ArXiv usually has most of the scientific papers that I read. Google Scholar gets the rest. But probably the best place is looking up the author an finding his personal website and downloading papers. EDIT: Also check with your university if they have an agreement for papers.

A few different places, depending on the situation:

1. Google scholar

2. Authors' personal websites

3. EBSCO Academic Alumni Edition via my alma matter's alumni association

4. Email requests to friends who are currently enrolled at major universities

5. Visits to the local university (Minnesota) library, which offers terminals for free public access to journals

Google scholar can usually find PDFs which authors' posted to their personal websites.

I use http://citeseer.ist.psu.edu/ and ACM digital library and sometimes google scholar

I thought I would try a couple of the suggestions here. What I was looking for was criticisms or alternatives to (Java) exception handling, an interest of mine. ArXiv may have the the paper but their search interface is pretty primitive. Google Scholar gave much more useful results after tinkering with search terms.

My usual approach to finding papers in a new area of interest is to google around until I find a Wikipedia article that looks like a good starting point. I have found Wikipedia itself to be pretty hopeless for searches. In the article or some of those that it links to, I usually find researchers working in my area of interest. Of particular value are survey articles because of the large number of references. Then I follow the authors to their research departments and projects and co-workers. That usually provides a week of reading.

If I'm looking for something serendipitous I read the comments here and follow the links.

In my (admittingly limited) experience with ArXiv, CS is not well represented. Most CS researchers just don't put their papers up on it.

Try searching on the ACM Digital Library. When you find a title, Google for it (doesn't even need to be Scholar), and you'll often find the author's copy on their webpage.

I'm interested in what turns up here, outside of arxiv - I have no clue, short of the odd self distributed paper.

I'm very much dependent on university journal subscriptions to get full text articles - mostly from sciencedirect. It's a shame, as it puts so much stuff out of reach if you're not in the same position.

I think in the CS and math fields arxiv can be very useful, but papers in my field (mostly CFD related) are fairly sparse.

It might be worth seeing if you can still get access from your university, even if you no longer attend or are affiliated/employed. I know mine offers library services to students after graduation (for life I think), which I assume includes access to journals.

Our university doesn't even offer this while you're studying there, most unfortunately.

filetype:pdf title

Sometimes you can get books this way too but a lot of the time it is just a preview of a first chapter. Another way is to search for a non-prominent quotation from the book with filetype:pdf.

I use this, mixed with Google Scholar, even though I have access to Web of Science through my university. Google's doing a good job.

Author's personal websites...

Usually universities have electronic subscriptions for most journals based on the IP range of the university network.

I ssh one of the computers on campus. cd to my public_html directory and use elinks to download the pdf.

I think many people do this trick because I keep finding good papers in various ~username/ folders indexed by google.

I wrote a firefox plugin to help solve this problem. See https://addons.mozilla.org/en-US/firefox/addon/45524/.

So far it only replaces the link on portal.acm.org with a google scholar search. Contributions beyond this are welcome.

I've managed to keep my university affiliations since graduating so I still have access to most archives for free via my university library.

Once that stops working I guess my next approach would be to mail a friendly request to one of my friends still in the academic world and have them download the pdf for me.

Many universities have alumni "library" programs, where you pay a certain fee to gain access to the university's library. As a student, I've never had to use it though and simply showing up to a site with my university's domain name (VPN) pretty much lets me in anywhere.

Google Scholar.

In California at least, you can also get pdfs of papers at most UC and CSU libraries without being a student. Stanford too, last time I tried. Of course this means you do actually have to physically go to the library in most cases.