I am a believer in Free-Will, but do not think this argument proves Free-Will. My objections are as follows:

1. If it is true that determinism is self defeating for the intuitive reasons suggested, this does not necessarily mean determinism is false. It just means that IF determinism is true then there is no valid argument for anything including determinism and everything else.
2. I do not think it obvious if determinism and free will are a real dichotomy, and I do not think every one agrees with the particular meaning of either term but understand them somewhat differently. For example for my own working definition of Free-Will and Determinism they are not exclusive.

I am sorry, but to me it comes across like you are pushing a false dichotomy.

Specifically it seems you insist I either:

1. nod along and say "yes" to your own language about what this moral obligation is including elastic concepts like "normative obligation" which I suspect you will eventually let me know the meaning of after I pre-agree to it.
2. Reject that there is any moral obligation of honesty in science, other than of course the one that you wish to keep control of defining.

>In other words, if you're presented with a body of evidence sufficient to establish a given scientific proposal to an acceptable degree of probability or certainty, do you then have an epistemic obligation or duty to endorse/accept that proposal...

You are mixing things that are right with things that are not quite right.

Not everyone has a moral obligation to look at all bodies of evidence for all scientific theories. Each of us has a limited amount of time and expertise. The obligation to follow the evidence in a given area of scientific inquiry for practical reasons must fall to a limited number of professionals whom we are asked to trust to examine, explore, and to simplify the evidence for us.

It is the obligation of those professionals to follow the evidence though, even if it means being banned by those who put pressure on them to support a narrative.

>So it wouldn't be a moral duty or virtue to accept any particular theory (e.g. a duty to accept, say, general relativity), only to accept theories well-supported by the evidence (and conversely, to reject theories that have been empirically falsified).

There is a moral duty to be completely open and honest about the evidence rather than be tempted to tweak it due to political and social pressures or for monetary gain or for fame and/or glory.

You seem to be implying that scientific theories are all necessarily of a high epistemic weight. If so, this is not remotely true. Most scientific theories turn out to be false or at the least only partially right. If this were not the case no more research would be necessary.

>as in the case of the increasingly clear negative personal and societal effects of, for instance, vaccine and climate change denialism.

Making the endorsement of a theory into a social virtue and its rejection into a moral crime is no part of science.

Supposing, for the sake of argument, the world really has absolutely no meaning in any transcendental or religious sense as is assumed here. Then indeed the desire to avoid suffering seems to fit very well into the role of a survival mechanism evolved in animals. Just as the desire for food, the desire for sex, the desire to protect one's children, the desire to protect one's tribe and so forth.

What does not fit is a hunger for meaning. It is bizarre that we would have our survival instincts that were finally tuned by evolutionary advantage sabotaged by a desire for imaginary vague notions of some non-existent thing?

Perhaps it can be argued that this need for meaning is actually an evolutionary advantage. Or perhaps it can be argued that it is not an advantage but an unfortunate by-product of other traits that are an advantage. But prima facia it seems to be something that is more than a mere side effect and not at all generally advantageous to evolutionary selfishness.

Before we make a bargain with ourselves to humor what we think a false desire in us, how about we make sure it isn't pointing to a real object?