AFAIK, licensing agreements between libraries and digital publishers typically include a right to perpetual access to articles in a journal or digital library they subscribe to. Any papers published while the licensing agreement is in place are to be made available even if the library cancels their subscription.
This also includes the right to make backup copies of the digital library.
What the library is allowed to do with this backup probably depends on country-specific laws.
However, if I understand the current situation correctly, ACM allows everyone to download CACM articles but not to redistribute them, right?
In theory, they could also explicitly allow anyone to redistribute all of their publications (they reserve the right to do this in the licensing agreement with authors).
However, I'd argue that it is not in ACMs interest to do so - and probably not in mine, either. Having a central, trustworthy source has benefits - e.g. when retracting papers.
I'm fine with the ACM DL being free to use and having library backups as safety nets if ACM ever goes under.
it's definitely in acm's interest to do so if they want to claim they're 'open access', and it's in your interest too
i agree that it's beneficial to have a central, trustworthy source. if redistributing cacm articles is clearly legal, and the acm goes under, you'll get them from the internet archive, wikisource, project gutenberg, google's new historical acm article archive, or some similar institution. if redistributing cacm articles is not clearly legal, you'll have to get them from library genesis, sci-hub, or the pirate bay, with the associated risks of incorrect contents—but only until those get shut down following a change of government
well, maybe you personally won't, because you have a position at a university, and you can easily just use its library. but the people you'd like to have as your students 20 years from now, people who are now living here in argentina or in ukraine or egypt, will have to get them from there
an additional issue is that, in many cases, the 'central trustworthy source' does a really bad job of scanning paper documents: they use bilevel scanning for photographs (sometimes even color photographs), scan at very low resolution, chop off the pages, and so forth. i run into these constantly; https://www.nasa.gov/history/alsj/a17/A17_LunarRover2.html is probably the most recent example. although allowing other people to scan their copies and distribute the scans does not ensure that they will correct this problem, prohibiting them from doing so reduces the chances further
analogous remarks apply to translation
and that's why the berlin declaration enshrines the right for readers to redistribute works and derivative works as central to open access
AFAIK, licensing agreements between libraries and digital publishers typically include a right to perpetual access to articles in a journal or digital library they subscribe to. Any papers published while the licensing agreement is in place are to be made available even if the library cancels their subscription. This also includes the right to make backup copies of the digital library.
(Here's such an agreement between ACM and the Norwegian universities: https://www.openscience.no/media/3449/download?inline )
What the library is allowed to do with this backup probably depends on country-specific laws.
However, if I understand the current situation correctly, ACM allows everyone to download CACM articles but not to redistribute them, right?
In theory, they could also explicitly allow anyone to redistribute all of their publications (they reserve the right to do this in the licensing agreement with authors). However, I'd argue that it is not in ACMs interest to do so - and probably not in mine, either. Having a central, trustworthy source has benefits - e.g. when retracting papers. I'm fine with the ACM DL being free to use and having library backups as safety nets if ACM ever goes under.