Thursday, August 6, 2015

Episode 5 is up, so here's a blog about robocorps, pitchforks, hairy monosyllabists, decision rules, and the tenth heuristic (with a footnote about pistachios)


Explaining first things first: I'm blogging on Tuesdays and Fridays, lately, because my serial novelet, "Silence Like Diamonds," is running at Light Reading, a tech-news oriented website covering advanced communications systems. You can browse back through this blog or just hop over there and start reading, but in any case, Episode 5 (out of 10) is now up and running, or will be by the time you read this, so if you haven't started yet,  you have a bit of catching up to do.
One idea that seems to bother people just a little is that my fictional NameItCorp, in about 2030, is actually a giant pay-to-access app, a corporation with no mission (other than "doing things for money") and no workers.  It does pretty much anything you can do on the net: answering questions, locating information, putting together bids of subcontractors to accomplish purposes, and so forth, all via robots, AI, and subcontractors. And in my imagined 2030, it is a big-as-Google-today-and-just-as-indispensable corporation.
Now, most people don't have that much problem with the idea on a small scale; after all, what is a sales or catalog website, especially one selling pure information products, if not a robobusiness? I've known a couple of people who owned a few dozen vending machines and serviced them all themselves; they're in the retail business without any employees, but most people would say they just have a way to sell things that doesn't require them to be there.
Information products are a natural for robobusiness. Already much of sports reporting and financial news is being written by robots, who take the box scores or the price data and convert it to a news story. Inference algorithms following web hits could easily figure out which content was most popular (and which was best at drawing people to advertising or to other news stories). In fact, they already do, with human beings then deciding overall policy on what to do about what the inference algorithms report.  Well, algorithms can also make decisions; aren't the humans only in the loop because the tech isn't quite right yet, or perhaps because the boss's nephew needs a job? Is there any reason why a web-based news publication would need human employees in another few years?
The often mentioned "analytics revolution" and the rise of "big data" are yet another example of what's already possible and will be much more significant soon: more and more business decisions are being made by processing the corporation's enormous store of data through analytic algorithms and simply picking the result with the highest expected value or minimax loss.* At that point, do you really need a management team?
Furthermore, it can fairly be said that although it's not legally required by fiduciary duty (despite myth), many corporations have been moving, as Harold Meyerson puts it very clearly, from a model of duty to all the stakeholders to a narrow duty to only the shareholders. There's a kind of political basis for this to be found in some capitalists and some extremist libertarians, a position that can fairly be called agorolatry**, i.e. the belief that markets are not just efficient or personal-choice-maximizing or fundamentally fair, but actually the source rather than the expression of moral and ethical judgment. I've noticed that some younger generation capitalists are beginning to talk more like agorolaters than their older brethren; this  might end about as the dance around the Golden Calf usually does, or then again it may herald a rewarding century or two for Golden Calf Corp.
Put all those trends together and you've got Silence Like Diamonds's NameItCorp: the company that does anything for enough money, just as your app does whatever it's supposed to when you push its button.
If it makes you a bit nervous, well, it's an idea that isn't far from Karl Marx's idea of capital as "dead labor" -- the residue of work done by somebody who is now completely cut off from it. For example, you own some product of  work, say, a shovel you bought from Ogg, the shovelsmith in the next hut, who has long since moved on to making other shovels.  You loan the shovel to Gronk, in exchange for a tenth of his potato crop; you trade the potatoes for more shovels, which you loan to more hairy people with monosyllabic names, gaining more potatoes, and eventually you are the Potato and Shovel Baron of the Hypotheticalithic Age.
But what if the shovel kept getting smarter? At first it just made it easier for Gronk to dig. But then it started identifying the best potatoes to dig ... and marking out the garden beds where Gronk should dig ... and digging in places that Gronk couldn't reach ... and one day, the shovel just grew potatoes all by itself, and brought them to you.  Sorry, Gronk. You're out of the potato business. But if you need any potatoes, I have them on sale, more of them and cheaper than you could ever produce.
Meanwhile, Ogg's shovelmaking tools have also gotten smarter. Nowadays the shovels just get up from Ogg's place, walk over to yours, and get to work. You, of course, faithfully pay Ogg in less and less valuable potatoes, and the day comes when you've got all the shovels you need, and Ogg sits around watching his last generation of shovels dig because that's all there is to do.
It's a great deal for you, but you have to wonder if Ogg and Gronk are talking to the pitchforks now streaming out of the house of the village's pitchforksmith.  
Or maybe the unease that the coming of the robocorps gives me is a more personal than social matter, as expressed by Emerson:
The horseman serves the horse,
The neat-herd serves the neat,
The merchant serves the purse,
The eater serves his meat;
'Tis the day of the chattel,
Web to weave, and corn to grind,
Things are in the saddle,
And ride mankind.
That might make one more heuristic, in addition to the nine I recently listed: if something makes you uneasy today, and you don't see anything stopping it, tomorrow it will be  genuinely scary -- and most people will think it's perfectly normal.
§
*Good-enough-for-here definitions: These are both decision rules. Expected value=sum of the products of the values of the possible outcomes and their probabilities. Minimax loss: theory of games based choice in which you make the choice where the most you can lose is the smallest. The expected value of a coin flip where you get paid a dime for heads and pay a nickel for tails is 2.5¢; in 1000 coin flips you'd make $25. Over the long run people get rich by maximizing expected value, but it will sometimes pick a very risky choice if the reward is high enough.  Minimax loss says that if you have three choices, A, B, and C, and the worst thing that could happen under A is losing $10, the worst under B is losing $8, and the worst under C is losing $6, you pick C, even if A or B has a higher expected value or a higher maximum payoff.  Minimax loss is the rule people follow when they "settle" in marriage, but also when they refuse to accept a ride home with a drunk.  Like expected value, it's a way to play, not necessarily the only or the right way all the time.
**You're an economist if you think that the way to handle high pistachio prices is to let them rise until more people plant pistachio trees, and that this will allocate existing pistachios to those who want them most and can afford to subsidize future plantings, and therefore a free market in pistachios is likely to be effective and convenient. You're a libertarian if you think your freedom to buy and sell pistachios is the real issue, and that the allocation that produces is just, and therefore a free market in pistachios is part of a free society. And you're an agorolater if you think high pistachio prices are the just reward for virtuous pistachio growers and the well-deserved punishment for undeserving impoverished pistachio lovers, who should have had the foresight to be able to afford pistachios, and that if there isn't a free market in pistachios, evil wins.