History of Contest Scoring

Recent discussions about contest scoring have shown that it might be
beneficial to review the history of contest scoring in the U.S.
Here is a very brief overview.

The first U.S. National Soaring Championship was held at Elmira, New
York, seventy-three years ago. Three tasks were used: duration,
distance, and gain of altitude. The pilot performance measures for
these three tasks were, respectively, minutes, miles, and feet of
altitude. Because soaring contests are multiday contests, the daily
performance measures must be added together for a cumulative score.
However, minutes of duration, miles of distance, and feet of altitude
cannot be added together. No common unit exists for the addition. This
problem was addressed by awarding 1000 points to the daily winner
regardless of the type of task and awarding points to the other pilots
in a proportional manner. The daily points were then added together
for a cumulative score. This is the 1000-point system with which we
all are familiar.

Much has changed in the last seventy-three years since the first
contest. Duration, distance, and altitude tasks are no longer used.
Soaring has matured into a racing-only sport. The pilot performance
measure with the Time Distance Task is distance (see my posting on
r.a.s. 9/23/03). The performance measure is the same each day -
distance. The daily results now can be added together directly for a
cumulative score. It is no longer necessary to assign points

In 1999, I wrote a paper analyzing scoring systems. It was accepted by
OSTIV and presented at the XXVI OSTIV Congress in Bayreuth, Germany.
It was published in the OSTIV Journal, Technical Soaring, and in the
Soaring Association of Canada Journal, free flight. The paper proved
that 1000-point scoring systems produce scores that do not accurately
represent the actual, measured performances of the pilots. Simply, the
scores are not accurate. No one has challenged the conclusion of that
paper. The conclusion is not surprising considering that 1000-point
systems were designed to score duration, distance, and altitude tasks
and are now being used to score races. I will send a copy of the paper
to anyone who emails me for one.

Scoring on distance eliminates the three major sources of inaccuracies
in 1000-point scoring. The first inaccuracy is that pilots' scores
depend on the performances of their competitors rather than just on
their own performances. This is caused by dividing the pilots' speed
by the winner's speed to assign points. The number of pilots who land
out also affects the pilots' scores. These calculations are not done
in distance scoring.

The second inaccuracy is caused by assigning the same score value to
each day regardless of the length of the tasks. For example, 1000
points may be assigned to a two-hour flight on one day and to a
four-hour flight on the next day. However, two does not equal four.
Trying to "Make" two equal to four is mathematically incorrect and
causes scoring inaccuracies. This is not done in distance scoring. The
actual distances attained each day are scored exactly has they happen.

The third inaccuracy is caused by the assignment of an arbitrary value
for distance for the pilots who land out. This problem is eliminated
completely with distance scoring. The finishers and the land outs are
scored on the same dimension - distance.

In racing world wide, a course is set, and the champion is the
competitor with the lowest elapsed time. By calculation, the champion
also has the highest speed.

The same result is achieved with distance scoring. A fixed time for
the race is set rather than a fixed course. The champion is the pilot
with the greatest distance. By calculation, he also has the highest
speed.

This is a unique time in the history of soaring. The current
combination of GPS recorders and the Time Distance Task gives the
soaring community a fantastic opportunity to move from a seventy-three
year old system to an accurate, simple, understandable, and uniform
scoring system based on distance only. The soaring community owes a
debt of gratitude to the Canadians for their leadership in being the
first to seize this new opportunity and to provide the development
that a new system requires.

Bill Feldbaumer 09

"John Cochrane" wrote in message
om...
Here we go again.

And what do you do about landouts? How do you avoid the incentive to
dive to 100 feet as time ends? What happens to the "beauty and
simplicity" of the scoring system as we add bonus points for making it
back to the field, so you have now to make a sophisticated calculation
about whether to drift to the downwind edge of the task area, or make
it back? Or when the optimal strategy is to time out 20 miles downwind
at 500 feet, then see if you can limp back to the airport for bonus
points? Yuk.


Then don't award any bonus for landing back!

Penalise things you dont want, landouts, low finishes, high starts etc.

Ian

We just had a long thread about this involving both theory and
experience from the Canadian nationals, the conclusion of which was
pretty much that handling landouts is an usolved problem. Let's not
start again without explaining how this fatal flaw will be solved.

