[CQ-Contest] Faster results for CW SS?

[Tree]

Bill, WB0O asks:

>   The question is: why does it take until June to get
> the results? The entire contest season will be over,
> and it may (or may not) be summer here, and my mind
> will be a million miles from contests.

I think it is easy to understand why this takes as long as it does.  First,
it takes about 2 months to get all of the logs, and fix some of the problem
logs up (those that are missing information or logged the contest in local
time or sent in the previous year's log).  Then, the log crunching has to
occur, and in the past few years, there has been work put into improving 
the capabilities of that process at the same time.  Once the numbers are
generated, then the writeup needs to be written and folded into the schedule
of the two people in the contest department at the ARRL.  Then, finally, it
goes off to the printers and gets mailed out using the slow mail.

And that doens't mention the paper logs that have to be dealt with manually
(still about 10 percent).

Many things are changing that can reduce this time and already this has 
happened.  The robots are helping clean up problem logs in real time.
Publication of the results on the web eliminates the printing and 
mailing delay.  The log checking process is near "perfection" and requires
less development in parallel.  It might be very possible to have "final"
results for all stations within two months of the contest.

The ARRL has come up with a new schedule for results, and things are moving
up about 6 weeks for this year I understand.

WB0O goes on to wonder about uniques:

>   The question of what to do with uniques. This is
> probably meat for a years worth of discussion. IMHO,
> it seems that there should be NO uniques among the
> top 100 or so scorers in the CWSS contest. 

This really isn't an issue in my opinion.  First off, uniques are only
a problem if a log has a lot of them.  All logs are going to have a few,
but when a log shows up with 10X the average (that aren't matched to 
busted calls), then this raises suspicions.  

For sure, any uniques a top scoring station has will be looked at.  There
have been some cases where it was obvious that "friends" got carried away
using multiple callsigns.  These QSOs were not removed, but in most cases,
these friends were encouraged to stick with one callsign, and to work some
other stations in the contest.

But - just to be clear - no uniques are removed without being judged to
be a busted callsign.

K4SB chimes in:

> I've gotten dinged for these "uniques" for the last 2 years. And
> afraid I sent a nasty
> email the last time. If it happens again, I'm going to send ARRL the
> QSL card!

I am not sure this happened in the SS?  It takes a lot of unprobable 
events to occur for any unique to be "dinged".  Please provide specific
data if this is the case - as I would be intersted in knowing more about
it.  Uniques are not removed from any log without being identified as
a busted callsign.  This applies to all contests I am aware of (except
the Internet Sprints and Keyman Club of Japan contests which require
both logs to show the QSO to be counted).

If you are concerned about uniques being listed, then please understand
this is a tool to detect people that have way too many uniques, a sign
of manufactured QSOs.  It also flags possible bad QSOs that can later be
judged to be busted either by the software or a human.

Finally, WB0O asks for my opinion (a dangerous thing):

> The results would be posted in a week. Tree, your comments? 

I would not want to make the call on who makes the top ten in a tight race
without having all of the information available.  Part of the log checking
process involves cross checking the QSOs with the logs we have.  This will
detect not-in-log situations, as well as incorrect serial numbers.  

If we did this before all of the logs were in, the stations with higher
error rates will look better than they should.  For each NIL not detected,
there is a 2 QSO impact.  Thus, an inaccurate log will be more competitive
without all of the logs being used in the process.

Therefore, I would suggest that if you wanted even faster results than 
supported by the log submission deadline, you work to move up the log
submission deadline for all logs.

Finally, AA4GA asks an interesting question:

> How would deleting uniques make the reporting process faster?

It makes it a LOT easier - because uniques are much easier to detect than
busted calls.  You have no judgement involved at all.  A unique is simply
a callsign that nobody else worked.  It takes a lot more work and time to 
match these up to busted calls.  In a contest like the SS, that is pretty
easy because of the "quality" of the exchage (lots of good information that
can be used to help you match up the call).  For other contests, like the 
ARRL DX and CQ WW, where the exchange is pretty useless, it is a lot harder.
These contests would be about 100X times easier to check if uniques were
just not counted.

Please understand I am not proposing this...  I am just answering the 
question Lee asked.

Tree N6TR
n6tr at arrl.org