[alma-config] dynamic scheduling, metric

[Min Yun]

An interesting paper appeared in the astro-ph this moring, "The
Number of Publications Used as a Metric of the NOAO WIYN Queue
Experiment" by Massey, Guerrieri, and Joyce (astro-ph/0001540).
These guys are trying to evaluate whether the dynamical queue
scheduling at the WIYN telescope is really more efficient and
has resulted in a higher scientific throughput than the classical
observing mode.  Beating to the punchline, their answer was NO
although you can judge for yourself how meaningful their analysis
is.  Nevertheless, it raises some of the interesting issues 
concerning theoretical "optimization" efforts confronted by
sociological (or real life) tendencies.  Also the issue of metrics.

Judging by the numbers of publications that resulted from the
queue observations versus classically scheduled time, these
authors find that the dynamic queue observations did not result
in any improvement in scientific output, at best.  This was
despite the excellent quality of the data delivered that users liked.
This contradicts 2.5 times improvement in publication rate 
predicted by a theoretical analysis.  Noteworthy results include:
1) target of opportunity observations did not have as much
impact as one may have expected; 2) advantage of dynamic scheduling is 
larger when the range of observing conditions is greater; and 3)
reduced sense of data ownership and data saturation may be 
important sociological causes.  

It is a short paper, so I recommend everyone to read it.  Clearly we
cannot apply this directly to our effort, but it certainly raises
issues with the incremental improvements we often argue over the
array efficiencies, etc.  This also makes an interesting contrast/parallel
to the ALMA/MMA memo on dynamic scheduling by M. Wright.  Lastly,
what a funny way to evaluate and quantify scientific throughput of
an instrument...  On the other hand, publication record is a major
evaluation tool for various funding agencies, at least in the US.
I was particularly amused by their view on the success of the HST
queue system and its apparent success -- money!!!  We should think
about this for the project....

Lastly, alma-config mailer has been collecting dust lately.  I hope
some progresses are being made by various people.  I urge you again
to volunteer to lead some of the topical discussions.  I will start
putting together the agenda for the Tucson meeting, and I don't want
people to be unhappy then.  Worse yet, I am going to push topics on
people if I don't get enough enthusiastic leaders.  Thanks.



					-- Min