USA 2010 Competition Rules Committee Minutes Posted

On Dec 16, 5:38*pm, John Cochrane
wrote:
Anyone care to provide a good explanation of the new 'long landout vs
early finisher' scoring rule?

TA

Take a look at the poll, question 4, which tries to explain it all
compactly.

http://www.ssa.org/files/member/2010...inion%20Poll%2...

Come back if that isn't clear

John Cochrane

There is a scenario I can't quite figure under the new rule. Say ALL
the finishers are MT15 and very short distances but a bunch of pilots
were able to rack up long distances but not get home. This can happen
with big weather systems moving through. The choice you have is stay
close to home so you can finish and risk a short flight or follow the
good conditions on the chance that you'll be able to get back home
later. I think under the new rules you might make the bet that none of
the long flights finish, but if even one of them succeeds it radically
changes the scoresheet because all the short finishers see their
scores cut down dramatically as BESTDIST goes dramatically up. Also
all the long non-finishers would see their scores go up if even one of
them gets home. It also potentially gets tangled up in devaluation
depending on the ratios.

I guess versus the old system it gives you some additional incentive
to be the hero and get around on a long flight even in dicey
conditions.

Any insights?

9B

There is a scenario I can't quite figure under the new rule. *Say ALL
the finishers are MT15 and very short distances but a bunch of pilots
were able to rack up long distances but not get home. This can happen
with big weather systems moving through. The choice you have is stay
close to home so you can finish and risk a short flight or follow the
good conditions on the chance that you'll be able to get back home
later. I think under the new rules you might make the bet that none of
the long flights finish, but if even one of them succeeds it radically
changes the scoresheet because all the short finishers see their
scores cut down dramatically as BESTDIST goes dramatically up. Also
all the long non-finishers would see their scores go up if even one of
them gets home. It also potentially gets tangled up in devaluation
depending on the ratios.

I guess versus the old system it gives you some additional incentive
to be the hero and get around on a long flight even in dicey
conditions.

Any insights?

9B

That's pretty much right.

Important note: In US rules, when there are any "finishers", BESTDIST
is still calculated as the best distance a finisher achieves. BESTDIST
does not reflect very long landouts. Thus, if the "finishers" go 100
miles, but some other guys all go 400 miles and land out, BESTDIST is
still 100 miles. The long landouts still only get 100 miles of
distance points. This is a separate problem, which maybe we'll think
about fixing someday, or maybe not. (Changing that to BESTDIST = the
long landout leads to another can of worms in terms of unintended
clever strategies.) One at a time, this is confusing enough!

The new rule only changes the scores of very short "finishers" when
there are other faster finishers. That's a good principle to keep in
mind. For example, it does not change the scores of your long landouts
above, nor of the 400 mile guys if one of them makes it home.

The only change is, a slow finisher is guaranteed the best of HIS
distance points or his speed points, whereas he used to be guaranteed
the best BESTDIST distance points, or his speed points. That's it.

What happens then is pretty much what you describe. If none of the
400 mile guys make it back, the 100 mile guys win the day, and the 400
mile guys ony get distance points as if they flew 100 miles. (And the
day will be strongly devalued).

If one of the 400 mile guys squeaks back to the airport, under old
rules the 100 mile "finishers" would have gotten 630 points, equal to
a 399 mile landout. Under the new rule the 100 mile "finishers" will
get 100/400*600 + 30 = 180 points, just as if they had landed out at
an airport at 100 miles, plus 5 points extra.

So, as you describe, the change does not guarantee that going longer
will win the day. But it does rather substantially increase the odds
that going longer will pay off. If you make it back after going
longer, you'll destroy the scores of the 100 mile guys. If you landout
at 399 but someone else goes 400 miles and makes it back, then your
399 mile landout will be worth 599 (+25) points, and you will destroy
the 100 mile guys.

This is an important strategic consideration that pilots need to be
aware of. Keeping going under a TAT / MAT rather than stopping very
early--say 1 -2 hours into a 3 hour task--is now a much more
attractive option. It's almost back to the way you would have thought
about it under an AST, where you would not stop and land at an airport
along the way unless things were really pretty desperate. It's not
quite that much. There is still a bit stronger incentive to cut short
a TAT/MAT than an AST because, as you describe, you can gamble that
nobody goes longer and makes it back. But that gamble faces longer
odds than it used to.

I don't think of this as a "change" I think of it as "fixing an
uninteded bug in the rules." We were happy with the tradeoffs pilots
were making under AST regarding stopping at an airport or keeping
going. When we ported the scoring formulas to TAT/MAT, as I view it,
we inadvertently opened this clever strategy to go back after 1 hour
and guarantee yourself 630 points even if the winners do 400 miles.
Loophole now closed.

