Once upon a time, I posted a video tribute to both Tetris and the turbulent history of the USSR from an indie neo-folk band from England. It featured a supposedly prototypical Russian worker singing through the downs and a few ups of Soviet history while doing Tetris-inspired versions of his work, referring to invisible bureaus, cabinets, and secretaries telling him to build walls from shoddy or top secret blocks, a typical fact of life in an intensely supervised command economy where every resource in the nation was to be used to fulfill a grand, all-inclusive, five year plan. It didn’t work of course, but somehow, the notion of a command economy ruled by precise computations of economic inputs, outputs, and relationships lived on, and even ended up as a rather odd novel called Red Plenty. On the conspiracy circuit, the Zeitgeist Movement and it’s Venus Project offshoot advocate a sort of neo-Marxist technocracy in which computers would control economies rather than humans to ensure a fair distribution of resources and labor. I’ve tried to explain why this is a really, really bad idea in a few skeptical forums, but luckily for me, a mathematician at the University of Michigan tackled the bulk of it in a post that won a 3 Quarks Prize for science writing, showing that such computations just won’t work.
Now, he built his argument around the idea of a perfectly computed communist economy of Red Plenty rather than the hypothetical ZM/TVP technocracy but his conclusion still stands. Modern economies simply have too many variables and are too connected to be virtually manipulated by even the mightiest supercomputers. His logic is pretty airtight, showing why parallelizing this problem is simply unworkable and why the notion of how to optimize an entire economy is inherently flawed because ultimately it relies on someone deciding what will constitute optimal performance. So what would happen if you were to try and create virtual models of an entire economy in motion to serve as your guide? The program would take far too long to run to be viable and what it decides is optimal, if it’ll ever finish running to come up with any oracle-like response, the output would be an enormous spreadsheet of hunches based on what it was told should be considered optimal. May as well just declare what you think is right and cut out all the unnecessary middle-bots that will take thousands of years to run a single simulation. But how do you objectively decide what is and isn’t optimal and why?
While the analysis doesn’t focus strictly on math and veers off into the politics of fairness and economics in a free or command market, it leaves off another crucial problem with a computer-commanded economy. There will always be a black market ran by unpredictable humans and without global wars, economic technocracies will always have to trade with human-ran markets. Suddenly, all the intricate computer models fail because in their assumptions, they’re dealing with rational, predictable actors, even if their behaviors are very complex. As we saw with the Great Recession’s opening act, the Subprime Meltdown of 2008, assuming rationality isn’t a good way to make money. All the formulas that told Wall Street it could keep the gravy train going ignored that brokers and debtors lied, attempted desperate schemes to stave off incoming financial body blows, or had to deal with those who lied or attempted desperate schemes to stave off incoming financial body blows. Dealing with fraud wasn’t part of the equation. Dealing with human panic wasn’t a part of the equation. Predicting what action each decision-maker involved in the implosion on Wall Street and its subsequent bailouts would make would’ve resulted in so many permutations, you may as well have resorted to tea leaves and coffee residue to plan the markets’ reactions. And all that still excludes the other actors involved in the global economy.
Finally, let’s circle back to the recurring issue of who’ll decide what is and isn’t optimal in the cyber-command economy as per the ZM/TVP wish list. Even in a nation state of equals, whoever finds him or herself in charge of the central planning computer can be, as the Soviets used to say, more equal than others. Why? Because humans want things and if they have an opportunity to enrich and empower themselves with a few keystrokes it would be the height of naiveté to assume that they wouldn’t. Ultimately as long as there’s a hierarchy of any sort involved in any social structure, there will be authority figures who have to somehow be kept in check. To trust them with a machine assumed to be an economic soothsayer would invite an abuse of power, and with debates about what the digital economic oracle in question should be optimizing and how, we shouldn’t even bother, especially when we consider that its outputs would be biased, unwieldy, and impractical. Deciding the best way to run an economy isn’t a simple matter and economics offers no clear cut or easy answers. And as tempting as it may sound to simply outsource the problem to a machine, it won’t work. When we don’t know a good way to solve a large and complex problem, even our fastest and more powerful computers will produce little more than a flood of noise in response, much like our minds do when faced with the issue…