[CQ-Contest] Log checking errors
Tree
tree at kkn.net
Sat Jul 19 11:54:18 EDT 2008
Thought I would share my perspective about log checking errors. As most
of you know, I have been doing computer log checking now for some 20 years.
I don't think it is a secret that a highly automated process is going to
have the following attributes:
1. It can do a LOT of log checking quickly.
2. It will follow some basic rules in how logs are handled.
3. Will be affected by "garbage in" data.
4. Improves with age.
The CQ WW log checking process has gone through several major changes over
the years. It is going through another one this year - which will have
some new issues not seen before - but will improve with age.
Feedback from all of you to the log checkers is very valuable for finding
cases where the log checking process can be improved - especially in the
area of detecting garbage data (like someone entering the wrong call in
their Cabrillo log). As the process matures - the error sources that
can be detected will be found and they will not occur again.
For example, the author of the current software has already added new
features to the process to detect the current problems being discussed
on this reflector (and some that are not being discussed here as well).
My final point is to reassure everyone that not ALL of the errors in the
process will be fixed. There will be times that you end up with a NIL or
a busted call when it wasn't deserved. This is just the nature of the
beast. However, we believe that this doesn't happen often enough to
really invalidate any of the results.
Most of the decisions that the log checking software make are biased
towards not penalizing you if there isn't strong data to indicate that
a QSO should be busted. Occasionally, there are times where a number
of small variables add up to making a QSO look like it should be busted
when it really shoudn't - but statisically, these do not happen very
often.
It is easy for these cases to get a lot of exposure here and make it
appear like the sky is falling. I hope that most of you can see the
big picture and realize that these cases are just the price to be
paid for having ALL of the logs checked using the same process.
I feel that is the most fair way to check them.
73 Tree N6TR
More information about the CQ-Contest
mailing list