John Cochrane
BB

On Dec 17, 11:33*am, John Cochrane
wrote:
There is a scenario I can't quite figure under the new rule. *Say ALL
the finishers are MT15 and very short distances but a bunch of pilots
were able to rack up long distances but not get home. This can happen
with big weather systems moving through. The choice you have is stay
close to home so you can finish and risk a short flight or follow the
good conditions on the chance that you'll be able to get back home
later. I think under the new rules you might make the bet that none of
the long flights finish, but if even one of them succeeds it radically
changes the scoresheet because all the short finishers see their
scores cut down dramatically as BESTDIST goes dramatically up. Also
all the long non-finishers would see their scores go up if even one of
them gets home. It also potentially gets tangled up in devaluation
depending on the ratios.

I guess versus the old system it gives you some additional incentive
to be the hero and get around on a long flight even in dicey
conditions.

Any insights?

9B

That's pretty much right.

Important note: In US rules, when there are any "finishers", BESTDIST
is still calculated as the best distance a finisher achieves. BESTDIST
does not reflect very long landouts. Thus, if the "finishers" go 100
miles, but some other guys all go 400 miles and land out, BESTDIST is
still 100 miles. The long landouts still only get 100 miles of
distance points. This is a separate problem, which maybe we'll think
about fixing someday, or maybe not. (Changing that to BESTDIST = the
long landout leads to another can of worms in terms of unintended
clever strategies.) One at a time, this is confusing enough!

The new rule only changes the scores of very short "finishers" when
there are other faster finishers. That's a good principle to keep in
mind. For example, it does not change the scores of your long landouts
above, nor of the 400 mile guys if one of them makes it home.

The only change is, a slow finisher is guaranteed the best of HIS
distance points or his speed points, whereas he used to be guaranteed
the best BESTDIST distance points, or his speed points. That's it.

What *happens then is pretty much what you describe. If none of the
400 mile guys make it back, the 100 mile guys win the day, and the 400
mile guys ony get distance points as if they flew 100 miles. (And the
day will be strongly devalued).

If one of the 400 mile guys squeaks back to the airport, under old
rules the 100 mile "finishers" would have gotten 630 points, equal to
a 399 mile landout. Under the new rule the 100 mile "finishers" will
get 100/400*600 + 30 = 180 points, just as if they had landed out at
an airport at 100 miles, plus 5 points extra.

So, as you describe, the change does not guarantee that going longer
will win the day. But it does rather substantially increase the odds
that going longer will pay off. If you make it back after going
longer, you'll destroy the scores of the 100 mile guys. If you landout
at 399 but someone else goes 400 miles and makes it back, then your
399 mile landout will be worth 599 (+25) points, and you will destroy
the 100 mile guys.

This is an important strategic consideration that pilots need to be
aware of. Keeping going under a TAT / MAT rather than stopping very
early--say 1 -2 hours into a 3 hour task--is now a much more
attractive option. It's almost back to the way you would have thought
about it under an AST, where you would not stop and land at an airport
along the way unless things were really pretty desperate. It's not
quite that much. *There is still a bit stronger incentive to cut short
a TAT/MAT than an AST because, as you describe, you can gamble that
nobody goes longer and makes it back. But that gamble faces longer
odds than it used to.

I don't think of this as a "change" I think of it as "fixing an
uninteded bug in the rules." We were happy with the tradeoffs pilots
were making under AST regarding stopping at an airport or keeping
going. When we ported the scoring formulas to TAT/MAT, as I view it,
we inadvertently opened this clever strategy to go back after 1 hour
and guarantee yourself 630 points even if the winners do 400 miles.
Loophole now closed.

John Cochrane
BB

John, thanks for the pointer to the poll - after some head-scratching
I finally figured out the proposed scoring column for the examples
shown. Now all I have to do is completely re-write the talk I'm doing
at the SSA convention. Adding insult to injury, I tried to give this
talk at last year's convention, but the scheduling guru couldn't fit
it in. Had that happened, I would have been safely out of town before
the rules got changed ;-).

Timing is everything ....

TA

On Dec 17, 8:33*am, John Cochrane
wrote:
There is a scenario I can't quite figure under the new rule. *Say ALL
the finishers are MT15 and very short distances but a bunch of pilots
were able to rack up long distances but not get home. This can happen
with big weather systems moving through. The choice you have is stay
close to home so you can finish and risk a short flight or follow the
good conditions on the chance that you'll be able to get back home
later. I think under the new rules you might make the bet that none of
the long flights finish, but if even one of them succeeds it radically
changes the scoresheet because all the short finishers see their
scores cut down dramatically as BESTDIST goes dramatically up. Also
all the long non-finishers would see their scores go up if even one of
them gets home. It also potentially gets tangled up in devaluation
depending on the ratios.

I guess versus the old system it gives you some additional incentive
to be the hero and get around on a long flight even in dicey
conditions.

Any insights?

9B

That's pretty much right.

Important note: In US rules, when there are any "finishers", BESTDIST
is still calculated as the best distance a finisher achieves. BESTDIST
does not reflect very long landouts. Thus, if the "finishers" go 100
miles, but some other guys all go 400 miles and land out, BESTDIST is
still 100 miles. The long landouts still only get 100 miles of
distance points. This is a separate problem, which maybe we'll think
about fixing someday, or maybe not. (Changing that to BESTDIST = the
long landout leads to another can of worms in terms of unintended
clever strategies.) One at a time, this is confusing enough!

