[mbarbar]

> So, this requires the author, and by extension, taxpayers in most cases, to pay a couple of thousand to ACM to put a pdf online. How does this make sense ?

Tangential, though at a recent SIGPLAN conference, the cost to authors to licence under CC-BY was listed at 400USD. In place was "The article processing charge (APC) is to be paid anyway for all PACMPL articles, and ACM SIGPLAN covers it if you cannot afford it. Your paper will be open access no matter which kind of publishing-rights agreement you choose.".

That said, from what I can tell, most of the main programming languages conferences have far better systems and communities in place than many other fields (a lot of open access, double blind reviews, etc.).

[bubblethink]

At the very least, I hope that all the top conferences (e.g., the ones in http://csrankings.org/) become open access by default at no additional cost to authors. There are a lot of strange discrepancies even between sister conferences. For example, VLDB is open access, but SIGMOD isn't. It's all the same people in both conferences.