Art for Computers

I was reading with interest the New York Times piece on The School for Poetic Computation (via an @aweissman tweet). The idea of the school is worth thinking about: the idea that art, culture, and work subvert each other. That they play off each other. That, indeed, using the tools of progress for play instead of for work is in itself an important component of progress. As Hank Tennekes observed in The Simple Science of Flight:

Among the redeeming qualities of our species is that we play. Indeed, we surround ourselves with toys, and we remain preoccupied with them throughout life… We display almost inconceivable creativity as we tinker with our playthings. The force of imagination and the passion for experimenting propel us toward outrageous designs and technological achievements.

This post isn’t about that. But it reminded me of something I wrote a year ago and didn’t publish. It’s about neo-Luddism, the idea that progress will not just put people out of jobs in the short-term, but that progress will permanently reduce the number of jobs that can exist. I didn’t publish it because any learned person who believes this has already demonstrated that they will not believe rational economic arguments, so spouting off even more rational economic arguments is preaching to the choir (I’m trying to do less of that in my life.) But book-ending the very long rational economic argument that I wrote (and ended up using elsewhere) is a different, humanistic one. It might sound a little strange, but it’s Summer, so maybe you’ll be a little patient with me. Here it is.


A question that’s been on my mind constantly over the last few weeks: if machines made art for other machines, what would that art be like? Let’s make it a bit easier to think about: if computers made art for other computers, what would that art be like?

No, seriously, I have thought about this every day for at least three weeks*.

Computers don’t have the same sense organs we do, they have different in-built biases towards the ‘beautiful’, they don’t have idealized physical forms, they don’t have the same perception of time and decay, they don’t have the same emotional reactions towards colors or combinations of colors, they don’t–in the sense we understand them–have emotions at all.

Would there be an analogue of Hopper, creating moody images of low-traffic URIs whose structure was scanned but not parsed? Or of Mondrian, Fourier-transforming novel time-domain ping patterns? Perhaps of Wyeth, with an idealized but enigmatic snapshot of the L4 cache. Of Duchamp, with partially overlaid, closely-spaced bubble-sort heap dumps. An existentially minimalized kernel panic. An idealized object repeated in a random but predictable way. A real, and carefully chosen for its generalizable realism, record of a SYN flood DoS attack. A program which seems as if it should, but in reality would never, work. An extremely large, but carefully counted, number of zeros in a row.

This is the kind of thing people don’t think about. Art for computers benefits only computers and nobody cares what benefits computers. That is not what machines are to us. We build machines to do things for us. We use them as an extension of ourselves, to achieve what we want but faster or better, more efficiently. Machines make us, as a species, more productive. They help us make more of the things we want, so we can have more of them or more people can have them. Machines are not ends, they are tools.

[Insert long, pointlessly duplicative of centuries of economic thought rational argument here.]

Neo-Luddites believe that machines exist for themselves. They confuse subject and object. They believe that there is some extrinsic mechanism that creates jobs and we are simply the dopey beneficiaries. The political dialog heightens this, making it sound like our government can create jobs and give them to us, like birds feeding their helpless young. That, in fact, that’s where the jobs all come from, as if the government were the stork and jobs were babies. The Neo-Luddites believe that if you empty the tub of water faster than the faucet fills it, then you only have half a bath, for the rest of time.

But we are not the bather. There is no bather. We are the water. And no matter how you move around the water in the tub, it remains full.

People lose their jobs because of improvements in the means of production. So, in that sense, progress destroys jobs. The prevailing wisdom has it that progress also creates jobs, and that these forces have balanced each other out in the long-run. Neo-Luddites say this is no longer true and most economists seem to argue with them on their terms. But arguing on these terms is ridiculous: there is no aloof god Progress. People create tools to make jobs more efficient, and then fewer people are needed to do them. And the people who might previously have done those jobs find other things to do, and those ‘other things to do’ is the job creation we also call progress. Progress is something we make, not something made for us. Economics has generally become more human-centered over the past few decades, but this particular subsector of growth theory has not.

The reductio ad absurdum of the Neo-Luddite argument is a singularity where machines end up doing all the jobs, leaving humans doing nothing. This is absurd in two ways: (1) if machines were not simply tools, they would not care to do the things we direct them to do, they would do the things they care about themselves (make art, perhaps); and (2) if they did continue to serve us, then the things they provide would be effectively free (the machines would not care to be ‘paid’ in any coin we have) and we humans would be free to create value in other ways without worrying that our needs would not be covered–what else, indeed, would the machines do with all the food and clothing they produced except give it to us?

At any point on the spectrum between now and this singularity, we can make a similar argument. Between when 80% of Americans were farmers and now, when some 3% are, we did not structurally lose 77% of all possible jobs. When each farmer could produce food for many more people than they used to, they did not hoard that food and let the non-farmers starve. They found some other good that they wanted that the non-farmers could produce, and traded food for it**. The non-farmers found ways to create things the farmers wanted but that they previously could not have because there was no one to produce it. And, indeed, what else would the farmers do with all that extra food except find things to trade it for?

To believe that there are no more jobs to be created you have to believe one of two things: that there is nothing else we as a species could use, that we already have everything; or that people are simply too unimaginative to create these things. If you believe the first then you should take a look at our healthcare system or our educational systems, among other places that have a crying need for more attention. If it’s the latter, then you need to have more faith in us humans.

Is the question of what art machines would care about interesting? Only in that it forces us to think about why we as people find art interesting. The machines themselves are not interesting, except in how they mediate interactions between people. The only thing that is really interesting in the end is that interaction between people.

* No, I’m not obsessive. Shut up.
** Through the intermediary of money, yes, but don’t let that obfuscate things.