When I spoke with Franz Rothlauf (chair for GECCO 2009) about this, he
pointed out the steady decline, which surprised me.  Every year, I
count the coev papers in the proceedings and until this year, they
seemed relatively steady (around 12 or so).

What appears to have happened is that people submitted coev papers to
other tracks, of course.   Also, several of us with active interests
in coevolution research are in the same place in our careers:  very
early, post-PhD.  This means that conference submissions aren't
particularly appealing to us at present, and we haven't necessarily
had the opportunity to build a program sufficiently large to drag ...
er, encourage ... GRAs into the field.  Sevan, Anthony, Edwin, and I
had no coev papers in GECCO this year.  I haven't published a coev
paper in a conference since 2006.

But there is clear interest in coevolution.  The tutorials are always
well attended, as are (usually) the sessions.  The group is small but
relatively active (this past year being somewhat depressed).  And
there are a few new researchers with serious interest in
coevolutionary algorithms.

It makes me wonder:  What was the average coev researcher's view of
the track?  For those of us interested in coev theory, we have (as of
last year) an uncomfortable choice (the EC theory track or the coev
track).  For those interested in applications of coev, I suppose they
are similarly split between the general GA or RWA tracks and coev.
Perhaps there is no niche for coev (pun intended)?

How many people are there out there where coevolution is their primary
object of study (as opposed to, say, merely a technique they happen to
apply to a class of problems of interest)?