[Skimmertalk] Fwd: Re: Identifying CQ-ers

Mon Nov 17 17:24:20 EST 2008

Jim, I don't think Alex was offended by your report at all.  I think you 
have helped put your finger on an area that can use some improvement, if 
and when Alex can devote more time to Skimmer.

On the broader subject of Skimmer's score impact, I stand by my original 
view.  In SS, using an appropriate master.dta file and the paranoid 
validation setting, I think it offers real advantages for Sunday score 
optimization.  In CQWWCW, I expect it to be worth many multipliers as 
compared to skimmerless operation.  One of the truly nifty features is the 
way that N1MM compares an incoming spot with the call history file (in SS) 
and the country.dta file for CQWW.  That way you know exactly when a single 
or double mult comes on the air by the color coding on the bandmap and in 
the available window.  You can even set the Available window to only show 
multipliers.

73, Pete N4ZR

At 04:59 PM 11/17/2008, Jim Baremore wrote:
>Hi Pete, Alex, and the group.
>
>In answer to Alex's questions, Here is what I had in my report about the
>settings of CW Skimmer.  Maybe you missed them in focusing on the results.
>
>"I set CW Skimmer to be more aggressive in the software evaluation of the
>callsign as well as requiring it to only spot verified calls. I ran the
>Skimmer for about 13 hours and in that interval it decoded 13,671 call
>signs, a rather impressive 1050 raw spots per hour. "
>
>AND
>
>"Initially I clicked on the next report coming in when I was ready to test
>another spot.  The second half of the test I only clicked on the latest spot
>coming in which CW Skimmer reported as 'CQ'ing'. This was an attempt to find
>only the station running the frequency and not the station he was working.
>In both cases I was always clicking on the newest spot available although at
>times several spots would go by while I was verifying a previous spot.  The
>results of using either technique were the same within 1%."
>
>As far as recording what I heard, I did do that and have all 13 hours of
>that I/Q data recorded.  It's 11.4 Gigabytes so a little much to forward via
>e-mail HI HI  Unfortunately I did not 'log' the 360 call signs I tested in
>my evaluation but merely scored them and then moved to the next available
>call spot on the list.  By not logging, it kept my Q Test rate at 140/hr.
>However, one could take the recorded data and redo my test with any new
>criteria they wished.
>
>Although contests are 'fun' the first time, I'm not so sure a replay would
>be that much fun!!  But, I also have an excel spread sheet that shows all of
>the call signs I received as well if they were "CQ'ing" or not.  It also
>shows frequency and time text received from CW Skimmer as well as the time
>stamp put on it by the Linux Box.  If you would like to pursue that data, I
>will send it to either or both of you.
>
>Finally, since not everyone may have read my report but may just be
>following the results here, I would like to reprint some words I said in my
>postscript.
>
>"Although my opinion may seem to be making light of CW Skimmer, I think the
>work Alex has done with CW Skimmer is amazing!  I bought a registered copy
>in part the show my support of his work and to encourage his continuation of
>it."  My other comment was: "I believe the continued advances in technology,
>be they hardware or software, will continue to attract new talent to the
>hobby.  Barriers to technology advances, perceived or real, will only
>stagnate the hobby."
>
>Again, my appreciation to Alex and what his technology is doing for our
>hobby.
>
>73's all
>
>Jim  K5QQ
>
>PS  Alex, Did you find the link to my report?  I can't tell if you are on
>the reflector or not.
>
>
>
>
>-----Original Message-----
>From: skimmertalk-bounces at contesting.com
>[mailto:skimmertalk-bounces at contesting.com] On Behalf Of Pete Smith
>Sent: Monday, November 17, 2008 1:58 PM
>To: skimmertalk at contesting.com
>Subject: [Skimmertalk] Fwd: Re: Identifying CQ-ers
>
>Excerpts from an e-mail just received from Alex.
>
> >Delivered-To: n4zr at contesting.com
> >From: "Alex, VE3NEA" <alshovk at dxatlas.com>
> >To: "Pete Smith" <n4zr at contesting.com>
> >Subject: Re: Identifying CQ-ers
> >Date: Mon, 17 Nov 2008 10:31:04 -0500
> >
> >Hi Pete,
> >
> >
> >>Hi Alex - I don't know if you have been following the discussion on
> >>Skimmertalk. K5QQ ran some tests during CW Sweepstakes
>
>
> >>....K5QQ did not tell us what validation option he used - this setting
> >>has a significant effect on the identification of the running stations.
> >>For example, in the Minimal mode the call is spotted after its first
> >>occurrence if it is found in Master.dta. Consider this scenario:
> >
> >- N4ZR TEST
> >- VE3NEA
> >
> >Since VE3NEA is in Master.dta, and it appears immediately after TEST, it
> >may be spotted as running if Minimal validation is selected. This depends
> >on many factors, such as the time interval between TEST and VE3NEA, and
> >the number of times N4ZR was previously copied on the frequency, but the
> >errors will occur from time to time. The Minimal  mode is more appropriate
> >in the pileups than in a contest.
> >
> >If the running station always repeats the caller's callsign, as in SS,
> >this type of error may occur in other validation modes as well, since the
> >Skimmer decodes the callsign of the calling station twice.
> >
> >
> >
> >>I have always had the impression that if anything Skimmer tends to false
> >>negatives.  For example, if a rare station is running and signing his
> >>call but only sending "UP" or "TU" then Skimmer doesn't realize he's
>CQ-ing.
> >>
> >>I wonder if you can elaborate on the rules that Skimmer uses to
> >>identifying someone as CQ-ing.  I know about "CQ" and "TEST" and (I
> >>think) "QRZ", but what else is there.  Does Skimmer in any way look at
> >>callsigns identified on a given frequency within a given time period
> >>after one of the "tag" words?
> >
> >CW Skimmer looks at the keywords before and after the callsign to
> >determine its CQ/DE status. Here are the rules:
> >
> >  //forward: {CQ,QRZ,TEST, SS} [garbage] [DE] <call>
> >  //backward: <call> [garbage] {UP, TEST, SS}
> >
> >
> >{} = any of
> >[] = optional
> >
> >
> >Garbage is one or two words that may appear between the keyword and the
> >callsign, i.e., in "CQ DX DE <call>", "DX" is considered garbage.
> >
> >Note that pattern matching does not work across the pauses, so that in the
> >messages like "<run_call> TEST <pause> <sp_call>" there is no association
> >between TEST and <sp_call> if the pause is 0.5 s or longer.
> >
> >
> >
> >>One theory being discussed is that because the CW Sweepstakes exchange
> >>requires each station to give the callsigns of both stations, that both
> >>stations in any given QSO are being identified as CQers, but one is
> >>moving on.  Is that likely?
> >
> >We do not have enough information to tell what exactly happened in K5QQ's
> >test. If he made an I/Q recording while checking the spots, he would be
> >able to play it back and see what exactly was copied, why the callsign was
> >posted, and why it disappeared by the time he tuned to the frequency. I
> >hope in the future tests the people will make such recordings and will not
> >have to speculate on the results.
>
>
>
>_______________________________________________
>Skimmertalk mailing list
>Skimmertalk at contesting.com
>http://dayton.contesting.com/mailman/listinfo/skimmertalk