It's about the same as everyone else: about 15 USD per paper, e.g.

https://dl.acm.org/purchase.cfm?id=2756548&CFID=761526954&CF...

They also have a somewhat restrictive licensing on their software:

http://www.acm.org/publications/policies/softwarecrnotice

This part is kind of weird, as they have a noncommercial clause on their copyrighted algorithms, which makes them non-free.

On a slightly different note, I was somewhat saddened by how the International Lisp Conference talks were recorded but unavailable because the ACM will not allow their publication.

How the hell can that be "non-profit"? Where is the money going? Free conferences in Dubai?

The ACM publishes annual reports with a high level breakdown of their finances. The most recent one is here:

https://www.acm.org/binaries/content/assets/about/annual-rep...

The finances are on pages 12 and 13. The short version: they're spending money on all the things you'd expect, but they seem to be slightly cash flow positive and have accrued around $100m in net assets.

What I'm wondering about is the economic justification for charging $15 per article copy. Other than "we can", I mean.

The usual "economic justification" for such things is that people are paying it.

In reality, very few people pay the individual article prices. The real market is in site licensing, as it were. The individual article prices are probably set to support the sales story on these institutional access licenses, not the other way around.

I think you can also get access to the ACM library via membership. When I was in college I paid for a student membership, I think its was $25/year or something generally reasonable. Probably a better value proposition than my actual CS degree, not that its a great point for comparison.

That's true. Many of these sorts of memberships are at least a few hundred dollars when you are no longer a student, so they add up pretty quickly (but maybe have tax credits).

Paying $25 annually for membership would be entirely fair. That's far less than site license for even one journal.

But what are criteria for student memberships?

Most definitely, here's a list of journals: http://dl.acm.org/pubs.cfm?CFID=591346815&CFTOKEN=85546644

And here's an article that, for instance, costs $15 to download: http://dl.acm.org/citation.cfm?id=954342&CFID=591346815&CFTO...

The ACM is paywalled for the majority of the publications it manage.

I found the ACM paywall to be one of the obstacle that limit access of the new wave of the code educators to older results in teaching programming: they are not in the academic sector and lack both knowledge of and access to this enormous amount of papers.

So, in this case, the paywall keep a new generation naive, as they reinvent independently research that have been conducted decades earlier.