The Blog

Featuring fresh takes and real-time analysis from HuffPost's signature lineup of contributors

Justin Cash Headshot

My Trip Inside the Dada Engine: Where Computers and Art Collide

Posted: Updated:
Print Article

Back in the 1990s, blogger, photographer, music producer and all round tech wizard Andrew C. Bulhak created a revolutionary computer program known as the Dada Engine. By combining some elementary grammatical rules with randomly generated bundles of text, the Dada Engine can spew out what sound like entirely plausible sentences of prose at will, sentences that sound so plausible in fact that they can easily be confused for those written by the human hand.

The Dada Engine performs this remarkable little trick by using something known as recursive grammar, where linguistic elements are regurgitated several times over to form coherent sequences. Think of a computer's version of recursive grammar as a nonsense feedback loop, where the language being defined is free to be used in the definition itself to form an ultimately meaningless chain of gobbledygook. These recursive structures are incredibly easy for technology to replicate, requiring very little processing power (though plenty of a very human sort of human ingenuity).

With the right programming, we can even be tricked into thinking we are privy to the words of some post-modern genius and not those of the entirely lifeless computer that was actually responsible for their creation. Bulhak's original experiment has spawned a whole host of kooky follow ups which have put the invention to some humorous uses, from generating bogus exam questions to curveball political speeches and expressions of pre-programmed braggadocio - all grammatically sound with the appearance of realism despite their genuinely random nature.

An NYU professor by the name of Alan Sokal even put the principles of the Dada Engine to the test by submitting its spontaneous whirring to an academic journal. Beautifully, the resulting paper actually went to print, even after it what given the characteristically impenetrable title: "Transgressing the Boundaries: Towards a Transformative Hermeneutics of Quantum Gravity"

I played around with a more recent version of the Dada Engine to generate the following 'legal' ruling, seemingly with all the necessary levels of specificity and clarity of language:

"Each telecommunications carrier has the duty: to provide dialling parity to competing providers of telephone exchange service and telephone toll service, except as otherwise expressly provided."

Sounds plausible, right? In fact, sounds like its been ripped straight from the pages of some dire regulatory manual. Yet it was all generated at the hands of a computer program, and to that extent, is utterly devoid of any meaning.

What I find far neater about Dada Engine variants, however, is that they can produce some pretty epic prose, not just bureaucratic drivel. Again, I played around with a variant of the Engine for a while and came up with such gems of wisdom as: "Art is nothing more than noise" and "truth is the greatest lie".

Imagine a world where we could have our entire literary discourse run by computers pretending to engage in intellectual discussions. Just close your eyes for a second and dream of it. Whatever would we do with all those musty thinkers sitting in even mustier studies, who expend so much effort to sound articulate and elegant only to have their trade hijacked by machines?

Another question worth asking is whether we should endeavor to make our language more scientific so as to eliminate any chance that some pesky tech nerd playing with a laptop would be able to recreate its content. Does therein lie the distinction between the literary world of rhetoric, and the scientific world of non-randomness? Should technology advance to such a level that it could even replicate a rigorously scientific treatise, this would be a leap into the unknown indeed.

Yet I actually find it rather comforting that machines can churn out the same vacuous crap as some writers (mostly those of the fiction variety). Human after all, it seems. But at the end of the day, who cares if a machine can pass some kind of literary Turing test? Technology and aesthetics have never been at war with each other, nor will they ever be. Beautiful prose is beautiful prose, end of story. Jargon laden pseudo science is a criminal offense whether committed by man or machine. The aesthete in me can still appreciate the language for its artistry, whilst the nerd in me still admires the fact that the words themselves were nothing more than the product of a computer program.

Since we have developed this ability to imitate human literacy, computer simulations have spoken to the core of debates concerning the role of authorial intention in aesthetics. What unsettled many about the original Dada Engine was not that its output seemed to make sense, more that its seemingly artistic prose was almost entirely devoid of input from any author.

An old guard still holds that art must have a thinking artist, that literature must have a conscious writer. I disagree. In fact, in find such views a little pretentious.

Take a Jackson Pollock painting, for instance. For those of you unfamiliar with his work, to the untrained eye it can be indistinguishable from the work of a maniac who has thrown an artistic tantrum and decided to launch his paint cans randomly in the general direction of his canvass. A large section of the art literati, however, hold him in highest esteem. The uninitiated cannot possible understand the intentions behind his unique brand of paint chucking, they cry.

But what if some non-sentient computer, following a set of formal rules à la the Dada Engine, squirted color at sporadic intervals around your garage, in which was placed a giant canvass. As luck would have it, the resulting pattern just happened to be an exact replica of Pollock's latest masterpiece. Would you still call it art?

For what my opinion is worth, I would. Sometimes human talent, skill or endeavor is relevant to how much one can appreciate an artwork, literary or otherwise. Other times, it just isn't. In these cases, whether my book, painting, music or whatever was entirely computer generated makes precisely zero difference to the quality of my aesthetic experience. I'd probably have exactly the same admiration for Shakespeare's words had they not actually come from the Bard's mouth, and I'd definitely hold the same opinion of a Pollock painting if I was unaware it was actually a Pollock.

Machination has yet to truly create autonomous art, but I don't think it would be such a bad thing if it did. An increase in quantity does not in and of itself imply that the quality of our art would decline. Put enough monkeys on enough typewriters, it is said, and they would, on an infinite timescale, bash out a copy Hamlet purely by chance. In the same way, enough computers generating enough random sequences of words should, in theory at least, produce a copy of a similarly compelling and deep tale by the grace of fortune alone.

Would our world be poorer because of this output? Not a chance. I like to see the Dada Engine as a call to allow the computer to intrude further into the world of literature and art. Bulhak's computer generated words have fooled some of the best in the business, but I am more than happy to be so fooled if the aesthetic reward is great enough. As we continue to refine the technology, the author may not be as relevant as he used to be, but our search for artistic beauty in prose and print will undoubtedly continue.