Briefly responding to a couple of points raised so far:
Yes, there is a need for a policy as otherwise the WMF would have no long term operational archive plan. "Self evident" is insufficient in order to budget and plan in a credible way. If as the planned outcome of a research project I had a large image donation to make and such a commitment was absent, I would prefer to mass donate images of public interest to an organization that had one, and assume that at some point e-volunteers at Wikimedia Commons would take the initiative and port in what they fancied.
The people I'm workshopping with tomorrow have research roles within a number of leading universities along with a number of research organizations under the umbrella of the Wellcome Trust (the largest charity in the UK) and a variety of semi-associated organizations such as Cancer Research UK, Open Research Computation, Bioinformatics Training Network and FlyBase. All these folks have large image assets to discuss and are keen to move forward with an open solution to recommend on their personal networks for the long, long term public good.
I appreciate the image deletion issue, what we are talking about here are planned batch uploads of high quality donations. Part of that planning would be to discuss the relevance to the public of large number of research images and compliance with existing Commons guidelines. There may well be cases, for example many thousands of similar images of mutant drosophila, where Wikimedia Commons is not the right place for a full donation and a more specialized database host is needed.
Cheers, Fae -- http://enwp.org/user_talk:fae Guide to email tags: http://j.mp/faetags