Scoring Discussion

To try and get this going separate from the discussion of the ongoing World Championships,I think we can agree:

Any scoring system will have an unintended consequences.

Current FAI scoring system used at World and Continental Championships tends to encourage group flying (reward for striking out on your own and completing when nobody else does is very small, but the penalty for coming up short is very large). It also does not provide speed point in proportion to the best speed. And it can compress (or expand) scores by having people intentionally land out.

Pilots don't like the idea of being 20 KPH faster than the slowest guy, but still getting the same number of points as him (minimum speed points).

Now, feel free to discuss various scoring system options, and be prepared for people to comment on the "unintended consequences" of that method.

Ready.... Go!

torstai 19. tammikuuta 2017 19.10.44 UTC+2 Steve Leonard kirjoitti:
To try and get this going separate from the discussion of the ongoing World Championships,I think we can agree:

Any scoring system will have an unintended consequences.

Current FAI scoring system used at World and Continental Championships tends to encourage group flying (reward for striking out on your own and completing when nobody else does is very small, but the penalty for coming up short is very large). It also does not provide speed point in proportion to the best speed. And it can compress (or expand) scores by having people intentionally land out.

Pilots don't like the idea of being 20 KPH faster than the slowest guy, but still getting the same number of points as him (minimum speed points).

Now, feel free to discuss various scoring system options, and be prepared for people to comment on the "unintended consequences" of that method.

Ready.... Go!

Clipped from Benalla thread:

Ok, let's imagine score formula that gives 900 points to all finishers and remaining 100 points are awarded according to speed. No more gaggles, problems solved?

Wrong. Nothing changes. Same pilots will win and others loose. All we change is point spread between pilots. If winner of the whole competition gets 10000 points and last one 6000, new formula gives 10000 to winner and 9500 to last one. Point spread is very small, but it is as difficult to make any difference by flying as before. Next we start calculating decimals.

"For every complex problem there is an answer that is clear, simple, and wrong."

Steve:

Thanks, yes, let's keep Benalla on Benalla and scoring on scoring. My concrete proposal

points = (day devaluation) x max [ 1000 x speed / winner speed, 750 x distance , winner distance].

Let's leave day devaluation out of this for the moment, as it's much less important.

The ratio of speed to distance points does not change as a function of land outs. This is the major innovation. Therefore, just finishing vs. just short is always the same thing. We do not have the situation that the only finisher is 1000 with the gaggle just short at 999 while the only just short is 300 with the gaggle at 1000. The lone wolf can strike out.

It's not magic. There still will be gaggles. 750 is still a bad outcome. But it tilts the playing field a bit to the lone wolf, early starter, etc.

The main "defect" mentioned in the previous thread, is that someone going 66% of winner's speed and someone going 75% of winner's speed get the same points. Reply: That's already in IGC rules. Very slow finishes are counted as landing out. Reply 2: Yeah, but so what. If you lower the 750, then you lower distance points, meaning that going further on distance days counts less, and that a land out is more of a disaster. On that basis 800 distance points makes more sense.

Notice also that by removing 2 x speed -- which was pointless, as it makes speed points just as easy to get as to lose, and does not affect the final ranking -- these distance points are the equivalent of half as many under current IGC rules. So even 750 is harsh. That's another argument for 800 or even more.

Day devaluation? Again, it's a separate issue, but I would also not base this on land outs. We only use land outs to measure things because we didn't use to have gps. A concrete proposal:

day devaluation = min [ median time from start to finish or landing of top 10% of pilots / 3 hours, 1]

We measure time on course for everyone. If the top 10% of pilots fly 3 hours, it's a fully valid day. If we all land out, we're fighting for the full 750 distance points. This keeps the current IGC philosophy that distance days are valid days. It's the median and the top 10% so that one pilot does not affect the devaluation formula -- no incentive to sit on a ridge and orbit to run up the clock and lower the devaluation of a hopeless day

Advantage 2: This is all really really simple! It also removes the quirks of current IGC rules that encourage pilots to intentionally land out on some days.

