The first problem is bibliographic. About 20 of my 75 citations do not
have any proper print reference, they are documents that are only on
the web. Now I can cite these fine (hey, I even hacked a bibtex style
to accept a "url" field and do the right thing), but they really
aren't very good citations. Most of them will be meaningless in a year
because the page will have moved. How do you cite a moving target?
There's been a lot written about this problem, including a nice post
to Risks recently. There's nothing approaching a solution.
There's also a problem with submission. All MIT wants is copies of the
paper on acid free paper (as Phil Greenspun says, "real dead trees!").
No place to give anyone postscript, no possibility for an electronic
copy. Not even any good way to submit a copy of the source code for
the program. They will take magnetic media and try to archive it, but
floppy disks and tapes are awfully unstable.
The funniest problem has to do with signatures. Theses here have to be
signed by three readers. A friend of mine's readers are scattered all
over the world. They won't take FAXed signatures, so he has to
chain-FedEx pieces of paper around the globe to get everything signed.
The Web and email has made for all sorts of amazing transmission of
information, documents, ideas. But academia hasn't caught up yet, or
else the net isn't good enough.