Here is a brief excerpt:
By owning exclusive publication rights to articles, ACM is able to develop salable publication products that sustain its top-quality publishing programs and services; ensure access to organized collections by current and future generations of readers; and invest continuously in new titles and in services like referrer-linking, profiling, and metrics, which serve the community. Furthermore, it allows ACM to efficiently clear rights for the creation, dissemination, and translation of collections of articles that benefit the computing community that would be impossible if individual authors or their heirs had to be contacted for permission. Ownership of copyright allows ACM to pursue cases of plagiarism. The number of these handled has been steadily growing; some 20 cases were handled by ACM in the last year. Having ACM investigate and take action removes this burden from our authors, and ACM is more likely to obtain a satisfactory outcome (for example, having the offending material removed from a repository) than an individual.
Personally, I gladly pay money every year to the ACM and the IEEE, as I feel I get excellent value for it. I don't begrudge them their business model, and I don't think there is anything particularly nefarious about their copyright policy (which explicitly allows authors to post freely available copies of their articles for non-commercial purposes.)