> I often hear that researchers don't want to limit access to the paper, but feel they have to, to publish and progress.

I have published in several ACM conferences and this is how I feel. I much prefer USENIX's policies.

> and that while its easy to say 'make it free', it costs money to run an organisation that manages publications.

The actual research is funded by the government or in increasingly rare situations, private companies. Reviews are volunteer labor. Post-review editing (what they used to call 'typesetting') is done by the author. Putting up a PDF somewhere costs more or less nothing. So yes, the essential parts could (and should) be free.

The actual money goes to printing paper (who cares), subsidizing conferences (an event where you hang out with your friends, and then give your speech to an audience of people waiting to give their speech), and subsidizing other ACM stuff (inappropriate for the taxpayer to incur this burden).

Until the ACM makes our work free to those who have paid for it, I will not be a member of their organization. I will publish in their conferences (you have to...), but I will never link to or reference their 'official' copy.