John Cochrane

On Thursday, 19 January 2017 21:19:52 UTC+2, John Cochrane wrote:
Steve:

Thanks, yes, let's keep Benalla on Benalla and scoring on scoring. My concrete proposal

points = (day devaluation) x max [ 1000 x speed / winner speed, 750 x distance , winner distance].

Let's leave day devaluation out of this for the moment, as it's much less important.

The ratio of speed to distance points does not change as a function of land outs. This is the major innovation. Therefore, just finishing vs. just short is always the same thing. We do not have the situation that the only finisher is 1000 with the gaggle just short at 999 while the only just short is 300 with the gaggle at 1000. The lone wolf can strike out.


Interesting proposal, but it creates massive incentive to glide home over that last unlandable forest no matter what altitude. Everyone landing out 5 km short on last good landing place, one tries risky glide home and scores extra 250 points. Current formula gives risk taker only small benefit.

It amazes that there is constantly reference to "planned outlanding" that would give more points than coming home, apparently by changing day factor (?). Such day has never occurred in my career as a competition pilot/organizer, nor have I ever heard of pilot who had even thought about landing out instead of coming home. I doubt that playing this game would require conspiracy of a huge gaggle. It is purely academic scenario and has absolutely nothing to do with real competition flying at any level. Yes, we can speculate AFTERWARDS that if this-and-that pilot would have outlanded, scores would be like that. It has nothing to do with flying tactic.

IGC formula is far from perfect but more I read about new formulas and changes, more I admire it. I think instead of inventing new formula ground up, it would be wiser to carefully adjust current formula to direction wanted (which is not clear at all). It might be slight change to distance/speed points relation, change to minimum time/distance giving 1000p day, day factor (it is very important tool) etc.

On Friday, January 20, 2017 at 10:09:15 AM UTC+3, krasw wrote:
On Thursday, 19 January 2017 21:19:52 UTC+2, John Cochrane wrote:
Steve:

Thanks, yes, let's keep Benalla on Benalla and scoring on scoring. My concrete proposal

points = (day devaluation) x max [ 1000 x speed / winner speed, 750 x distance , winner distance].

Let's leave day devaluation out of this for the moment, as it's much less important.

The ratio of speed to distance points does not change as a function of land outs. This is the major innovation. Therefore, just finishing vs. just short is always the same thing. We do not have the situation that the only finisher is 1000 with the gaggle just short at 999 while the only just short is 300 with the gaggle at 1000. The lone wolf can strike out.


Interesting proposal, but it creates massive incentive to glide home over that last unlandable forest no matter what altitude. Everyone landing out 5 km short on last good landing place, one tries risky glide home and scores extra 250 points. Current formula gives risk taker only small benefit.

It amazes that there is constantly reference to "planned outlanding" that would give more points than coming home, apparently by changing day factor (?). Such day has never occurred in my career as a competition pilot/organizer, nor have I ever heard of pilot who had even thought about landing out instead of coming home. I doubt that playing this game would require conspiracy of a huge gaggle. It is purely academic scenario and has absolutely nothing to do with real competition flying at any level. Yes, we can speculate AFTERWARDS that if this-and-that pilot would have outlanded, scores would be like that. It has nothing to do with flying tactic.

IGC formula is far from perfect but more I read about new formulas and changes, more I admire it. I think instead of inventing new formula ground up, it would be wiser to carefully adjust current formula to direction wanted (which is not clear at all). It might be slight change to distance/speed points relation, change to minimum time/distance giving 1000p day, day factor (it is very important tool) etc.

Very difficult to do a sneaky outlanding as you would have to know, while still in flight and with a choice possible, not only how many other pilots had already made it home but also how many pilots behind you might still make it home.

If you're the guy who could devalue the day by landing out instead of finishing, then it only takes one other pilot still airborne to scratch home at MC=0 and spoil your plans. So you have to be sure that everyone else has already landed, one way or another, and how many made it and how many didn't.

