Trish, Reiner and the Politics of Open Data

Ongoing debates over open data remind me of Cory Doctorow’s short story, “Human Readable”:http://craphound.com/?p=3275, which depicts attitudes to technology through a disagreement between lovers. Reiner is a classic hacker – he thinks of the world in terms of technological fixes for technological problems, and has difficulty in believing that the algorithms can be systematically skewed. Trish is a classic lefty, who thinks of the world in terms of power relations, and, specifically, in terms of how smart powerful people figure out ways to gimmick the system so that it works to their advantage. They don’t understand each other very well but end up, sort of, figuring out a way to cooperate. Clearly, real life people have more complicated views than Doctorow’s characters – even so, he’s put his finger on an important tension. When I went to Foo Camp a few years ago, I found it incredibly intellectually exhilarating – meeting a bunch of people who were both (a) smarter than me, and (b) intensely practical, interested in figuring out how to do stuff rather than study it. But I was also a bit nonplussed by how enthusiastically many of them believed that the Obama administration was going to usher in a new era of Big Data led technocracy. A lot of them (not all of them) didn’t seem to have any very good idea of how politics actually worked. They were mostly Reiner, without much admixture of Trish.

Of course, there are plenty of Trishes too. Debates over open data, like many other debates over technology and politics, have calcified around a Reiner-Trish confrontation – techno-utopian naivete versus politics-led skepticism. This is _not_ to say that the people on the one side of the argument are Reiners, and the others are Trishes. The smarter people on both sides of this argument have more interesting and complicated understandings. But it is to say that the Reiner-pole and the Trish-pole are the attractors – it’s harder than it should be for people to escape the gravitational pull of the one or the other position for very long. Furthermore, even when Reiners try to become Trishes, they can retain a rump utopianism. Larry Lessig, driven into shrill, unholy Trishdom by the horrors of the American political system, still quietly yearns for technocracy. His book on American corruption is an excellent and compelling indictment of our current system, but he tends to conflate political corruption with partisanship. Here, he is plausibly influenced not only by his earlier work on technology, but by a broader tradition of American progressivism, which is (as Nancy Rosenblum has documented) inherently suspicious of partisan contention. Trishes for their part tend to blend a specific, and well justified skepticism about the political uses to which technology can be put with a more general disinclination to believe in the social benefits of technology, data. Actually existing Reinerism blends (a) a belief in the social benefits of technology, with (b) a technocratic understanding of politics. Actually existing Trishism blends (a) the belief that the social benefits of technology are limited, with (b) a power-based understanding of politics. Again – Reinerism and Trishism by no means necessarily describe the _actual beliefs_ of people on the one or the other side in these arguments. Instead, they describe the _attractors structuring debate._

This annoys me, not least because it’s highly inconvenient for my own politics. Like Reiner, I’m broadly optimistic about the social benefits of technology. But like Trish, I’m not a technocrat, and believe that politics is necessarily a struggle between factions with different wants and interests. More precisely – I argue that information technology can provide extraordinary benefits – but only under the right political configuration. Cosma Shalizi and I have been trying to articulate the case for what we call “cognitive democracy”:https://crookedtimber.org/2012/05/23/cognitive-democracy/ – political arrangements which recognize the irreducible diversity of individuals’ interests and perspectives, and try to take advantage of them. Briefly, we argue that democracy, to work well, has to (a) minimize power disparities and (b)retain and harness cognitive diversity. Technocracy will not do this particularly well – what one wants is not bland consensus, but vigorous political contention, in which different factions struggle and engage with each other, likely never reaching agreement, but learning from each other, and setting out clear, alternative approaches to dealing with collective issues.

Here, open data can have three crucial benefits. First – it helps limit power disparities. Lobbyists’ main advantage is often less their selective control of funding than their selective control of information. Making politically-relevant data available can (with caveats: see below) make it easier and cheaper for interests that are under-represented to make better and more compelling arguments for their perspective. Second, open data, where it is high quality, can help limit the tendency of factions to make up their own information as well as their own interpretations of that information, hence improving democratic politics (diversity of opinion and understanding is not cognitively helpful when it is unmoored from reality). To be clear – neither of these is a cure-all. Even under the most optimistic assumptions, open data cannot correct for various economic, organizational and political disparities of power. As debates over global warming demonstrate, _no_ data will be sufficient to convince a group that has dived into the deep end of crazy. Yet open data can help (and would help even more, if it were combined with structures such as, in the US case, a reborn “Office of Technology Assessment”:http://en.wikipedia.org/wiki/Office_of_Technology_Assessment).

Third, opening up specific _kinds_ of data – data about what Cosma calls processes of “collective cognition”:http://cscs.umich.edu/~crshalizi/notabene/collective-cognition.html could help foster a general democratic experimentalism. Cosma and I don’t think of the Internet as either a force for general emancipation or another instance of commercial and political cooptation. Instead, we think of it as something like an abandoned mad scientist’s laboratory, in which various experiments in cognitive processing have been left to fizz and overflow together. Some of these experiments are turning into monsters, others unviable chimeras, others yet interesting hybrids. Figuring out why different experiments had different outcomes is going to be a chancy process – but nonetheless can provide highly valuable data on when processes of collective information processing and decision making work, and when they don’t. The problem is, of course, that most of the really interesting data is not readily available for commercial and political reasons.

If this understanding of politics is right, open data has great promise. However, it also has clear limitations. First and most obviously, it’s going to work best under different democratic arrangements than those we have in the US and other democracies, which are becoming increasingly sclerotic thanks to inequalities of power. Open data doesn’t create its own politics – instead it requires sweeping politic reforms, which it can’t plausibly generate itself, in order to achieve its full benefits. Second, in the actually-existing-democracies of today, the ‘if you build it, they will come’ attitude of some of the breezier open data proponents is badly misplaced. Open data is likely only to be taken up to the extent that it is _directly useful_ to the agendas of _existing movements, factions and organizations._ Politics is about struggle between factions. This isn’t likely to change, and arguably it _shouldn’t_ change. Hence, data will be politically relevant only when it’s relevant to the political goals of particular interests, or, perhaps, to newspapers if there’s a particularly juicy scandal.

Of course, this view of politics may be profoundly or subtly mistaken (it’s a work in progress – we make no guarantees). But at the least, what it does is to open up the debate a little. If we try not to conflate our views of technology with our views of politics, in the ways that Reiner and Trish do, we may be able to think more clearly about how they relate to each other. Specifically, we should be able to think better about how different forms of politics interact with different regimes of data access. There are a whole variety of possibilities, which the current set-piece battle blinds us to. It would be nice to move on from it.