This TechCentralStation piece muddles an interesting discussion of strategic behavior by suggesting that behavior that’s strategically random—as when an agent plays a Nash equilibrium “mixed strategy”—is “irreducibly uncertain,” which is to say, an exercise of “free will.” First, it’s worth noting that even genuinely “irreducibly uncertain” behavior would not be the same as “free will”—certain behavior could (in principle) be keyed to inherently probabilitic micro-behavior in the brain, which wouldn’t be the same as “freedom” in the Cartesian sense of being the act of an agent as a “prime mover unmoved.” But even that is rather unlikely: in fact, the attribution of strategically random behavior to anything as robust as genuine nondeterminism, or “freedom,” is a classic example of theorizing beyond what the data will support. What game theory requires is that the randomizing process be opaque to third party observers, not that it be genuinely random. Computers generate “random” numbers all the time that, of course, are not genuinely random at all, just based on an elaborate enough algorithm taking arbitrary inputs like the clock time or the number of milliseconds between the last two keystrokes. Speaking of this as “free will” just helps to put a pseudoscientific patina on a bad (indeed, incoherent) metaphysical idea that deserves to have died off ages ago.