John

Please tell us what your Canadian experience is.
In my relative short time flying contests, most of which was flown
in the US, I thought that the US system was the way to go.
But in recent years the relative new Canadian way of flying contests
caught my attention and I like it.
It is flexible relative easy for the CD to administer.
it combines a number of task elements which the AST and the PST
advocates like, it leave a lot of room for the more aggressive competitor.
It can be used in a sports class setting as well, with a relative wide
handicap spread.
Landouts are reduced too.

Regards
Udo

The Time - Distance scoring system has been used for the last three Canadian
Nationals. It has been well accepted by the majority of the competing
pilots. So far, we have still normalized the score to 1000 points, in order
to make the TDT compatible with traditional scoring systems for Assigned
Speed Tasks and Pilot Selected Tasks which are still part of the Canadian
Nationals Rules. However, we are now comfortable enough with TDT to
seriously consider to eliminate the 1000 point system altogether and express
the score in kilometers instead of points.

Here are the main reasons why we went to a scoring system where speed is
expressed in distance over a set time TDT:

1. Fairness
The pilot's score depends exclusively on his distance achieved and is not
influenced by the scores of other pilots. The scoring bias in favour of
gaggles (see Bruno Gantenbrink's lone wolf article) of traditional scoring
systems is eliminated.

2. Logic
Traditional scoring systems attempt to score two dimensions: Speed and
Distance. The ratio of speed to distance points is arbitrary and depends on
the ratio of pilots who complete the task to those who land out.

3. Simple scoring
Since the score is based on distance every contestant knows where s/he
stands as soon as the distance is totaled up.

4. Suitable for Assigned Tasks as well as Pilot Selected Tasks and AAT
Obviously an Assigned Task must be open ended to avoid that competitors run
out of task before their time is up. Such a task could be set as a triangle
that brings the competitors back to the contest site after about 80% of the
estimated achievable distance. A smaller triangle (approx. 40% of achievable
distance) can be added with multiple laps to ensure even a very fast
competitor will not run out of task.
In practice most contestants will "time-out" near the competition site. The
slower ones will be on the last leg of the first triangle, while fast pilots
will be somewhere in the smaller triangle.

The traditional scoring systems need different scoring formulas for
assigned, pilot selected or area tasks.

5. Classic task setting problems eliminated
The margins for error are significantly greater since the major decision is
on the task time (i.e. 2, 3 or 4 hrs) and the distance is only secondary.
Even if the task setter gets it somewhat wrong, it may still be a great day.

6. Ideal for handicapping
The traditional Assigned Speed Task is problematic when handicaps vary
widely. In such a scenario it is quite possible that high performance ships
burn around the course in less than two hours, de-rating the day, while
lower performance ships cannot complete the course. Since the best handicap
factor cannot compensate for the loss of speed points, the low performance
ships have no chance. An assigned TDT can be set in such a way that
everybody has a good chance to make it home and even if he doesn't he still
gets a reasonable score.

Is it a race?
Of course, the pilot who goes the furthest in a given time is also the
fastest.


And what do you do about landouts? How do you avoid the incentive to
dive to 100 feet as time ends?

Landouts are scored in the same dimension as finishers -- distance. It has
not been our experience that people drive it down to the deck at timeout.
This strategy may win one day if one is lucky enough to get back up and make
it home but over several contest days the loss of the home bonus would
outweigh the benefits.

We just had a long thread about this involving both theory and
experience from the Canadian nationals, the conclusion of which was
pretty much that handling landouts is an usolved problem

In our experience landouts have not been a problem at all. Actually, we have
seen fewer landouts on assigned time distance tasks than on assigned speed
tasks. The distance of an assigned speed task should be such that it makes
good use of the day's expected conditions. If the day doesn't turn out as
expected a number or all of the contestants will not be able to finish the
task and land out. We have never had a mass landout under TDT scoring --
assigned or pilot selected.

There is little incentive for "make it or break" it final glides because
landing on the wrong side of the fence doesn't eliminate you from the race.
Therefore final glides in the "coffin corner" have not been a problem.

Nobody seems happy with the kind
of flying that total distance requires.

As I said before, the majority of competitors who have flown the last three
Canadian Nationals are quite happy with this scoring system.

There are various possibilities to eliminate the incentive to be low on
time-out. Such as incorporating the time out altitude in the score.

Joerg Stieber

Sporting Committee
Soaring Association of Canada

P.S. For responses to my personal e-mail address remove everything before
the "j" in my e-mail address.

"Joerg & Renate Stieber" wrote in message ...

