful.
4. The result is stored as a file on Kyle’s server, so that when people view Type­tweets, he’s not using any of Twitter’s servers to supply the data.
5. A bit of Javascript parses the data when you load the page and you enjoy reading.

Although there’s no time­line for updates to Type­tweets, Kyle already has a few upgrades fig­ured out and just needs to find time to add them. Type­tweets still is an alpha ver­sion, a proof of con­cept, so there might be quite a few bugs left in it which may account for the occa­sional rogue non-​typographic tweet, dis­play bug, etc. But I share Kyle’s opin­ion: “It’s still good fun to play with in my humble opin­ion, even in its rough form.”