I think your reading is in fact over-charitable. He is clearly referring to "weird creatives" as people who behave oddly without explicit negative effects, and is trying to argue that oddness has some diffuse, unobservable cost on society; if they are simply (creatively) stealing there would be no need to argue this. I think it's also important to remember in these discussions that there's often skin in these games. Trying to find baileys to which bigots can retreat might promote dialogue and openness that results in wandering truth-wards, but may simply spread misinformation. It's worth flagging that there's more than 50 years worth of consistent evidence that children from same-sex families do as well as [1, 2, 3], if not better [4], than those from straight families, and his statements based on weird extrapolations from straight couples contribute to their marginalisation. In this specific case, Peterson is explicitly arguing against updating in the face of evidence and arguments ("conservatives do things because it's how we assume we've always done the and we can't be expected to remember why"), which breaks a lot of otherwise good Bayesian argument advice.
[1] https://www.tandfonline.com/doi/full/10.1080/1550428X.2013.869486
[2] https://www.tandfonline.com/doi/full/10.1080/15504280802177615
[3] https://pubmed.ncbi.nlm.nih.gov/12361102/
[4] https://psycnet.apa.org/doiLanding?doi=10.1037%2Fsgd0000203
Adding religion to the equation immensely complicates this topic, since most religions posit the continued existence of the soul after death. In this case you're not really respecting the wishes of past people, per se, but merely people you can no longer see. When we talk about respecting the graves of the dead, I suspect it's a little of this persisting: maybe they're still there somewhere, watching. In this case, include their utility in relevant calculations, times the probability that they exist.
The main topic that I feel is missing from this discussion is consent. There are always reasonable boundaries to what any contract, with yourself or others, can force someone to do, and we now accept that even legally-binding promises such as marriage can't bind our future selves unconditionally. Insofar as your future self is similar to your past self, your previous consent binds your future self, but even when contracts don't have official get-out clauses there are practical things that will prevent them from mattering. Although you're the same legal entity, I'd argue that in cases of Hollywood amnesia, for instance, the separation of past and future self is sufficient to make marriage vows morally dubious. It is only because of the similarity between 17! and 27!Austin that you can bind yourself to anything at all: I see no reason to generalise this obligation to your past self to your ancestors. They don't get utility out of your actions, so the only reason to worry about their opinions is where there is a contract you've consented to. You didn't consent to be created, so have no intrinsic obligations to your ancestors.
Regarding whether your descendants are forced to fulfil a will: they have not consented to the contract, but can simply not take the money if they wish. If they take the money, they are consenting to reasonable requests in the will, although if the requests are clearly absurd (such as saving something that has already been destroyed) then the conditions will just be ignored.
"The people in the past were roughly as intellectually capable as you are": maybe in genetical potential, but not phenotypically, as shown in the Flynn effect. "The people in the past had similar modes of thought, similar hopes and dreams to you." They were mostly farmers with very limited capacity for abstract thought. To the extent that their hopes and dreams are similar to mine, I see no reason to double-count them. Furthermore, they don't have working knowledge of the current world, and giving them it would render them effectively different people, by your own arguments.
I wrote a rather long article in which I show that almost every aspect of this study is flawed from both a statistics and EvPsych point of view, some of which other commenters have mentioned but several of which are new. TL;DR: RVs as defined in this paper completely ignore parenting, so can hardly explain the evolution of a parenting behaviour; the data collection is poor for doing proper statistics (hence they don't) so the total population count is not a good measure of the error; and it's easy to get absurdly high correlations between two time series because they autocorrelate, which is why you should never do this sort of analysis even under a "correlation is not causation" gloss. https://scienceisshiny.wordpress.com/2020/09/11/everything-wrong-with-the-paper-human-grief-is-its-intensity-related-to-the-reproductive-value-of-the-deceased/
Agree on the first point. On the second point, my comment doesn't rely on Peterson arguing in bad faith, merely that he is arguing with excessive faith in his priors - Bayesian reasoning doesn't work if one person has 100% confidence in their initial position, and may be very inefficient if you have extremely strong priors and don't update well. He may sincerely believe that same-sex families can't bring up children properly, but if his position is unlikely to change much from the argument, the social effects of how you engage with it (the effects on onlookers who may be insulted by the argument or marginally update towards his viewpoint) may be larger than the benefit of his marginal update.