When the Soviet Union was created, beginning in 1917, its rulers wanted all profits, all returns to capital, to accrue to the state. This would not have been possible with decentralized markets, since the state could not know the profits made by individual entities, regardless of whether they were privately or publicly owned. This, they anticipated, would lead to underreporting of income and defeat the rulers’ objective. To circumvent this problem, they turned to scientific planning to replace the market with algorithms.
Production units were asked to declare their plans and capacities. Retail chains were asked to project the demand for consumer goods. Government ministries were asked to stipulate their needs for investment in various kinds of infrastructure. These numbers were then put together in matrices, and a mathematical method known as linear programming was used to optimize the allocation of workers and productive capacity so as to achieve the objective of the plan at the lowest resource cost.
While this system could more or less handle large planned investments in heavy industry — steel mills, coal mines, and the like — it was woefully insufficient at handling consumers’ demand for varied goods and services once the Soviet economy started to develop. The problem was that the rulers at the time had only simple mathematical algorithms and no computers. Running an entire complex economy via algorithm was simply not feasible. This probably explains why after several Eastern European countries fell into the orbit of the Soviet Union in the late 1940s, some of them did not even attempt to adopt its full-fledged central planning system and instead allowed a modicum of market exchange to remain. By the time the Soviet Union was disbanded in the early 1990s, its central planning system of exchange was judged a failure, and it was rapidly replaced with a decentralized market system.
Spin the clock forward another 30 years, and we are now in a world where a large and increasing fraction of economic exchange is run, quite successfully, by algorithms. Amazon and Alibaba perform essentially the same tasks as the Soviet central planners: They collect information on items available for sale and, by doing so repeatedly over short time intervals, effectively track productive capacity. They collect day-to-day information on consumers’ requests for various goods and services, thereby tracking the evolution of demand over time. They match supply and demand in real time. And they organize the dispatch and delivery of these items to consumers across the globe. If the Soviet Union had had such tools at its disposal, it would no doubt have used it to run its economic exchange system. By observing all transactions and all payments, the rulers would also have observed all profits and incomes and thus been able to appropriate whatever they thought was due to the state.
As a matter of fact, the United States and other Western countries have begun to use sophisticated algorithms for central planning purposes, with a view toward appropriating economic surplus. The best example is the auction sale of bands of the airwaves for mobile-phone and other communications services. Companies interested in buying communications licenses are asked to place bids on the airwave bands they want, and a computer allocates the licenses so as to both minimize the chances of airwave interference and maximize the government’s income from the licenses. Similar auctions have been used to allocate foreign exchange by central banks and the procurement of goods and services by government agencies. Matching algorithms are used to assign medical interns to hospitals, students to schools, and workers to jobs — and even to arrange meetings or marriages between people. Lenin could not have been more pleased.
But such algorithms now do much more than simply matching individuals or firms based on their stated preferences: They also forecast demand. Artificial intelligence and big data make it possible to predict preferences at the individual level in real time, thereby enabling algorithms to facilitate trade by reducing consumers’ search costs. This is achieved by collecting vast amounts of individual data, often without anyone being aware of how revealing this data is about their political views, sexual orientation, health conditions, criminal activity, wealth, and income. Not even the East German Stasi or the Romanian Securitate could have dreamed that such information would be surrendered without resistance.
So where do we stand? We are at a crossroads. The technology necessary to put in place a centrally planned police state is now a reality. Whether this technology is used for this purpose is but a matter of time. At the end of last year, Jack Ma, the creator of Alibaba, went missing for three months. We later learned that he had been chastised by the Chinese authorities for announcing an ambitious plan that would have strengthened Alibaba’s position in the economy. This warning shot only serves to remind us that algorithms of exchange can one day fall under direct or indirect government control. Alternatively, private interests could just as easily capitalize on this technology for huge private gain.
The question then is: How can we protect civil liberties in the face of these enormous challenges while continuing to enjoy the benefits that all these algorithms generate? This is the biggest question facing us today. How we resolve it will determine how we — and our children — will live tomorrow.
Marcel Fafchamps, an economist at Stanford University, is senior fellow at the Freeman Spogli Institute for International Studies.
This article was updated on March 24 to clarify the reference to the Soviet Union’s creation.