My coauthors and I have just had a paper accepted by a journal. The journal in question is published by a nonprofit professional society and offers two options in their copyright agreement.
We transfer copyright to the publisher, as is traditional. The journal has exclusive publication rights but we retain the rights to post the paper on our web pages or arXiv (post peer-review), distribute to colleagues, etc. We also have the right to release our paper to the public domain after 28 years.
We retain the copyright for ourselves, but grant exclusive publication rights to the publisher. We still have the rights to post and distribute the paper privately as before.
I am trying to determine the pros and cons of each option. From a practical standpoint there doesn't seem to be much difference: in each case, both we authors and the journal have the rights to do the things that we respectively care about.
In principle, it would seem better for us to retain the copyright rather than transfer it, just on the grounds that it is better to keep as many rights as possible. I could only think of one possible drawback: suppose that a number of years from now, some person X wants to republish our paper, or use it in some other way that requires the permission of the copyright holder. If we transfer it, X can work it out directly with the journal, who should remain easy to contact. If we don't, X has to track down all three of us authors (or our heirs) to get permission, and then we have to decide what to do about it.
I'm not asking for legal advice, and I know you're not a lawyer, etc, etc. But I would be interested to hear about any experiences with this, or other issues I haven't thought of.
The publisher is the American Mathematical Society and their copyright agreement form can be found at http://www.ams.org/authors/ctp.pdf