[Cryptography] "Public Accountability vs. Secret Laws: Can They Coexist?"

[Jerry Leichter]

>> It's not a lack of common sense, it's what you need to do to get a paper
>> published.  If you have a neat idea you can't just publish it because it's
>> a neat idea, you have to have some problem for it to solve.  So you invent
>> a problem and then publish your idea as the solution to it [0]
> 
> I've not generally seen this be the case in mathematics. A lot of
> subfields do not have any direct application. You can formulate
> solutions like "find X in Y better than Z" but maybe X and Y haven't
> been linked to any physical phenomena.
Fields of inquiry define their own standards for what makes an acceptable paper.  Mathematics long ago separated itself from practical applications.  (Though there have been some partial exceptions.  During the Soviet era, there was a push to show that work could actually be applied.  I recall once seeing a translation of a Russian text on, I think, functional analysis.  This is highly theoretical stuff - and most of the book was as well.  But then the last couple of chapters applied some of the theory to engineering analyses of concrete shells.  How widespread this kind of thing was, I don't know - but there was some really good mathematics done in the SU in those days.)

There's a whole field of "applied mathematics" (which overlaps with "numerical analysis" in uncertain ways) in which work is expected to have practical application - though of course sometimes that's a stretch anyway.  (I know of at least one respected practitioner who refuses to call himself an "applied mathematician" because he thinks most of the work done under that rubric is neither usefully applied nor good mathematics.)

Cryptography as a field started off heavy on the practical applications - but any field with a strong mathematical component will inevitably feel a pull toward abstraction which will result in papers judged on some mathematical notion of elegance rather than practical application.  The same occurs, of course, throughout theoretical computer science.

Two stories, both from many years back:
1.  I knew a bunch of guys working on functional languages.  They were working on some algorithm for a compiler - I think some kind of type resolution - and were very happy one day to announce that they'd gotten the algorithm down to only doubly exponential time complexity.
2.  At some STOC or FOCS, there was a "rump session" talk announced with the title "Application of VLSI Theory to VLSI Practice".  The speaker came up, put up a slide with that title, stood for a bit, moved to his next slide - which was completely blank; stood for a bit more; then bowed, thanked the audience and sat down.  Much laughter and applause ensued.

                                                        -- Jerry