Re: Searchable Crypto Paper Archive?

[tcmay@got.net (Timothy C. May)]

At 12:48 PM 9/6/95, David J. Bianco wrote:

>I was trying to dig up some cryptography papers cited as references
>today, when a thought hit me; there seem to be a fair amount of crypto
>papers available on the Net, but they're pretty scattered.  Bell Labs
>has some online, which is great!  The cypherpunks FTP archive has a
>few, though you can't perform keyword searches against them.  In
>short, it's hard to find papers unless you already know what you want
>and where it might be.
>
>Having had some experience in designing and implementing technical
>report retrieval services, I naturally think there's room for
>improvement here. 8-) What I have in mind is something like NASA's
>NTRS ("NASA Technical Report Server",
><http://techreports.larc.nasa.gov/cgi-bin/NTRS>), which I helped
>design and implement at my last job.

It's an idea with some attraction. But some issues need discussing. Being
an analytical sort of person, prone to looking for flaws in ideas, I'll
mention a few:

1. First and foremost, _copyright_ issues. Most articles are copyrighted
(automatically, by Berne Convention) and the permission of the authors must
be obtained. Authors may also collect royalties, or the conferences may, so
unlimited electronic distribution is a potential problem.

NASA can publish its reports (and those of other government agencies)
electronically because it has the copyrights, or the copyrights are free
and clear. Try putting someone's article on the Net without their
permission and look out.

Indeed, there are a couple of the most important papers on the soda archive
site, some of them scanned-in and OCRed by "The Information Liberation
Front." There are so few that the authors likely don't even know they are
there, or care. But try to put lots of copyrighted material on a site and
get ready for actions. Remember, most nations are party to the Berne
Convention(s).

2. Many of the papers have complex typography, lots of equations and
diagrams. These reproduce poorly on most screens, and really need a new
level of display presentation. (Yes, I know about Adobe Acrobat, which I
have. Ditto for FrameMaker, and a few other such systems. But not many
others have them.)

I happen to know the ILF member who posted the Chaum "Dining
Cryptographers" paper, anonymously, and know that he picked that paper both
because of its importance to his interests and because it was pure text,
with no equations and no diagrams. This made it a natural for scanning.

3. In the crypto domain, the papers are much more conveniently concentrated
into a handful of conference proceedings, nearly all published by
Springer-Verlag. (Those great silvery-grey paperbacks.)

This point about Springer-Verlag relates to Item #1 above. Namely, that
copyright holders (Springer-Verlag, through publishing arrangements with
the conferences) will not take kindly to folks making the papers available
electronically.

This point, about the limited number of main crypto volumes, also implies
another point: many of these papers refer to other papers in the same
volume or set of volumes (e.g., papers in the "Crypto '93 Proceedings" will
refer to papers in that volume or earlier volumes). This makes it *even
more advantageous* for a serious researcher to buy the complete set of
volumes.

4. Authentication issues. Electronic versions of articles will need to be
signed, to prevent unauthorized modifications. The infrastructure for this
is beginning to build, but is clearly not available to many.

I am confident that someday most journals will be published electronically.
Many people think this likely, whether in 5 years or 15 years. Just too
many advantages.

However--and this is my point--before that happens a huge amount of
negotiation about author's rights to reproduction, about verification of
copies, about royalty payments for copies, etc., has to happen. And, the
display software/hardware is not quite there yet....too many people would
be unable to see the equations and diagrams on the screen. In 5 years, less
of a problem.

Many authors make their papers available by anonymous ftp, or via the Web.
I think this is the way to do it: let those who feel their papers need
electronic dissemination do so. The author makes the choice.

In summary, this project is probably premature (technologically), has
numerous copyright issues to be resolved, and is probably less needed in
the crypto community than in some other areas.

(Granted, we are not following those other areas, necessarily. But that
other domains have not yet gone fully electronic is indicative that others
see some of these same problems, and are likely to address them before the
math/crypto community does.)

Sorry to dissect this proposal so thoroughly, but it's one of the things I do.

--Tim May


(P.S. The copyright problems can possibly be skirted by using anonymous
remailers and offshore data havens in jurisdictions that will not raid the
sites, or by message pools. But these are major steps, mostly untested. A
"Scientology" site is probably a better test than a site with crypto
papers. I wouldn't want to run either of them.)



---------:---------:---------:---------:---------:---------:---------:----
Timothy C. May              | Crypto Anarchy: encryption, digital money,
tcmay@got.net  408-728-0152 | anonymous networks, digital pseudonyms, zero
Corralitos, CA              | knowledge, reputations, information markets,
Higher Power: 2^756839      | black markets, collapse of governments.
"National borders are just speed bumps on the information superhighway."