[reeses]

Google Scholar is pretty good for indexing the major libraries/journals, but you still run into the paywalls in places.

It also turns up copies of paid papers that someone has stored in a public web/ftp server. Usually it's a student obviously working on a project. I must admit that I've used these on occasion to decide if a paper is worth buying.

[mjn]

I guess it depends on your personal views, but I don't find much wrong with just using these articles. Some people even leave them "accidentally" open not really by accident (I have some colleagues who conspicuously fail to take much care to keep Google out of indexing their course reading lists).

If you really want to cross your t's legally and still not pay, you can almost always email the author and ask for a copy, and they are typically authorized by their publication agreement to privately send you a preprint. So in practice just using a version you found online is not much different, except that you save the author some time responding to your request for the same PDF by email.

[reeses]

Everyone wants the citation. Good point. :)