it's national novel writing month again. you may recall last year i entered the fray, with the intent of composing an entire novel from strings of text found on google. i kept running into problems and after exhausting my daily limit of searches on the google API, i had to wait until the next day to start work again. as i tend to hit my unpaid web development in binges, this didn't produce a novel within the month.

a couple weeks ago, tom coates posted five years worth of his writing on in the hopes that someone would create visualizations of it, and so it was that i gained the source material for my second attempt at a computer-generated novel. what i have so far is "virtual tom coates," though currently it would probably be better titled "virtual drunk tom coates" as it tends to speak in incomplete and/or incoherent sentences -- though i must admit i have never even met tom coates, much less experienced him inebriated. every two adjacent words it spits out are found together somewhere on plasticbag, but that's the current extent of the similarities between tom coates and virtual tom coates. here's an example of the output, which is different every time the page is loaded:

Of work on is finally my hair cut suits my stride the short-lived before the best project I'm going to

there's obviously a lot of room for improvement. since i have no significant limits on my work, i will only have myself to blame for any failure to produce a 50,000 word novel based loosely on the last five years of writing by tom coates. and if all goes well, i should have a system i can then apply to any text source to produce similar text.