Re: [TUHS] How to Kill a Technical Conference

["Theodore Ts'o" <tytso@mit.edu>]

On Mon, Apr 05, 2021 at 03:30:25PM -0400, Clem Cole wrote:
> 
> The issue for the USENIX BOD had always been getting butts in the seats to
> pay the bills.   Academics had a path, so following what IEEE and ACM did
> was well-trod and understood.    Get the more pure hackers like yourself,
> even the engineers inside of places like Sun or Masscomp was often
> difficult and we know was not well understood.  USENIX was hardly the only
> firm/org that made some bets in those days that in the long run might not
> have been the best.

In some ways, this was in large respect a great example of the
innovator's dilemma.  In the mid '90's, before the .com crash of '98,
Unix sales were starting to trend down, but they were still quite
respectable, and while Linux had a huge amount of buzz, a lot of that
was because Linux was cheap as in beer --- which means that when it
came down to the trade show floor revenues at an ATC conference event,
Unix was still bringing in healthy amounts of $$$ at least as compared
with Linux.

And similarly, there were still some --- but not as many as before,
during the "Golden Age" --- decent industry papers getting published
at ATC.  I sometimes suspect (although I have no proof of this) that
one of the reasons why ATC papers were not double blinded ala FAST was
because if there was an important paper from Sun, even if it didn't
meet the high standards of a tenure-track conference, it was desirable
to let a few such papers "slip through".  For the most part, those
industry papers were still better (from the let's not embarass the
tenure-track professors) perspective than a typical Linux paper, but
papers which were *really* not written to the academic style would
certainly either get rejected, or there would be strong pressure
placed on the writers to make them seem more "academic".

Ultimately, ATC (with or without its USELINUX track) was trying to
serve two masters, and while there were some techniques such as having
invited talks track which worked at least moderately well, at the end
of the day, if one of the goals of an attendee who happens to be a
Linux developer is to be able to meet a critical mass of other Linux
developers, it was always going to be hard for ATC to be able to serve
that need as well as the needs of Usenix's other constiuents --- which
were, after all, far more established within Usenix.

      	    	     	  	      	     - Ted