There’s another path here that you haven’t mentioned: take advantage of content generation technology and combine it with your own efforts. (This isn’t a new idea: a few months ago, at least two or three ‘creative autosuggestion’ projects were floating around; Burroughs edited the output of cutups, as did Bowie; after Deep Blue became the world’s best chess player, top-tier players started playing ‘centaur’ or ‘cyborg’ chess wherein they collaborated with chess programs.)

That said, being a part of the generative writing community, I think you slightly overstate the state of the art. Generative poetry is pretty good, in part because poetry is pretty flexible and unusual stylistic choices are typically seen as meaningful: generative writing does well in forms that are either extremely experimental or extremely formulaic. Machine-generated short articles performing numbers-driven reporting of factual events (such as sport or financial stories) are more or less indistinguishable from human-written stories of the same type, but also represent a fairly uninteresting application of this technology (in part because, while the AP and other news wires started using this technology only recently, it’s been possible for even a mediocre programmer to implement this kind of thing since the 50s). Long-form narrative-driven works don’t tend to be readable, although in the past few years there’s been a lot of progress (for instance, there was an entry in National Novel Generation Month in 2015 called MARYSUE that generated pretty convincing pastiches of bad Star Trek fanfiction, most of which is not much worse than human-written bad Star Trek fanfiction).

There are certain things that generative writing systems are very good at. These systems are capable of being intensely creative (in the sense that they can generate juxtapositions of ideas that could never occur to a human being), but lack any good approximation of human taste: as a result, writing systems can be used as a creativity prosthesis, where human judgement isolates good ideas from bad ideas while the machine performs the heavy work of producing ideas. Likewise, these systems are good at being exhaustive: it’s trivial to take a couple corpora and a format and produce every possible variation on a theme, if someone is willing to wade through the output looking for the interesting material. Human editing makes machine-generated writing much more feasible for even long-form work.

Written by

Resident hypertext crank. Author of Big and Small Computing: Trajectories for the Future of Software.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store