Utility is what those Bernoulli experiments led up to. The part about increasing variance for a chance at more potential upside, when we don’t care about the potential downside, is known, and is moral hazard.
The borrowing/lending constraints used importantly here are not generally present. Assuming borrowing and lending of money and shares in real life is generally a good assumption (but not in Gamestop's case, granted).
Also, ‘prisoner’s dilemma’ is a term already used in game theory for a long time to describe a different thing. Probably should call this something else to avoid possible confusion. There’s no dilemma if all prisoners’ optimally desired choices can be accommodated in a single outcome.