The new rule only changes the scores of very short "finishers" when
there are other faster finishers. That's a good principle to keep in
mind. For example, it does not change the scores of your long landouts
above, nor of the 400 mile guys if one of them makes it home.

The only change is, a slow finisher is guaranteed the best of HIS
distance points or his speed points, whereas he used to be guaranteed
the best BESTDIST distance points, or his speed points. That's it.

What *happens then is pretty much what you describe. If none of the
400 mile guys make it back, the 100 mile guys win the day, and the 400
mile guys ony get distance points as if they flew 100 miles. (And the
day will be strongly devalued).

If one of the 400 mile guys squeaks back to the airport, under old
rules the 100 mile "finishers" would have gotten 630 points, equal to
a 399 mile landout. Under the new rule the 100 mile "finishers" will
get 100/400*600 + 30 = 180 points, just as if they had landed out at
an airport at 100 miles, plus 5 points extra.

So, as you describe, the change does not guarantee that going longer
will win the day. But it does rather substantially increase the odds
that going longer will pay off. If you make it back after going
longer, you'll destroy the scores of the 100 mile guys. If you landout
at 399 but someone else goes 400 miles and makes it back, then your
399 mile landout will be worth 599 (+25) points, and you will destroy
the 100 mile guys.

This is an important strategic consideration that pilots need to be
aware of. Keeping going under a TAT / MAT rather than stopping very
early--say 1 -2 hours into a 3 hour task--is now a much more
attractive option. It's almost back to the way you would have thought
about it under an AST, where you would not stop and land at an airport
along the way unless things were really pretty desperate. It's not
quite that much. *There is still a bit stronger incentive to cut short
a TAT/MAT than an AST because, as you describe, you can gamble that
nobody goes longer and makes it back. But that gamble faces longer
odds than it used to.

I don't think of this as a "change" I think of it as "fixing an
uninteded bug in the rules." We were happy with the tradeoffs pilots
were making under AST regarding stopping at an airport or keeping
going. When we ported the scoring formulas to TAT/MAT, as I view it,
we inadvertently opened this clever strategy to go back after 1 hour
and guarantee yourself 630 points even if the winners do 400 miles.
Loophole now closed.

John Cochrane
BB

Just to clarify, is BESTDIST the longest distance of any finisher or
the distance of the fastest finisher? I think it's a pretty big
difference.

9B

On Dec 18, 5:33*am, John Cochrane
wrote:
The only change is, a slow finisher is guaranteed the best of HIS
distance points or his speed points, whereas he used to be guaranteed
the best BESTDIST distance points, or his speed points. That's it.

Is it maybe time to retire the separate concepts of speed points and
distance points? IN particular, wouldn't it be better if outlanders
got credit for speed too?

As far as i can see, the only reason not to is the practical one that
in the old days there was no evidence of exactly when an outlanding
was made, making it impossible to reliably calculate speed to that
point.

In these days of GPS traces that is no longer true.

It's 01:40 here and I only gave this a few minute's thought, but I
can't immedately see major unfairness in the following proposal:

raw points = S * (D - L/2)

Whe

D = the scoring distance as defined by the task rules
L = the distance from the landing point to the finish line (0 for
finishers)
S = speed achieved over the scoring distance

The raw points could be simply kept as is and totaled up over the
contest (this would devalue bad days in a natural way), or the maximum
could be scaled to 1000 or some lesser value according to existing day
devaluation rules.

This seems to me to have the following nice characteristics:

- if you fly the same distance as someone else then it's better to do
it faster, regardless of whether you both complete the task or both
land out at the same place.

- if you achieve the same speed as someone else then it's better to
maintain that speed over a longer distance.

- speeds tend to have a fairly small spread on a given day (except for
those who spend a long time on a low save), so the preferred method to
more points is more distance.

- the penalty for landing out just short of the airfield is very
small, reducing the incentive to try to stretch and just scrape over
the fence.

- once you stop making forward progress it's better to land out
promptly than to waste a lot of time scratching at low level. This may
be true even in the case of an eventual save. (I'd have to run the
figures)

- if faced with a long, slow, skinny, final glide it may in fact be
better to fly quickly to a good outlanding area that you can reach
easily. (once again I'd have to run the figures)

- distance flown away from home counts for half, distance towards home
counts for 1.5x. If you're going to land after 100 miles it's better
to do it out and return than straight out.


What do you think? Totally stupid? Perverse and unsafe incentives I
didn't notice? Too complex?

I'm certainly prepared to debate whether that "2" is the right value.
For sure the number needs to be bigger than 1, otherwise a straight
out task is worth zero.

I also wondered about a slight variation:

raw points = (D^2 - (L^2)/2) / T

Where T is the flight time.

This is less different than it first appears. S = D/T, so the first
version can also be given as:

raw points = (D/T) * (D - L/2) = (D^2 - DL/2) / T

This is the same in the event of a straight out flight but the
alternative version penalises landouts near home relatively much less
after a long flight than after a short one.