Wordle visualization of my Tweetstream

Mike Hendrickson (@mikehatora) sent me a nice Wordle visualization of my tweet stream, dating back as far as Twitter keeps it. As you can see, I retweet a lot. It’s also interesting that many of the repeated words are not concepts or topics, but people’s names (in the form of twitter handles.) This is one of the interesting things about twitter: it’s a reflection of a community of shared minds, rather than of shared ideas.


(Tweets were retrieved using this python script.)

  • What I see is a jumble of ideas/memes and a social graph of names. What I think would be interesting to see is a flow of these words through the twitterverse, using the time dimension.

    Twitter might be a very interesting platform to study the real world (vs computer model) dynamics of the spread of ideas. Is it through the social graph, or is the spread by different means? How dense do the connections need to be to ensure transmission?

  • Also interesting: the prominence of “timoreilly”, which I assume comes from re-retweeting, or retweeting comments about your tweetstream and other activities.

  • Jack – that’s right. I took a look with twitter search to see what was there, and it is indeed me retweeting people tweeting stuff about me to bring it to my attention.

    This is very useful for me, btw, since my tweet stream is a kind of memory archive, which helps me to keep track of links. It’s amazing how often I learn about a video or interview with me because someone else tweets about it.

  • Just for fun, I took the code Tim pointed to and then filtered out Twitter users and got this Wordle.


    I wanted to do the same for the pages linked in there, but that will take a little more time…

  • Tim: “my tweet stream is a kind of memory archive, which helps me to keep track of links”

    A distributed mind.

  • Neat idea. I was interested enough to re-write the scraping code to use the Twitter API calls directly and also respect the Twitter rate limiting. It is downloadable, from here Also supports fetching multiple users and easy to add in additional filtering to extract links, RT’s only etc.

  • Gordon thanks for the rewrite and making your code available. Another scrape of twitter, written by our Tim Allwine, for #followfriday tags shows how much tweeting of peoples names is indeed happening. The Wordles here: Referers and Referred To shows for the past 30 days [March 30 – April 30 2009]. Perl code here.

  • Josep Yao

    if you are interested in these forms of twitter research, i suggest checking out this link. http://www.dosenation.com/listing.php?id=6114

    Twitter dreams really.

  • Josep – that is very cool. Retweeting it!

  • Tweaking a bit more with the API, I came up with this indication of what Twitter is afraid of, today, May 1st

    Fear or Afraid wordle Probably not that big a surprise. Wonder what it’ll be next week.

  • I just blogged about Twitter, distributed brains and minds here:



    well i would like to say that i love this wordle thing we talk about it is really fun i like making them then and putting them i the front of my binders for school it is a easy way to make COOL AND SIMPLE book covers.


  • Jam

    That is quite interesting, Check out this wordle projcet it gives a new perspective of looking at the inuagural addresses. Its much easier to digest than reading them all, however looking at wordle may just intrigue one to read the whole address. You may even be suprised by some of them. There is a stark contrast between lincolns first and second.