Ideally, yes, but at a minimum he should provide a list of changes and their rationale(s) in the newer editions.

If he edited his book to rebut his own argument and lays out the case then maybe not but it should be stated in the forward that the change was made.

We are moving forward into a time where media is only digital and can be changed at will, it is important to establish a system where when we make changes we record them and also why we made the changes.

If we do not, the only future is one where information is as ethereal as the morning dew and subject to the zeitgeist of the times.

In other words, the diff between editions may be even more interesting than the either edition by itself.

If someone else changes the text on a book from a dead author... how do you know the author would have liked that?

I remember a while ago I was reading on a stage play that's supposed to be only 4~5 men (no women on the play), and the - now dead writer - specified that this should never be changed. And some ppl try to do women-only, or mixed cast, and then the writer's estate sues and tries to intervene...

The only real counter-argument I can make to the position that the author can edit his works but his heirs cannot is that the author chooses his heirs; the author could have instead chosen to will his works into the public domain after his death, or establish a trust or foundation with a strict mandate to republish his works in their original form while returning the profits to his heirs.

It's not a particularly strong argument -- many authors may not have the resources or foresight to make such provisions, and you could just as easily insist on authors who do wish their heirs to manage their works and not simply profit from them make explicit declarations on that point -- which is why I lead with the more extreme case of an author editing his own work.

I can at least talk to you about whether it's reasonable for eg Brian Herbert or (the now-late) Christopher Tolkien to continue to exercise editorial control over their fathers' works, or under what conditions a foundation could be established for the duration of the 90-years-post-author-death copyright the US currently uses.

If it's permissible for Dr Seuss's widow to inherit editorial control of her husband's work along with the copyright, can she pass that control to Dr Seuss Enterprises; if not, could Dr Seuss have established it before his death?

(Personally I think the other reply was better, in stating that it is arrogant to assume that the author's heirs do not know his wishes better than us.)

I guess the widow (or if they are still alive, the editors that worked with the author when he was alive) is the person that probably knows the author best to know if they would've liked their work to be updated or not. Or at least, make decisions on allowing changes that can have impact on how the author is seem (e.g. progressist, non-racist, sexist or not sexist, etc...).

My problem is when someone totally unrelated starts making changes, even in cases where author was explicit that they would never want the gender of their characters changed, and people go ahead and still want to change the gender

Edit: source for what I mentioned on this and my previous comment https://www.irishcentral.com/culture/beckett-waiting-for-god...

There's an obvious counterargument: how do we know [better than the author's estate] that a dead author would have been so committed to the retention of a particular stereotype or word that has acquired different connotations that they'd have angrily refused editors' suggestions they change it a bit for the next edition? In the cases discussed here, it seems particularly unlikely - would the bloke who accepted the suggestion that the workforce of African pygmies he imagined should stop being characterised as African pygmies really feel the purity of his writing was being destroyed if an even later revision made the comparatively minor switch to describing them as "people" rather than "men"?

There's an odd dichotomy between arguments that authorial intent is being destroyed by substituting or omitting a couple of words and various arguments that retaining them serves to create "teachable moments" about "how bad things use to be". With a few notable exceptions, "teachable moments" isn't at all what the author intended by a character using a word considered offensive in 2023 or casually assuming that women or foreigners were constitutionally unsuited to certain things, and generally they weren't trying to imply the character was a bigot (as modern audience might well misinterpret), or say anything at all about the contemporary attitudes incidentally referenced. Authorised translators use very non-literal translations, add or remove gender references or omit minor details that local audiences would completely misunderstand whilst living authors remain unconcerned and usually blissfully unaware; why are we so certain that the dead ones would deeply resent their publishers and estates making equally minor tweaks for the culture of 2020s English speakers?

But as you mentioned, the author's estate, family and editors that worked with them when they were alive are the ones that can probably make the best guess.

My problem is when publishers make changes even at times when authors are explicit in saying they don't want changes, without (not sure about this specific case, I'm just assuming here) getting approval from the estate

Automatic compromise: your copyright to the original is compromised or shortened in a way that makes its survival no longer dependent on you.