Game Theory and Tech's Rightward Shift
An Application of Prisoner's Dilemma and Iterated Play
There have been a lot of spilled bits about tech’s rightward shift in the 2024 election cycle. There are several caveats to this swing:
tech workers still voted for Democrats in higher numbers than Republicans
it’s better to be on a vindictive president’s good side, so some executives and VCs went towards Trump when he appeared to be winning for purely self-interested reasons
the libertarians, neo-reactionaries, and National Soci…, ahem, National Conservative members of Silicon Valley felt more emboldened to speak up and speak with their dollars
You might recognize my nome de plume as an ode of sorts to one of those right-wing geeks, although it should clue you in on what I think of his ideas.
With those caveats, let’s proceed with the premise that tech indeed went towards Trump in 2024. What does game theory have to say about why and what to do next?
Prisoner’s Dilemma
You’ve likely heard about prisoner's dilemma. If you have, you’d likely rather gouge out your eyes than read another explanation of it. For that I apologize. Please feel free to skip to the section on iterated play. For the other readers, you have two criminals who committed a crime together held in separate interrogation rooms to prevent their coordination. If they both keep their mouths shut, they will be convicted of a lesser charge and serve a short sentence. If one rats the other out and the other one remains silent, the tattle-tale gets off scott-free whilst the silent one gets a harsh sentence. If they both rat on each other, they both get a medium sentence. A payoff matrix holding their utilities in each scenario is below. Let’s call our criminals Alice and Bob. The cell contents hold tuples of Alice’s utility and Bob’s utility in that order.
So what happens? The rational thing for each criminal is to defect (i.e. drop a dime on the other). Why? From Alice’s perspective, if she tells on Bob and Bob stays silent, Alice gets no punishment. If she tells on Bob, and Bob also sold her out, she gets a medium sentence which is better than the harsh sentence she’d have got if she stayed silent. The same is true from Bob’s perspective. Defecting is the Nash equilibrium; Alice or Bob is better off defecting no matter what the other player does. This lands them in an outcome where they are both worse off than if they had stayed silent.
Iterated Play
Iterated play of prisoner’s dilemma changes the game. If Alice and Bob are playing an unknown or infinite number of rounds and they can remember the earlier rounds, cooperation without direct communication becomes possible. In fact it becomes a better strategy. Do not defect on the first round. After that, do what the other player did in the previous round. Classic tit-for-tat. This allows Alice and Bob to end up in the upper-left quadrant of the payoff matrix (the good quadrant). There are optimizations to be made, like randomizing whether you punish a defection in the prior round, but the main idea to take away is tit-for-tat. So how does this relate to tech going right?
Democrats Defect
Imagine US elections as iterated games. There are going to be an unknown number of them (fingers-crossed there are many). The two Obama administrations and Clinton’s campaign were tech friendly. Then Clinton lost. The backlash against tech, particularly social media companies, for the spread of disinformation in the 2016 election changed things. Biden’s administration was actively hostile to the industry.
The Federal Trade Commission under Lina Khan went after murders and executions. Sorry, mergers and acquisitions. It targeted a good portion of the FAANG/Magnificent 7/whatever new term Wall Street cooks up for alleged monopolistic practices that didn’t harm consumers. This obviously pissed off the big dogs, but as Noah Smith notes it was not popular with the wannabe monopolists (start-up founders) and their investors.1 The Securities and Exchange Commission under Gary Gensler was skeptical of crypto (as am I) and cracked down on the industry. Biden’s executive orders on AI safety were not well-received in Silicon Valley. In short, the Biden administration, and by extension Democrats, defected. The response by some in the tech industry was to go tit-for-tat.
How Democrats can win back Tech
Many people to the left of center online want blood from the tech industry. In the framework of prisoner’s dilemma in iterated play, this would be a mistake. “Defecting” again in the next round might lead to a defection spiral and firmly entrench a portion of Silicon Valley on the right. In this framework, Democrats should cooperate. Adopt friendlier policies and where good policy conflicts with tech’s interest, be nice about it. For example, don’t go after big companies simply because they’re big. However, if a company is hurting consumers or stifling creative destruction, by all means try to cut them down to size. But also take their calls, which Biden officials apparently didn’t.
There are some who think Democrats don’t need to win back VCs and tech executives and I’d agree they don’t need them to win elections. They may even make good targets for populist ire, although I abhor populism. But it would probably be easier to win a popularity contest when more people like you. In the world of Citizen's United, that especially means rich people, regardless of your opinions on the matter. Sometimes you have to set aside your ideas of how the world should be, and act strategically in the world in which you actually live.