Interesting proposal, but it creates massive incentive to glide home over that last unlandable forest no matter what altitude. Everyone landing out 5 km short on last good landing place, one tries risky glide home and scores extra 250 points. Current formula gives risk taker only small benefit.
____________

I believe this is only true if practically everyone lands out and you are the sole finisher. The more common case is you are (or believe you may be since you don't have perfect information) one of a few landouts. In Sean's example landing out cost him nearly 700 points. In general IGC rules are much harsher on outlandings than US rules which are harsher on outlandings than John's proposal (most of the time except in the case of a distance day with near 100% landouts). You are really describing only the case of the one guy who manages to get close to home on a difficult day where IGC rules say "don't bother".

Having looked at the issue of marginal glides home and scoring quite a lot I've concluded that most pilots, if they get within a marginal glide from home, will go for it - for reasons unrelated to points. A landout and retreive is a significant risk and hassle all by itself.

Andy Blackburn
9B

On Friday, January 20, 2017 at 6:43:59 PM UTC+3, Andy Blackburn wrote:
Interesting proposal, but it creates massive incentive to glide home over that last unlandable forest no matter what altitude. Everyone landing out 5 km short on last good landing place, one tries risky glide home and scores extra 250 points. Current formula gives risk taker only small benefit.
____________

I believe this is only true if practically everyone lands out and you are the sole finisher. The more common case is you are (or believe you may be since you don't have perfect information) one of a few landouts. In Sean's example landing out cost him nearly 700 points. In general IGC rules are much harsher on outlandings than US rules which are harsher on outlandings than John's proposal (most of the time except in the case of a distance day with near 100% landouts). You are really describing only the case of the one guy who manages to get close to home on a difficult day where IGC rules say "don't bother".

Having looked at the issue of marginal glides home and scoring quite a lot I've concluded that most pilots, if they get within a marginal glide from home, will go for it - for reasons unrelated to points. A landout and retreive is a significant risk and hassle all by itself.

If you've got any brains you'll have scouted safe landout places 5km, 10km, 20km from home field in various directions -- or, better, responsible contest organizers will have done that for you. So hassle, yes, but hopefully not as much risk as pressing on with zero safety margin.

I've concluded that most pilots, if they get within a marginal glide
from
h=
ome, will go for it - for reasons unrelated to points. A landout and
retrei=
ve is a significant risk and hassle all by itself.

Andy Blackburn
9B


It all depends on the landable fields on the final glide to the finish
line. Given good fields and no obstructions, I think most WGC
contenders will drive straight ahead hoping for some "help" plus
ground effect in order to get across that finish line. If they fall
short,
they just land safely straight ahead. It's when there are obstructions
or poor landing choices on short final where that decision comes more
into play.

RO

First impression - I like it !

Heres my perspective.

1 safety decisions are enhanced by a smaller difference between landout points and speed points

2 gaggles will always happen - its less the scoring mechanism (though its true it is a factor) than being mainly due to the basic fear of losing

3 higher distance points will also help reduce gaggles even with FAI scoring

Ive said this before but 600pts is too high a penalty for landing safely but just short.

Perhaps a less steep scale on the speed points (1.5 factor not 2?) with the higher distance points would be something we could really use in US contests

It would encourage FAI dynamics for speed (7T and others) while reducing the incentive for pushing safety limits (BB and others)

Best

2T

On Thursday, 19 January 2017 11:19:52 UTC-8, John Cochrane wrote:
Steve:

Thanks, yes, let's keep Benalla on Benalla and scoring on scoring. My concrete proposal

points = (day devaluation) x max [ 1000 x speed / winner speed, 750 x distance , winner distance].

Let's leave day devaluation out of this for the moment, as it's much less important.

The ratio of speed to distance points does not change as a function of land outs. This is the major innovation. Therefore, just finishing vs. just short is always the same thing. We do not have the situation that the only finisher is 1000 with the gaggle just short at 999 while the only just short is 300 with the gaggle at 1000. The lone wolf can strike out.

It's not magic. There still will be gaggles. 750 is still a bad outcome. But it tilts the playing field a bit to the lone wolf, early starter, etc.

