As other people keep pointing out, investment decisions don't much care about what might happen 50 years in the future. Windfalls that happen in the future increasingly inure haphazardly; they don't actually figure into the finances of how the industry works, not even as magic helicopter money.
Economists have run the numbers. They have models for this. That's why extension advocates principally hide behind theory, not empirical evidence, because the empirical support is poor at this point.
EDIT: Let me put it a different way. If all that matters is any marginal effect (and especially if we ignore opportunity costs in the public domain, which is admittedly difficult to quantify), why not make the copyright term 1000 years? Or in perpetuity? If every time someone says, "let's extend it by 10 more years, and here's a hypothetical by which it might incentivize someone in theory", by what principle are you prepared to put your foot down? My principle would be evidence-based, and what evidence exists favors stopping terms extensions, and probably rolling them back somewhat. And fortunately that evidence is also consonant with economic theory, no less so than the hypothetical arguments for continued extension.
> the winners pay for the enormous number of losers.
The original point from someone else way upthread was that the vast majority of revenue streams from published works, especially film and TV, are front-loaded. Yes, most works don't return much value; and, yes, the big winners cover the losers. But the context of this whole subthread is prospective marginal values of copyrights decades in the future (including windfalls in the distant future--I'm trying to strongman your argument) and the degree to which they incentive production of more content.
Your original point (AFAICT) was that people ignore the prospective value of extended terms and its effect on pricing and incentives in the present. And my point was that the FPV of such marginal returns are de minimis, even taking as a given that they're positive and include huge windfalls. Your point makes most sense as a retort to people wishing to roll back copyrights to, say, 20 years or especially 0 years--i.e. only have the right of first publication.
The more meta point is that we could argue hypotheticals all day, but what matters most of all is the quantifiable. Arguments from strong patent and copyright advocates are often analogous to, e.g., the shoe bomber hypothetical in security theater--yes, shoe bombers pose a hypothetical threat, and in fact a shoe bombing incident has even come to pass, but that alone doesn't necessarily justify the airport security regime established in response. To justify it we must look (as best we can) at the quantified costs + benefits, and then make judgements accordingly. So the fact that, hypothetically, marginal returns from copyright extending past 50+ years in the future might effect pricing regimes in the present--or even if we know they in fact do--doesn't resolve the question; it only begs the relevant question: does that value exceed the costs?
The parents point I think is that if you make 100 films today you don't know which will be hits but you can expect X to be and they will pay for the losers.
I don't think that requires the trickle down effect that you seem to be referring to with Seinfeld.