[CQ-Contest] ENC: Report on WRTC 2006 log checking

[PY5EG]

Dear WRTC 2006 Friends:

This report is also available on our page for download as Release 059/06



This memorandum is to record an anomaly discovered by the Judging and
Log-Checking Committees in the course of adjudication of the WRTC 2006
competition and the action taken as a result, and to offer a recommendation
for future WRTCs.

Immediately after the end of the competition at 1200 UTC on 9 July,
electronic logs began arriving from participants in the IARU HF World
Championship from all over the world for inclusion in the master database.
Thanks to the extraordinary cooperation of the international contesting
community, in a matter of hours we received more than 1,000 logs for use in
adjudication of the 46 WRTC logs.

When initial trial discrepancy reports for the 46 WRTC stations were run,
four logs were found to have unusually high numbers of "unique" QSOs. (A
"unique" QSO is one with a station that appears in few other logs - it is
not literally unique and the appropriate threshold for labeling a QSO as
"unique" may vary from one contest to another. In the case of WRTC 2006
there were, as noted above, more than 1,000 logs used to determine "uniques"
- that is, we did not limit our comparison only to other WRTC logs.)

In three of the four cases the reason for the high number of "uniques" was
clearly that the nationality of the WRTC station had been identified and
communicated within the home country of the team. The three logs in this
category were PW5B (SP7GIQ and SP2FAX), PT5P (DL6FBL and DL2CC), and PT5N
(9A8A and 9A5K). In some cases, multiple QSOs clearly had been made by the
same station under different call signs. This is contrary to the spirit of
WRTC, which is to provide a level playing field for all. QSOs that are
available to one team but not to any of the others tilt the playing field.
The purpose of WRTC is to determine the best operators, not who has the most
friends back home. At the same time, it must be said that the WRTC
competitors do not do anything wrong unless they do something that is
intended to identify themselves. They are not expected to disguise their
voices or their accents. So, in past WRTCs these "national uniques" have not
been deleted from competitors' logs.

The fourth case, PT5L (YT6A and YT6T), was quite different. In this case
there were hundreds of "uniques" with call signs from many countries, nearly
all of them in Europe, beginning about two hours into the event. Not only
that, but many of these "uniques" showed up on several bands, in some cases
on both CW and SSB. Only rarely did these "uniques" show up in any other
logs. The referee, G3XTT, confirmed that he had heard nothing suspicious or
unusual. Most of the call signs were not familiar to the judges as appearing
in CW contests, but this was apparent only when the "uniques" were
identified as such and this was not especially noticeable when they were
scattered among valid QSOs.

The audio from this station had been recorded, so it was possible to confirm
that the QSOs had in fact taken place. It appeared to the judges, from
listening to the recording while examining annotated log extracts
identifying the "uniques," that there was a small number of stations,
probably more than one, feeding "phantom QSOs" to PT5L. Again, however, this
was apparent only in retrospect.

The motive for feeding these "phantom QSOs" was and is unknown. PT5L had
only operated CW up to the time the "uniques" began to appear; we have no
evidence that the identity of the operators had been communicated to anyone
prior to that time, either by the operators themselves or by anyone else.

Identifying "uniques" is a standard part of the adjudication of any contest.
Detecting QSOs added to a log "from the Callbook" was one of the very first
objectives when computer analysis was first performed on contest logs.
Anyone familiar with UBN or LCR reports - in other words, any active
contester - would realize that adding such QSOs to the PT5L log was bound to
come to the attention of the judges. Therefore, in the absence of any
evidence to the contrary, the judges could not conclude that the operators
of PT5L were implicated in any way. They had to know that such manipulation
would be detected and that there could be no acceptable explanation for so
many people choosing, independently of one another, to work only PT5L from
among more than 1,000 stations active on the bands during the contest.

It seems much more likely that the "phantom QSOs" were intended to be
sabotage, either of the PT5L operators specifically (although this would
have required knowing who they were) or of a WRTC station selected at
random. Such behavior, by amateur operators outside the WRTC event itself,
is both reprehensible and illegal and deserves to be thoroughly
investigated. However, doing so within the time frame of the WRTC event was
impossible.

Thus, the problem facing the judges was how to deal appropriately with the
situation within the context of WRTC itself. It was decided to reduce the
threshold for labeling a QSO as "unique" to a relatively low number of logs
and to delete from all 46 logs, without penalty, all "unique" QSOs. Doing so
resulted in the deletion of, at one extreme, just 3 QSOs from a log
containing more than 1,900 QSOs to, at the other extreme, the deletion of
240 QSOs from the PT5L log of more than 2,300 QSOs. Most logs lost about 15
QSOs. No doubt a few of these were valid QSOs (others are busted calls where
there is insufficient evidence to label them as busted). However, deleting
them is very unlikely to change the relative positions of any teams with
"normal" logs and definitely did not affect the top three positions, which
is the principal concern of the judges, other than to move PT5L out of 3rd
place. PT5L moved from 3rd to 11th place, some of this movement the
consequence of error rate. (If the intent of the "phantoms" was to hurt the
PT5L entry they certainly succeeded; the operators spent more than two hours
working stations for which they received no credit.)

The impact was much greater on the three logs with significant numbers of
"national uniques" (which were also a very minor factor in the PT5L log -
among the 240 "uniques" are just 9 "national uniques" that showed up very
late in the contest after the operators were identified on the BalkanDX
Yahoo forum at 1024 UTC). PW5B lost 125 "uniques" and moved down five
positions, from 11th to 16th, some of this movement the consequence of error
rate. PT5P lost 76 "unique" QSOs but actually moved up one position in the
final results because their error rate was lower than average. PT5N lost 67
"uniques" and moved down one position.

While it is unfortunate that the operators of these four stations all spent
time making QSOs for which they received no credit, the judges and
log-checkers firmly believe that the appropriate remedy was applied to
correct an unanticipated situation. Disqualification would have been
inappropriate in that there is no proof - indeed, there is no evidence at
all - that any of the operators were at fault. On the other hand, allowing
"phantom QSOs," once detected as such, to be credited also would be
inappropriate and would make the whole WRTC event pointless. The same
standard was applied to all 46 logs.

The judges and log-checkers recommend that in future WRTCs, all "uniques"
should be removed from logs without penalty and that this policy should be
announced in advance. All stations operating in the IARU HF World
Championship should be encouraged to work all WRTC stations, without
favoritism and without using more than one call sign from any station
(unless family members share a station).

We also recommend that "unique" not be precisely defined in advance. The
judges must be allowed discretion to determine the appropriate threshold for
"uniques" for a given event, as this will depend upon the number of logs
available for crosschecking and other factors.

While we regret that this attempt to manipulate the WRTC results occurred -
for whatever reason the perpetrators may have had - we are pleased to have
detected it and to have been able to take appropriate corrective action.

The judges would especially like to thank Larry "Tree" Tyree, N6TR, for his
analysis and Phil Goetz, N6ZZ for his invaluable contribution in uploading
the WRTC logs and maintaining the communication link between the judges in
Florianopolis and Tree at his home in Oregon. Trey Garlough, N5KO, also
contributed significantly to the log-checking process from his home in
California, as did Larry Hammel, K5OT, on-site. 

Respectfully submitted,

Judges:					Log-Checking Committee:
David Sumner, K1ZZ				Larry "Tree" Tyree, N6TR
Roger Western, G3SXW			Phil Goetz, N6ZZ 

WRTC2006 Steering Committee
Atilano de Oms PY5EG

-- 
Esta mensagem foi verificada pelo sistema de antivírus e
 acredita-se estar livre de perigo.