Reading the introduction to von Neumann’s Games and Economic Behavior. He’s very in touch with the poverty of the state of economics, the huge challenges involved, and the need to focus on making a difference where differences can be made.
His portrait of how physics and mathematics coevolved is crisp and beautifully motivates how difficult applying the mathematical tools of the day to social problems is. I think he’s after something like the Asimovian dream of psychohistory—of actually being able to predict human behavior on generational timescales and engineer societies based on those predictions. That same dream has drawn cybernetics nerds into econ for generations.¹
One thing that stands out: he smacks down so many criticisms of microeconomics still bandied around today. He does it very well, moving fluidly from one point to another, always hemming the opposition in. I’m happy, because he puts these arguments in such wonderful context. I’m sad, because people still make them now and don’t seem to overcome his responses when they do so.
“Why does econ focus on these toy problems?” Because they’re tractable and let us compare theory with both observation and intuition, which is unskippable foundation-laying (compare probability theory, which was first used to characterize obvious problems before we got things like Buffon’s needle).
“Why doesn’t sprinkling math on economics work?” Well, applying calculus works when doing marginal analysis, but most of the time, we mostly don’t know what’s happening. There’s often no setup—no ansatz—we can do to gain new insight. When economists do this, they’re often just putting fancy mathematical clothes on their verbal arguments, not discovering anything new. And calculus itself emerged from the need to solve kinematics problems in physics, and the kinds of problems we want to understand in economics often seem pretty different from this. Von Neumann really hopes that we’ll discover new kinds of math to better understand economics, and Games is meant as a step in that direction.
“Why bother with math at all? Trying to reduce humans to a bunch of numbers is foolish!” Well, we can observe humans exhibiting preferences and making choices, which immediately suggests there’s quantitative data (ordinal utilities) we can work with. And studying the impact of ordinal utilities at the margin using calculus is no more problematic, von Neumann argues, than studying clumps of atoms and other indivisible quanta as continuous bodies.
“Why don’t we try to understand more important and complicated systems, like the US economy?” Because the system is complex and the data is pitifully sparse for that complexity²—and there’s nothing to be gleamed from very complex data that we cannot theorize about. “Observation is theory-laden” isn’t language von Neumann had, but he seems to be reaching for this idea. Von Neumann even does a David Deutsch-esque maneuver of saying, “we scanned the heavens for millennia in vain before it gave us ideas, which made all the difference.”
—
1. The boldest form of this vision has a serious problem, which Karl Popper elucidated: as long as new knowledge is being created, and as long as predicting human behavior depends on understanding the state of human knowledge, prediction will always be limited, because the discovery of new knowledge is unforeseeable by definition.
2. I think a dynamicist breaking this point down further would talk about things like the number of parameters (and the enormous phase space that the economy must live in), the lack of stationarity on the timescales we can look at, and how few of the driving processes can be observed. As of 2023, my understanding is that most macroeconomic models taken seriously need to capture both behaviors we know must happen (like capital costs reflecting technology efficiencies) and strongly suspect must happen (like hysteresis in labor markets leading to sustained unemployment). Just capturing those behaviors makes the models so capacious that falsifying them, never mind fitting them to reality, seems hopeless. What’s amazing is that von Neumann must have known much less about these things when he was writing, but he understands this phenomenon of data poverty extremely well.
why does no one ever talk about being able to physically feel the last thing you copied. like youll right click a url and copy it and be able feel the weight of it like its in some kind of inventory
leave my sylladex out of this she didn’t do anything to you
