From the sporting code:

15. 4.5 End of Speed Section

Where the point being claimed is the end of a speed section and the track-log
has 2 points either side of the sector boundary at most 30 seconds apart, then
the finish time will be interpolated from these points (constant speed being
assumed). Otherwise a finish time is taken from the first (in time) point
outside the boundary of the speed section; if the end of the speed section is a
goal sector then this will be the first (in time) point within the goal sector.

15. 5.2 Minimum Track-log Evidence

The verification software must confirm that all points used to verify the flight
occurred at reasonable times (e.g. on the day in question, between the start of
the task and the end of the task, and showing the correct chronology of start
and turn points).

15. 5.9 Rejection of Track Log

The competition organiser has discretion to reject any track log, or part
thereof if he/she feels it does not show sufficient evidence that the claimed
data is genuine. In such cases the pilot is to be awarded zero points for the
round.

The Sporting Code does not provide the scorekeeper with much
guidance about how to deal with track logs that have position errors in them. As
a scorekeeper I have seen many GPS track logs that have what in my judgment are
obvious position errors and I discount those track log data points. Scott
Barrett's track log going into goal on the first day of the Worlds presents
interesting data interpretation issues that are not normally encountered. The
question becomes which track log data points are "valid" and which are not.

In addition, can track log data points that have obvious positional errors still
contain valuable data that can help one determine the actual track of the pilot?
What evidence can the scorekeeper use to help him make a judgment to determine
which track log data points have validity and which do not?

Are witnesses useful in making this judgment call and is it valid to use
statements from witnesses? Can the scorekeeper attribute to the pilot his
implied intent to get to the goal cylinder directly and as soon as possible as a
way of judging which track log data points are valid or partially valid?

It is clear to me that there is no automated way to determine the validity of
any of the track log data points and that it is up to the scorekeeper to
determine which points he considers valid and which he doesn't.

Regarding Scott's track log near the goal:

It was the judgment of the scorekeeper that the tracklog started containing
invalid position data after 17:58:10 as the calculated airspeed to the next
point was 293 mph. In addition, the jiggles in the glide path were not what you
would expect from a pilot gliding into goal. Also, this jiggle pattern
reminded the scorekeeper of a problem where the GPS is blocked by the pilot's
body (something that happens when flying fast into goal) and the positional
information gets unreliable.

So the scorekeeper relied on his judgment to determine that the track log data
was not to be taken literally as defining the pilot's actual position using
information that he knew about the pilot's circumstances and information that
was not necessarily contained with in the track log data.

But the question becomes, which of the track log data points are invalid, and
which may contain actually correct position information as they do appear to
jiggle around a presumed course line into the goal cylinder.

The points time stamped at 17:58:25 and 17:58:27 seem reasonable looking at the
implied speed, direction and the decrease in altitude. Also the points time
stamped at 17:48:47 and 17:58:48.

The points time stamped at 17:59:06 through 17:59:33 also appear to be
reasonable looking at implied speed, direction and decrease in altitude, but
they are as a group significantly displaced twenty kilometers to the south. We
don't know the mechanism in the Garmin GPS/satellite interaction that would
cause this displacement but then display reasonable looking data points.

Using a reasonable looking data point time stamped at 17:59:01 (direction,
position, and altitude) and moving the displaced points north 20 km indicates
that the pilot crossed the goal cylinder before 17:59:25. This would be
consistent with witnesses that Scott finished ahead of Christian who finished at
17:59:26.

There are clearly at least two track log data points, one inside the goal
cylinder, and at least one outside the cylinder that are less than or equal to
thirty seconds apart. The previous track log data point is 2 seconds before the
data point inside the circle, but it is the judgment of the scorekeeper that
this is not a valid data point, i.e. it doesn't represent the pilot's position.

It is 1.22 km away from the point inside the goal cylinder and that distance is
indicated to be covered in 2 seconds. Not possible.

All the other points within 30 seconds of 17:59:38 are displaced twenty
kilometers. The last reasonable point outside the goal cylinder is time stamped
at 17:59:01. Seven seconds outside the thirty second limit.

Therefore the scorekeeper would choose the first point displayed inside the goal
cylinder as per the rule above.

An interpolation between the 17:59:01 point and the 17:59:38 point (a distance
of .9km) would indicate a goal crossing time of 17:59:26.

I have used my best thinking and best judgment and the track log data points to
come up with my best estimate of when Scott crossed into the goal cylinder. The
scorekeeper is restricted by the Sporting Code to make use a data point that in
my judgment is not as good a reflection of reality as the calculation that I
have made.

Now it is quite reasonable to state that it is the pilot's responsibility to
deliver a track log that doesn't cause one to use this amount of thinking to
determine when he crossed into the goal cylinder, but I would suggest that it is
my experience that I often have to use my judgment as a scorekeeper to determine
what best reflects reality. Usually it is a lot easier than this.