The main "defect" mentioned in the previous thread, is that someone going 66% of winner's speed and someone going 75% of winner's speed get the same points. Reply: That's already in IGC rules. Very slow finishes are counted as landing out. Reply 2: Yeah, but so what. If you lower the 750, then you lower distance points, meaning that going further on distance days counts less, and that a land out is more of a disaster. On that basis 800 distance points makes more sense.

Notice also that by removing 2 x speed -- which was pointless, as it makes speed points just as easy to get as to lose, and does not affect the final ranking -- these distance points are the equivalent of half as many under current IGC rules. So even 750 is harsh. That's another argument for 800 or even more.

Day devaluation? Again, it's a separate issue, but I would also not base this on land outs. We only use land outs to measure things because we didn't use to have gps. A concrete proposal:

day devaluation = min [ median time from start to finish or landing of top 10% of pilots / 3 hours, 1]

We measure time on course for everyone. If the top 10% of pilots fly 3 hours, it's a fully valid day. If we all land out, we're fighting for the full 750 distance points. This keeps the current IGC philosophy that distance days are valid days. It's the median and the top 10% so that one pilot does not affect the devaluation formula -- no incentive to sit on a ridge and orbit to run up the clock and lower the devaluation of a hopeless day

Advantage 2: This is all really really simple! It also removes the quirks of current IGC rules that encourage pilots to intentionally land out on some days.

John Cochrane

On Thursday, January 19, 2017 at 9:17:09 AM UTC-8, krasw wrote:
torstai 19. tammikuuta 2017 19.10.44 UTC+2 Steve Leonard kirjoitti:
To try and get this going separate from the discussion of the ongoing World Championships,I think we can agree:

Any scoring system will have an unintended consequences.

Current FAI scoring system used at World and Continental Championships tends to encourage group flying (reward for striking out on your own and completing when nobody else does is very small, but the penalty for coming up short is very large). It also does not provide speed point in proportion to the best speed. And it can compress (or expand) scores by having people intentionally land out.

Pilots don't like the idea of being 20 KPH faster than the slowest guy, but still getting the same number of points as him (minimum speed points).

Now, feel free to discuss various scoring system options, and be prepared for people to comment on the "unintended consequences" of that method.

Ready.... Go!

Clipped from Benalla thread:

Ok, let's imagine score formula that gives 900 points to all finishers and remaining 100 points are awarded according to speed. No more gaggles, problems solved?

Wrong. Nothing changes. Same pilots will win and others loose. All we change is point spread between pilots. If winner of the whole competition gets 10000 points and last one 6000, new formula gives 10000 to winner and 9500 to last one. Point spread is very small, but it is as difficult to make any difference by flying as before. Next we start calculating decimals.

"For every complex problem there is an answer that is clear, simple, and wrong."

I think where we crossed is the assumption that speed points need to equal max points for the day minus distance points. In that case where you separate the two doesn't matter, but that's not how scoring needs to work. If you allow overlap you can dial up or down the relative disaster of a landout to your heart's content. The price you pay is the point of overlap - dealing with the conflict between very slow finishers and long landouts. You can either set limits to prevent the overlap or not - that is, either allow some ties for the very slow (or extreme nonlinear scoring of speeds below 80% or so to forestall it), or allow some long non-finishers to score higher than very slow finishers.

Devaluation formulae complicate matters further, particularly if they are asymmetric - leading to all kinds of undesirable results and pilot behavior - as was demonstrated at Benalla.

I don't get the obsession with pilots who finish less than 75% of the winners speed. Missing the granularity of scoring 750 points versus some pilot who scored an even lower number when the bigger point is both pilots have likely been knocked way down. I'd rather keep them both more in contention. If we were looking at a situation where much of the field was on a consistent basis hitting the limit, I'd feel differently, but in a system where the points are 1:1 proportional to the speed ratio, that just doesn't happen unless someone really got into trouble - in which case it was more likely bad luck anyway so why score it like it's really a representative score of that pilot's abilities that day.

9B