UBA log checking

Although I don’t receive any emails from CQ-Contest anymore (hell no), I can’t hold myself from skimming (pun intended!) the online archives for possible interesting topics. This week my UBA log checking partner ON7SS posted a plea for checklogs which, in true CQ-Contest tradition, triggered some commotion. I tried to reply (although I’m not a subscriber) but after 36 hours, the post seems to be trapped too (again, pun intended  🙂 )  or simply rejected by the moderators. Anyway, here it is:

I’m not a subscriber anymore so I hope the moderator lets this pass? TNX.

Being half of the UBA contest log checking process and reporting (ON7SS is the other 50%), I’d like to add the following.

The UBA is a nice contest, and being a passionate contester and an ON, I’d hate to see the popular contest fade away. So together with Marc ON7SS, I decided to try to keep this contest shining. We took over the log checking in 2009, the year when I started writing our own log checking software. No idea how the logs were checked before that. I can’t even tell who did it. But I thought it could be better.

First thing we did find out is that some people try to cheat. Big time. The claimed winning log in 2009 (from Ukraine) had well over 50% uniques, of which a big number of 10pt ON calls never been issued. And almost none of the other calls were known to QRZ. Of course this log got DQ’ed. The two years following that, our software did not reveal any cheating of that degree.

What does the log checking actually show then?
– Many typos! RT3 becomes TR3, HB9 logged as OA9, a JA that suddenly becomes a Dutchman…
– UTC seems to be a problem as well as an accurate track of time
– no CAT = plenty of QSO logged on a wrong band
– People sending wrong exchanges (yes, sending)
– People copying wrong exchanges
– People cutting out dozens of QSO out of their ‘all band’ log to submit Single Band. DON’T DO THAT!
– People using software that is NOT suited for the job, or misconfiguring their good software to produce a mutant form of Cabrillo
– People submitting a log for a call they didn’t actually use in the contest, resulting in 100% NIL (mostly result of the above)
– People mutilating their Cabrillo by hand, rendering it completely useless (or not processable)
– But most of all honnest and sincere operators logging the right call with the right exchange, or at least trying to.

Moral of the story A: if people know a log will be thoroughly checked, hardcore cheaters will stay away.
Moral of the story B: I spend 90% of my time trying to solve problems caused by 1% of the dreaded non-standard Cabrillo logs.

Log checking is an easy job if you have the logs of both calls in the contact. Every log checker’s job would be a lot easier if he has (almost) all of the logs for each participant. So it is VERY important that as many
people as possible in the contest submit a log. Hence ON7SS’ call for checklogs.

But what if you only have one log? If the other call appears plenty, it probably is a valid call. If it only appears a few times, it might be a busted call. You can check that. For the UBA contest, if it appears a few
times and it’s not a bust for another call, it IS VALID. But what if a call only appears exactly once and cannot be traced to a busted call? This is the infamous UNIQUE.

You could say: “all uniques must be cut”. Fair if applied to everyone, right? I’ll always remember KL9A’s view (hi there Mr. C) on removing uniques in Russian DX. He said: many people called me for the KL7 DXCC, not making any other QSO. KL9A would lose all these contacts. Valid, honnest contacts coldly removed in the log checking process. Haven’t heard KL9A in RDXC lately… Exactly what P43E points out too: they get called by DX’ers who never submit logs. Uniques à gogo.
I had a similar thing with one of my logs in ‘my own’ log checking: a WA2 called me on 20m SSB, his call never appeared again making it a true UNIQUE but I got a direct QSL card: ‘TNX for Belgium, new one’. The QSO could be removed because of its uniqueness, although I had the proof in hand that it was a valid QSO. That made me wonder.

We have discussed this a few times and there is no fair way to treat a real unique, nor an easy one.
Our philosophy is: the claimer gets the benefit of the doubt, we prefer to let a bogus QSO pass rather than to remove a valid QSO. After all, before a call is labeled as a unique, we have checked for busts and errors to eliminate that. We DO count the number of uniques and a log gets flagged for utter scrutinizing if it has a more than average number of uniques. Which HAS NOT been the case after 2009 in the UBA DX contests.
The only criterium we handle for crediting or removing a unique, is callsign databases. We use QRZ.com (since it has almost all US calls, many of the uniques in UBA are American callsigns) and the offical Belgian callsign database. I guess it’s safe to say that if a call in the contest is not on QRZ, it *probably* is an invalid or inactive call. Uniques are a minority of which callsigns unknown to QRZ are even a smaller minority. I know that this is not 100% water proof, but this small relative uncertainty seems more fair than removing each and every unique or using a threshold like ‘a call must appear at least 3 times or it is removed’. I tried the latter but there are many calls that only appear two times so there you go: even more chance of
removed valid contacts. No go.

Some numbers for UBA CW 2011, with a total of 1102 accepted logs and a total of 314 017 QSO:
94,69% are valid and credited QSO
0.16% valid and credited uniques (NOT removed, and not counted above)
0.81% invalid and removed contacts, inlcuding rubbish calls that cannot be traced to a bust
1.21% Bad call (busted)
2.30% Bad serial
0.36% Not In Log
0.33% Dupes
The rest is logged on wrong band, wrong ON province.

I hereby hope that the AB7E interpretation that “they don’t only remove Uniques for the UBA … they intend to remove contacts for any callsign that they cannot verify via any other independent database.” has been proven wrong. The only QSO that is rejected is the pure unique that we can’t trace to be a real active call. That’s the 0.81%.

Checking the logs is hard work, and time consuming. Writing the software is too but that is the fun part (for an amateur-coder) ;o)
I have learned a lot about contesting in the past 2 years by seeing a lot errors and finding a pattern that leads to the error. No we don’t catch all errors. Yes we probably make a small number of mistakes. But we need to live with that. Both as a contester as well as when wearing my log checker’s hat. Our goal is to assure you that if you end up #2 in the UBA contest, the #1 will have made slightly more genuine and valid contacts than you. And to be #1, you need a clean log with not too much rubbish removed.

That said, 2011 was a record year for both of the UBA DX contests (SSB + CW). Participation noticably increased from outside Europe. USA, JA and PY logs are on the rise. Even Oceania pops up in the logs. You can now download your own certificates from the UBA website, and all time high scores are updated.
We can try to report and provide feedback, but the success of a contests depends on the participation. So we sincerely hope you’ll all join us again in 2012?

73 de Franki ON5ZO = OQ5M
log checker for UBA DX Contests