It's not their obligation to track down cheaper or free sources. That's silly. After all, the ACM isn't going to waste time keeping track of the various sources of every paper they publish. It's not their job.
If someone is motivated by the fee to find another source, they'll do so. Maybe they'll ask their colleagues. Maybe they'll search the web. Maybe they'll take the time to visit a university library and track down a print copy.
But if they don't care about the price, they'll just do the easiest thing and pay the ACM in order to save time.
It's not the ACM's job to do the legwork. If you want to save the money, you have to do the work yourself.
Who do you think should pay for the servers, maintenance & upkeep of the ACM site? It's a non-profit organization supported (barely in some years) by membership dues.
For example, there is an ACM Portal page ( http://portal.acm.org/citation.cfm?id=1553489 ) for a paper I wrote that is freely available from the conference site ( http://machinelearning.org/archive/icml2009/abstracts.html ).
There is no indication on the ACM page (the DOI on that page just returns the ACM page) that this content is available elsewhere without the $15 fee.
It's bordering on unscrupulous in my opinion.