No, the publishers don't make it availble; you need to look for it on the authors' home pages. Most of the time, if you Google the title with "filetype:pdf", it'll show up.

In the last several years there's been maybe half a dozen times that I failed to find a computer science paper online. I'd estimate that less than 1% of papers in the top conferences are unavailable freely online.

Computer science journals are pretty unique in that they let authors get away with that. In any other industry the papers are owned by the publisher, not the author, and putting a link to your own paper on your own website is a copyright violation they will take you to court for (or just blacklist you so you never publish again).

Not true. In astrophysics, at least, all journals I know of let you keep rights to noncommercial publication like putting the paper on arXiv and on your web page.

Sorry, that's not my experience. And I read a lot of papers. Most of the time they'll link to a walled site or have no link at all.

You have to look for personal or institute pages. For example, one of my papers is behind the ACM paywall [0], but you can also get the pdf at our institute website [1], where we publish everything [2].

As a last resort, mail to the authors.

[0] http://portal.acm.org/citation.cfm?doid=1878921.1878926 [1] http://pp.info.uni-karlsruhe.de/publication.php?id=buchwald1... [2] http://pp.info.uni-karlsruhe.de/publications.php?orderby=1&#...

And even if you have ACM digital library membership, sometimes the paper refered to in the ACM DL is hosted over at Springer or the like and you must pay 30$ or more to access the paper.

I rejoined ACM specifically to get one particular paper only to realize exactly this. The next two papers I looked for the exact same thing happened.