The Time - Distance scoring system has been used for the last three Canadian
Nationals. It has been well accepted by the majority of the competing
pilots. So far, we have still normalized the score to 1000 points, in order
to make the TDT compatible with traditional scoring systems for Assigned
Speed Tasks and Pilot Selected Tasks which are still part of the Canadian
Nationals Rules. However, we are now comfortable enough with TDT to
seriously consider to eliminate the 1000 point system altogether and express
the score in kilometers instead of points.


This will have a profound impact on contests. Right now, despite day
devaluation, contests are won and lost on the weak days. If we add up
the separate days based on distance achieved rather than 1000 points,
this will end up putting much more emphasis on the strong days.

For example, think of a weak day in which the winner goes 50 miles vs
a strong (ridge!) day in which the winner goes 300 miles. Suppose
noone lands out, so we have 1000 points both days. If you do 90% of
the winner's speed on each day, you do 45 miles on the weak day and
280 miles on the strong day. Under current scoring, you get 900 points
for each day. Under TDT scoring you get 45 points for the weak day and
280 points for the strong day. Automatic, and very strong, day
devaluation!

The big issue is that the difference between 90 and 91 mph on a strong
day becomes much more important than the difference between 30 and 31
mph on a weak day. It's 3 times as important if the tasks are the same
length of time, since you cover 3 times as much distance going from 90
to 91 mph than you do going from 30 to 31 mph. If the fast day is a 4
hour task and the weak day is a 2 hour task, it becomes 6 times more
important to go the extra mph on a strong day.

Advocates might say this is good. (Though it could also be achieved by
changing the mind-boggling devaluation formula we currently use to one
based on distance achieved, if that's the only benefit.)

I (for once) don't have a strong opinion, but I'm curious if the
supporters have thought this through, and why they think putting so
much more emphasis on strong days is a good idea.

John Cochrane

Interesting point, John. We will know more how this plays out when we have
flown our first contest using accumulated distance.

Personally, I believe longer tasks should have a higher weight because the
luck factor is averaged out much better over a longer task. For example on a
day where contestants fly only 50 miles, a luck thermal that gets one pilot
2000 ft when everyone else is landing, will give this pilot an additional
distance of 13 miles which is a significant percentage gain over the field.
There is no way that one lucky thermal can make such a difference during a
long task.

As I said, so far we have scored based on distance but normalized to 1000
points. This has led to interesting results during the 2003 Nationals where
the weather was extremely weak:

On a day where the winner flew 200 km, one km was worth 5 points (no
devaluation)
On a day where the winner flew 100 km, one km was worth about 0.7 points due
to severe devaluation
If we had a 500 k day, one km would have been worth 2 points

Pretty strange, I would say.

BTW, I believe the reason why contests get decided on weak days is simply
that some pilots land out and take a significant landout penalty under
traditional scoring, and other pilots are lucky enough to get the extra
thermal that gets them back.

All the Best,

Joerg Stieber

(John Cochrane) wrote
The big issue is that the difference between 90 and 91 mph on a strong
day becomes much more important than the difference between 30 and 31
mph on a weak day. It's 3 times as important if the tasks are the same
length of time, since you cover 3 times as much distance going from 90
to 91 mph than you do going from 30 to 31 mph. If the fast day is a 4
hour task and the weak day is a 2 hour task, it becomes 6 times more
important to go the extra mph on a strong day.

3 times more important????
91t-90t=t
31t-30t=t
((V+1)*t)-(V*t)=(V+1-V)*t= t

6 times more important????
91*4-90*4=4 91*2-90*2=2
31*4-30*4=4 31*2-30*2=2

Under this type of scoring 1 extra mph is worth the task time whether
it's a 10 or 200mph day. It's the 1000 point scoring which gives the
point difference between 90 and 91 as one third the difference between
30 and 31. And then throws in another effective 50% devaluation of
the good day if it happened to be 4 hours of racing instead of 2.
Anyway, you could look at it that way.

However, on a 4 hr task you waste only 2.64 minutes to get from 91mph
down to 90, but you waste a full 7.74 minutes to get from 31mph down
to 30. How in the world can TET purist Bill Feldbaumer stomach both
differences being worth 4 miles? A minute is a minute, right?

Jonathan Gere


Advocates might say this is good. (Though it could also be achieved by
changing the mind-boggling devaluation formula we currently use to one
based on distance achieved, if that's the only benefit.)

