The Anatomy of a Tweet: Metadata on Twitter

On ReadWriteWeb there’s an interesting map of the metadata that accompanies every one of your 140 character messages on Twitter, which I’ve reproduced below. (The map is the work of Raffi Krikorian. Click on the image to enlarge it.) There’s nothing terribly shocking here, perhaps: much of this metadata can be learned from visiting the Twitter page of the person sending the message. Even so, it’s sensible for those of us who are privacy conscious to be reminded from time to time that what seems to us to be a very minimal exposure to the unblinking glare of the internet carries with it a large amount of “body language,” so to speak.

Click on image to enlarge.

[Here’s the link to the same file as a PDF, so that you can enlarge it even more if you wish.]

These extras—this metadata—is of considerable interest to corporations that want to know which way the winds of commerce are blowing and how best to trim their sheets to capture them (and you) for profit. The ReadWriteWeb post talks about DataSift, a company that can consume the giant Twitter flow along with other gushings and extract information of an aggregate nature that might be useful to corporate clients. One imagines that governments, too, are mining this spate for purposes of their own.


  1. Wow! This is fairly scary stuff. Talk about too much information. Had no idea all this was available. Makes you want to think about commenting.

  2. What part of that is scary? Pretty much every single bit of that information is already publicly available via the Twitter website. The only bits I can see that might not be available through the web interface are your user ID# (which I’ve never seen, but which may be available somewhere) and your account creation date, plus some ancillary information (time zones and language). Things like location will only be included if you enable it on your account and on the app creating the tweet.

    Still, interesting to see how much data there is for app developers to play with. Surely someone has come up with some novel (and actually useful) uses for this information?