From: <HIDDEN>
Path: news.mathworks.com!newsfeed-00.mathworks.com!webx
Newsgroups: comp.soft-sys.matlab
Subject: Re: now that was interesting
Message-ID: <ef2fb57.154@webx.raydaftYaTP>
Date: Fri, 14 Apr 2006 09:59:54 -0400
References: <ef2fb57.133@webx.raydaftYaTP> <ef2fb57.134@webx.raydaftYaTP> <ef2fb57.135@webx.raydaftYaTP> <ef2fb57.136@webx.raydaftYaTP> <ef2fb57.137@webx.raydaftYaTP>
Lines: 23
NNTP-Posting-Host: 151.204.227.63
MIME-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
Xref: news.mathworks.com comp.soft-sys.matlab:344901


Ned Gulley wrote:

> Lots of good ideas here.

I think the contest team is getting a lot of good feedback from these
discussions. (Is "contest team member" a full-time position at
MathWorks, or do you all have to go back to your "day jobs" now, like
the contestants do?)

I'll add one more point, and I am not the first to say this. I think
that by far the most frustrating part of the contest is the CPU time
variability. Quantum mechanics dictates that there must be some
variability, but every effort to reduce it would mean more genuine
(but admittedly small) efficiency improvements would be recognized,
instead of being swamped by noise.

This would simultaneously help solve the queue flooding problem,
because a lot of "parameter tweaking" is not actually finding better
parameters, but rather just getting lucky on timing.

I would like to see the contest team focus all their efforts on this
problem, rather than fighting obfuscation, etc., which in my opinion
has only a minor impact on the contest.