From: srach <s.rach-nospam-@iu-bremen.de>
Path: news.mathworks.com!newsfeed-00.mathworks.com!webx
Newsgroups: comp.soft-sys.matlab
Subject: Re: now that was interesting
Message-ID: <ef2fb57.140@webx.raydaftYaTP>
Date: Thu, 13 Apr 2006 15:54:29 -0400
References: <ef2fb57.133@webx.raydaftYaTP> <ef2fb57.134@webx.raydaftYaTP> <ef2fb57.135@webx.raydaftYaTP> <ef2fb57.136@webx.raydaftYaTP> <ef2fb57.137@webx.raydaftYaTP>
Lines: 29
NNTP-Posting-Host: 134.106.76.159
MIME-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
Xref: news.mathworks.com comp.soft-sys.matlab:344797


In my opinion the best way to stop people stop spamming the queue was
suggested by Matthew in comment to the blog:

\begin{quote}
 We were kicking around the idea of only scoring one entry per
contestant on each pass through the queue. That is, we’d only score
one entry from each author until everyone waiting has had one of
their entry scored. The drawbacks are that it is a little tough to
explain and it would encourage contestants to submit under different
names.
\end{quote}

The possible drawback he sees might be eliminated by awarding an
additional price to the contestant who makes the biggest cumulative
improvement to the top score over the whole curse of the contest.
(However, it might be difficult to find a way of weighting
improvements relatively to contest phase - improvements are much
greater at the beginning of the contest.)

Submitting entries under different names would decrease ones chances
to win such an award.

Another good idea I've read somewhere here is to run entries multiple
times instead of one single run and to score the mean performance.
This would make spamming identical unneccessary at least for entries
depending on random numbers.

Finally I would like to thank the contest team as well as the other
contestants - this contest (actually, my first) was really fun!