I (for once) don't have a strong opinion, but I'm curious if the
supporters have thought this through, and why they think putting so
much more emphasis on strong days is a good idea.

John Cochrane

I was wondering how long it would take for TET to rear
its head in this thread. Flew it for a whole season
- hated it. This seems only somewhat better in that
it seems easier to measure distance in a set time than
to convert distance into time - but is misses the basic
viceral appeal that racing is about speed over a course
- not whatever part you've completed by some indeterminate
time. Particularly given that the final glide normally
is at ~30 knots faster than average X-C speed, so it
would seem to give winners a non-linear point spread
over those who didn't get far into final glide.

If it were me, I'd monitor the radio for the fastest
finishers - then dive for the deck if I weren't close
to home -- no point in having any altitude in the bank
and not turn it into points, plus you'd want to end
your flight on a glide rather than a climb for speed
averaging reasons.

As for weighting, why wouldn't you measure distance
on a daily basis and weight the days by elapsed time
for each day? I agree that time is the main factor
that determines how many chances a pilot gets to make
good decisions or bad ones.

Worth noodling on as a concept - but at first blush
it seems to add more problems that it solves and might
require too many band-aids to make workable.

9B

At 07:18 07 October 2003, Jonathan Gere wrote:
(John Cochrane) wrote
The big issue is that the difference between 90 and
91 mph on a strong
day becomes much more important than the difference
between 30 and 31
mph on a weak day. It's 3 times as important if the
tasks are the same
length of time, since you cover 3 times as much distance
going from 90
to 91 mph than you do going from 30 to 31 mph. If
the fast day is a 4
hour task and the weak day is a 2 hour task, it becomes
6 times more
important to go the extra mph on a strong day.

3 times more important????
91t-90t=t
31t-30t=t
((V+1)*t)-(V*t)=(V+1-V)*t= t

6 times more important????
91*4-90*4=4 91*2-90*2=2
31*4-30*4=4 31*2-30*2=2

Under this type of scoring 1 extra mph is worth the
task time whether
it's a 10 or 200mph day. It's the 1000 point scoring
which gives the
point difference between 90 and 91 as one third the
difference between
30 and 31. And then throws in another effective 50%
devaluation of
the good day if it happened to be 4 hours of racing
instead of 2.
Anyway, you could look at it that way.

However, on a 4 hr task you waste only 2.64 minutes
to get from 91mph
down to 90, but you waste a full 7.74 minutes to get
from 31mph down
to 30. How in the world can TET purist Bill Feldbaumer
stomach both
differences being worth 4 miles? A minute is a minute,
right?

Jonathan Gere


Advocates might say this is good. (Though it could
also be achieved by
changing the mind-boggling devaluation formula we
currently use to one
based on distance achieved, if that's the only benefit.)

I (for once) don't have a strong opinion, but I'm
curious if the
supporters have thought this through, and why they
think putting so
much more emphasis on strong days is a good idea.

John Cochrane

Hello John,

In making an evaluation between two scoring systems, it is important
to focus on all the advantages and disadvantages of both systems. If
we focus solely on one disadvantage (or advantage) of one of the
systems, we will not make a net evaluation of both systems. After a
net evaluation, we might be willing to accept a disadvantage in one
system in order to get its' many other advantages.

I believe that the most important function of a scoring system is to
produce scores which as closely as possible represent the measured
performances of the competitors. This is unquestioned in other racing
sports. They go to great lengths to score accurately. With all the
effort we put into a soaring contest, we deserve to have the proper
champion chosen and all the other pilots ordered properly.

I am sending you a copy of my paper on scoring. It contains examples
which show that 1000-point scores do not represent measured
performances accurately.

I would like to add a personal example also. I flew in a regional in
which the 1000--point scoring put me in second place. The actual
measured performances - elapsed times - showed a different result. The
1000-point champion spent 40 minutes more in the air than I did (we
both finished all the tasks). In no other form of racing would a
result like that be accepted.

John, if you do not want to embrace distance scoring as a solution for
these inaccuracies, what solutions do you propose?

Accuracy is an issue underlying your discussion with Joerg Stieber
about scoring systems emphasizing weak or strong days. To be accurate
a scoring system must score what actually happens in the contest. If a
day is short it must be scored as such. If a day is long it must be
scored as such. Calculations of "points per minute" are an artifact of
1000-point scoring and its' arbitrary setting of both short and long
days to the same 1000 points.

Bill Feldbaumer 09