Online databases maintained by print publishers (AMS's MathSciNet, AP's IDEAL, Springer's LINK, JSTOR, etc.) aren't free. Sure, you and I can use them, but that's because our universities paid for access. What about smaller schools that can't afford access? What about readers outside academia?
Please put your papers and reviews and other writings on your own web pages, or in free archives such as arXiv or the K-theory archive or the algebraic number theory archive or the cryptology archive. I'd like to see each department commit to maintaining a permanent archive of its papers, suitable for citations: for example, http://cite.math.princeton.edu/1995/wiles/flt. This will scale much more smoothly than a centralized global archive.
If you have pre-TeX articles, please make an effort to put them online. The International Mathematical Union has issued a Call to All Mathematicians to do this. Scan the articles, or hire an undergraduate to retype them. Retyping is better, because the resulting files are much smaller.
Librarians have several big scanning projects such as http://www.numdam.org/en/. Check these databases; perhaps a librarian has already scanned your old articles and put them online!
One way to show your support for papers on the Internet is to sign the Budapest Open Access Initiative.
Please don't sign any contracts that prevent you from authorizing these activities! In several cases I've said something like
This paper is entirely my own work. I have put it into the public domain. Luxury Press is therefore free to publish it.instead of signing a copyright transfer agreement. If you ever encounter a publisher that doesn't accept this, let me know, and I'll be happy to blacklist that publisher here. I'm now blacklisting IEEE and ACM.
Perhaps you transferred a copyright to a publisher years ago. You can ask the publisher to put the paper into the public domain, but what if the publisher doesn't agree? You can't put your paper online without violating the copyright. Fortunately, copyright law in some countries lets you terminate the transfer, no matter what you signed:
If you have questions or comments about copyright, try the mathcopyright mailing list.