So there are several things not accounted for in the infinite value math:
First, direct opportunity cost in the form of compounded interest. My estimate is that if you took $20 out of your high yield savings account (or some other highly liquid savings instrument) to pay for GamePump, and you bought day one, you'd be out about 4 cents interest. But many people didn't buy day one, so I would guess the expected loss is actually closer to about 2.8 cents interest integrating across the time-space of purchases. And of course most people didn't take from their savings account, so they voluntarily made a decision to ignore the opportunity cost. As a result, I feel that 3 cents is a conservative opportunity cost.
Second, opportunity cost in form of reduced capital liquidity. This only impacts purchasers who used their last marginal $20 in liquid capital during the period (i.e. emptied their accounts if they have no access to credit, or maxed their credit instruments during the period). Setting aside the very obvious impulse to beat those people over the head because of their terrible life decisions, I would guess that for most people this cost does not apply. Just the simplest version to look at this is that imagine interest doesn't exist and credit doesn't exist and you have $100 and loan a friend $20 for a month and then they give it back. There was only ever a cost if you planned to buy other things worth between $81 and 100 in the month AND there's a non-interest time value of money to have those purchases early. I am unconvinced this is worth modelling, except for Rhaknar.
Third, the naive assumption fails to model consumption value of getting hyped for GamePump. It's clear that constantly talking about GamePump and following the epic saga gave people enjoyment, and so any model that purports to be fair by taking into account interest must also take into account positive utility from this. You might say "yeah, but weren't a bunch of those people anxious and disappointed about the lack of communication and ultimate cancellation of GamePump?" -- that's true, but I have no idea why you're assigning anxiety and disappointment a negative net utility. Some people are clearly dispositionally anxious, and spend their whole lives being constantly having a conniption about things that don't matter. I think two possible interpretations here are either that they derive utility from anxiety, in which case adding anxiety to their life is EV+, or that they do not derive utility from anxiety, but sources of anxiety are fully substitutionary goods, and so being anxious about GamePump is EV neutral because if they weren't worrying about that they would have spent the same time worrying about something else.
In summary, I think a view of this situation fully informed by the considerations of behavioural economics would vindicate GamePump as having an extraordinarily high positive utility value, even if finite.
