Saturday, February 4

the incipient model

I wrote last month about being embarrassed by all the stupid shit I’ve written:

I’m not going to delete all this stuff, though. I’m not sure if I can quite say why. I don’t blame anyone else for that impulse, but it’s not for me right now. Maybe I’m just too obsessed with memory to deliberately efface one of the few artifacts I’ve made out of it. Or too conscious of how little else I’ve built than a pile of words, as I near an age when it’s just as likely the bulk of my accomplishments are in the past.

But having written that, I keep thinking about the value of putting anything on the public internet at all.

This is rooted in a bunch of half-formed thoughts: I used to think the internet was in some deep sense a benign phenomenon, with some kind of root ethos or principle that tended to erode power and build a more humane community. I’m now pretty sure I was wrong.

The first thing is that networked computation is an epidemic.

Cellphones have already rewritten civilization; beyond that, the epidemic is expanding into every remaining domain where it can be supported by market speculation, narrow technological advantage, fast prototyping, spying on users, and/or generalized b-school hype-machine idiocy. Processing power is cheap, wireless networks begin to approach ubiquity, software is easier than hardware, business is full of gullible power-mongering assholes, and data collection is like catnip to every system of control on the planet. That people and industries will keep putting computers in shit is hilariously overdetermined, and there is nothing short of an apocalyptic calamity that can be done to stop it.

Network-scale capital has built us a baby panopticon, and it’s maturing fast. So is the technology to extract meaningful structure from all of the data it eats.

We already know, for example, that photos exposed to the internet are scanned for recognizable objects (like faces) and correlated. Google does this for image search. Social networks do this to associate media with individual users. Law enforcement agencies are applying similar tech to an ever-growing firehose of surveillance data. Facial recognition and license-plate tracking have only just begun having consequences, and they’re the tip of an iceberg. Given enough storage and compute, the facts and associations latent in things we think of as opaque blobs of content are about to become visible to everybody with a big enough database. It’s all metadata, given enough of the stuff.

Of course there’s a vast amount of opportunistic bullshit being spun about machine learning and AI right now, but all of this is really just an emergent effect of a less shiny-marketable thing: Databases and networks have consequences. Kinds of data that our culture still think of as requiring human interpretation are rapidly becoming easily-queried tags and indexes.

So the network at large contains this cumulative model of me. It’s not all contiguous and cross-referenced yet. Much of it is hidden from direct view, it’s lossy in a bunch of particulars, and it’s still fractured in part by legal apparatus and social norms. But it’s a big thing and it’s not going away. It’s going to T-1000 itself together from all those pieces just about no matter what. There are very few remaining technical impediments, and no political institutions seem willing (or able) to create effective legal ones.

So I wonder: Should I be feeding this model my writing and pictures? Should I volunteer, for that matter, any kind of self-surveillance?

The second thing is that we’re starting to have good empirical evidence about the kinds of social systems the internet generates. Plenty of them are good, or at least fairly benign. Others of them are, for example, a resurgent fascism. Hateful shitheads turn out to do just fine in the network environment. Reactionary backlash and the inelastic properties of damaged cultures propagate as well in new media as anything does. Sometimes better.

There’s something here about systems. I was wrong about how this system operates and what’s possible within it. I think I’ve been naive about systems in general.

So I wonder: What part of the self-surveillance I volunteer isn’t an attack surface?

Anyway: I want to put writing and pictures out here. But there’s this competing impulse to hide all the stuff as far as possible and take down all of the public accounts. Go dark and operate only in the quiet corners of things, keep longform writing and pictures for letters to family and a handful of friends.