[CQ-Contest] Log Checking Software
Ward Silver
hwardsil at centurytel.net
Wed Sep 17 20:26:16 EDT 2003
>>I agree it isn't a perfect piece of software and can improve. I try to
>>improve
>>it each year, but so far, there have been more important areas to focus
on.
>>
>>It seems to me that giving credit for properly-logged QSOs should be an
>>important enough area to get right.
Come now. If *anyone* can define "right" sufficiently well such that a 100%
iron-clad, won't-fail, catches-everything-misses-nothing algorithm can be
developed from the definition it will be a first.
Pushing beyond the point where true-positive and true-negative rates are in
the high-90% region is an very unreasonable expectation. I have worked with
cardiac diagnostic algorithms where a false-positive and false-negative rate
of one or two percent is considered excellent and took millions of dollars
and tens of thousands of man-hours. The fact, and it is a fact, that we can
send in all these different logs and have such consistent results should
speak for itself. The 8P1A log contained nearly 10k QSOs and the software
goofed on what, ten, twenty, thirty QSOs? Fifty QSOs would be a 0.5% error
rate. Um...that's good enough for W2SC and good enough for me. This isn't
missile interception you guys, get a grip.
The number of variations that exist once you have to start guessing at
intent or arbitrating between conflicting data (call versus exchange) is
very large and increases the number of parameters which the algorithm must
use to decide. This introduces error sources in and of itself. Once a
change to the cross-checking algorithm is made, then it has to be verified
as well. The more parameters used, the more time consuming the verification
process. The amount of work gets very large very fast for very little real
improvement.
Are there more important things? You bet there are. Catching cheaters,
reducing reporting turnaround time, making it easier for people to submit
logs, extending log checking to more contests, etc. etc. etc. Not worrying
about sub-1% minutia that can't even be defined except on a QSO-by-QSO
basis.
In addition, if the operator screws up or the logging software screws up and
records the wrong information why is there ANY expectation AT ALL that the
log checkers should be the ones to clean up the mess? Gee whiz...what a
thankless job! All done for free, by the way, in case there's any mistaken
impressions out there.
That's spelled Foxtrot, Romeo, Echo, Echo. When was the last time anyone
said Tango Hotel Alpha November Kilo Sierra?
73, Ward N0AX
More information about the CQ-Contest
mailing list