Markov Chains, Horse e-Books and Margins

In discussing trajectories, elements of engineered serendipity, “thought vectors in concept space” with Gardner and Jon yesterday the following occurred.

Gardner shared this video (which is well worth watching and I rarely have the patience for videos).

That led to a discussion about creating and using a MOOC/hashtag specific Twitterbot (like horse e-books but real1) using Markov Chains2 to create algorithmically driven conversations/connections that occur in the margins of intention and result.3

So I began messing with the idea last night. Given I have a completely illusionary knowledge of programming I looked for people to tell me how to do this. I found the metaphor a minute tutorial which will help me out with the Twitterbot end of things in the near future. I also found this PHP based Markov generator which does very nearly what I want absent the Twitter-ing part.

I did want to automate the connection to a particular Twitter hashtag rather than adding the content manually so I started wandering around looking for ways to do that. Step one was trying to use curl. I eventually semi-melded some curl examples with the Markov generator. I was using the Twitter search for #ds106 as the source initially. With curl you are pulling the html so I got some interesting pieces but a fair amount of code fragments as well. Stuff like . . .

I liked the code to some degree but figure a larger audience would probably ignore it. So I harassed Alan, Jim and Martin early this morning and got access to the #ds106 Twitter spreadsheet archive. I pulled it down as a txt file and used it for the source material. That started to get cleaner results like . . .

You can mess around with the semi-working (just refresh page and hit resubmit form- I did say semi-working) manual/random #ds106 tweet generator over here.


1 Really fake, I mean. I guess.

2 I’m not really sure if that should be pluralized or not.

3 There’s a whole additional piece where you think about larger scale curricular design which incorporates random elements and assignments that use algorithms to push people in new directions. That starts to get really interesting. I am considering how the assignment and maybe a browser plugin could create contextual variables based on what site you were on at the moment that would then be incorporated into the larger assignment- kind of a #ds106 remix on contextual steroids.

2 thoughts on “Markov Chains, Horse e-Books and Margins

Comments are closed.