My bad experiences stem mostly from established publishers such as Springer or Oxford University Press, who are often very important for certain application areas, such as as bioinformatics. The point about equations is very much true, though: _despite_ the fact of existing LaTeX templates, they take your paper and typeset it once again---often with detrimental results :-/
Concerning the peer reviewing: I fully agree! I myself am seeing reviewing as an honour and a duty to support my field. I am objecting to the reviewers/editors doing all this work and the publisher charging ludicrous prices to access a paper. Wiley still charges 42 USD for a PDF download and online only access of a paper I wrote 6 years ago! This is absolutely bonkers and not useful for people in countries/universities that are less well-off...
That's the problem really: ACM viewing a digital library, journals, and conference proceedings as a cash cow to be milked for many unrelated purposes.
ACM seems to have little interest in real open access as long as they are hooked on milking that cow.
What's also terrible/embarrassing is that the ACM digital library doesn't even have an API, and its ToS forbid programmatic access.
I would pull the plug on ACM if I could.
This is actually a bad thing, and a sign of standards slipping. Even if one uses a LaTeX template there are all kinds of quality-typesetting nuances that many authors are not aware of: where non-breaking spaces are necessary, proper hyphenation of foreign-language names, en dashes instead of hyphens in ranges, etc. I have seen so many publications in maths and sciences where the author was expected to provide camera-ready copy and the final result was sloppy.
I agree that learned societies can do a great job of publishing journals and Festschriften, and it is the norm in my own field where our learned societies never handed over their journals to a for-profit publisher. But I am happy that in my own field one's submissions still get hands-on work by trained copyeditors and typesetters so that the final result is perfect regardless of the author's own typesetting competence.
In ACM / CS, the publishing pipeline is now fully based in LaTeX, so authors effectively do their own typesetting for their own articles. There is now a standardized template for submissions, so the version of the draft you send to peer review is typeset in the same way it will be published in the final version. In my experience, almost always the version that gets published is identical to the final "camera ready" version I submit after peer review.
Finally, peer reviewing is indeed not paid, but the way my fields treat peer review is that you are supposed to review proportionally to however much reviewing you obligate on others by the papers you / your students / your lab submit to peer review. If your paper goes out to three reviewers, you sign up for three reviews. Then for the reviewer coordinator / meta-reviewer burden, if your group/lab is submitting 6-8 papers a year, then you have an obligation to be on the editorial board / program committee, which comes with the reviewer coordinator burden of what a decent sized group/lab obligates on others. Of course, some people are still free riders, and many people submit publications who are not qualified to peer review, but it does change how you think about the "peer reviewing is free labor" issue.