Algorithms, Prices, and Economic Calculation
Ozimek’s article is a somewhat disjointed list of firms that use algorithms to price things, and some observations on other examples of when markets work pretty well. It’s pretty short, so perhaps he just didn’t have enough space to fully develop the argument, but I’m not sure what he was going for.
The idea that you could run Uber without markets is so distant from reality that neither of the Adams even consider it. The non-market Uber would not just algorithmically compute prices, it would try to actually allocate rides to specific people at specific times, potentially years in advance. Central planning is not about coming up with a single price, given an environment where all other prices are determined by competitive markets. That’s easy. It’s about coming up with all prices, quantities, and allocations simultaneously. That’s the vision Hayek and Mises were refuting, not the idea that companies should set prices according to supply and demand via algorithm.
A Firm is not the market. Algorithms are a pricing strategy. Yes, it’s better that Uber’s price goes up when it rains, but algorithms can’t replace markets, because they require the context of a market to work. For example, Uber’s algorithm would have spit out higher price if not for Lyft. The reason Uber gains so much benefit from algorithmically pricing its product is precisely because their main competitors, taxis, don’t vary their prices at all. They are the margin, and as such, they must react more than if everyone varied their prices a bit less.
Gurri approvingly quotes Harry Collins: “[Algorithmic pricing] only works because the surrounding social organism makes up and ‘repairs’ its deficiencies.” Exactly right. Without the price system to transmit the information necessary to feed into the algorithm, it would not have the raw information necessary to make a reasonable calculation. Customers need the prices of alternative goods to decide whether they want to spend their money on Uber rides. Drivers need not only to consider the wages of alternate jobs, but also the subjective value of their leisure time.
Firms are islands of central planning. They don’t operate as markets, because of the transactions costs of using price signals. Firms have always used algorithms to allocate resources within themselves, this is nothing new to the computer age, it’s just that our algorithms are better today. I don’t think make or buy decision is as critical to the debate as Gurri implies, but it is an important part. The totality of the victory of the free market side of this debate is demonstrated by the fact that slight variations of their vision is considered a refutation, whereas the socialist side is not considered at all since it is too absurd.