LLC 2007

User avatar
Vertigo
Posts: 1080
Joined: Fri Sep 16, 2005 11:17 am
Location: Belgium

LLC 2007

Post by Vertigo » Fri May 05, 2006 11:20 pm

SL and myself, conceived, prepared and organized the previous LLC in just over a week, including making the rules, scoring, (basic) website, etc. For the next Cup, we have a bit more time :) and I’d like to use it to improve on its shortcomings and make LLC 2007 as good as we can get it. Below is a list of things I think could be improved, with proposals, but I’d appreciate your opinion, insights, idea's. Note this is just my personal idea's at this moment, nothing final in here

Here goes:

The team
I was (and am) extremely happy about the teamwork with Bjorn and SL and think we worked together as a perfect and very complementary trio. I would love to work with you guys on the next cup again (assuming you will still want to), but I’m wondering if we shouldn’t consider a forth ‘full timeâ€
Image

User avatar
mac
Posts: 394
Joined: Tue Mar 28, 2006 8:25 pm
Location: Strasbourg, France

Re: LLC 2007

Post by mac » Sat May 06, 2006 2:31 am

Dear Vertigo,

I just flought a few LLC tasks, having joined Condor community less than two months ago, but here below my contributions:
I propose taking only 1 result per pilot per day: the worst result with a valid race start per pilot, even if the worst one is a crash. Any others (duplicates on different servers, and flights with no start) will be dismissed.
I agree with the principle, but:
1. I don't understand what a "flight with no start" means.
2. I have some reservation in case of early disconnected flight: I personally preferred in the past to quit the race and re-start on another "late start" server, as playing online is so much more fun than spending 2 hours on your own...
to give each pilot a webinterface that allows him to upload the IGC, and fill out his own result in a form. We will take this result as the truth, automatically score it and let all other pilots (who can download and check the IGCs) report irregularities.
I disagree, as I believe is up to the organising committee to guarantee a systematic and reliable ranking. A self management based system would open the door both to mistakes and cheats, and both would be a fireback on the credibility and prestige of the competition.
This has been debated extensively, and I know SL disagrees, but when not counting each day, I still think each day should have equal weight. I therefore propose to drop the dayfactor (set it to 1), but keep the speed/distance points system. Another alternative I might find acceptable is a place scoring system for days (as well as months).
I'm not an expert in scoring system and I believe I only raced on dayfactor 1 competitions. Anyhow a possible alternative solution could be to adopt a scoring system that take into considearation also day factors. Something like for example "result of the month=best 3 tasks with dayfactor 1 + best 3 tasks with dayfactor=2".
introducing a place point penalty for each crash, regardless if the dayresult or even month result would be counted or not. For instance, first crash of the season would cost you 10 ? overall points, second crash 25, third one would reset your score to zero or even below zero if you had less than 25 points, otherwise the first two/three races people could crash with no real penalty.
I agree only partially as frequent crashes can be related to poor soaring skills: a beginner could take 8 competitions/month, crash 4 times and arriving 120th for the other 4... why to take away from him the satisfaction of benefitting of those few points?
I see your point, but then I would rather penalise crashes in relation to other results. Two possible factors to be considered here:
1. Only those crashes that are in the "dropped part" of the results of the month should be counted for penalty.
2. Pilots who are hight in the rank should get higher penalties for crashes (as it's more likely these are the fruit of having taken too much risk), pilots that are low in ranks should ge lower penalties (as it's more likely their crashes are due to inexperience).
the above rule (place score penalty for each crash) would encourage people disconnecting on purpose to avoid this penalty. To counter that, I would consider *each* disconnection a crash (!) until proven otherwise.
I understand the concern but I disagree on the solution, because there are conditions (certain kinds of PC crashes) that make impossible for the pilot to demonstrate (s)he didn't disconnected because of an imminent glider crash. Of course in order to implement the crash penalties (previous point), we have to tackle this other problem first... but I don't have any good idea at the moment, sorry... :(

Also, if my computer crashes and a server is stil open, I would definitively try to race again, rather then let my computer decide my ranking...
i propose to drop cloud flying penalties since these can be arbitrary/incorrect in condor (esp patch 8 ), and the benefit of cloud flying might become questionable with patch 9.
I have no sufficient experience to express an informed opinion, but I remember some advocated in other threads for eliminating CFP because in their country, in RL, this is the normal way of flying. Why not to keep it as a factor taks-dependent (meaning that you could chose if having it or not according to the task)? This would also allow to work around possilble temporary bugs of Condors introduced with patches.
It needs an overhaul, with a simple content management system behind it and automated navigation so one doesn’t need to know FrontPage to maintain it
I would also say: a free/open CMS, so that other members of the community could also re-use the code for organising other contests... My personal choice would be Drupal (drupal.org). Very well coded, accessible, modular, with a built-in system of tweaking performance according to server workload...
A redesign wouldn’t hurt either, if anyone feels the desire to submit some designs, I would certainly welcome them. Drop me a line if you have some skills in this area and are willing to make a template
Two things that personally I find improveable:
1. frames (that keep the URL never changing) and make impossible to retreive the URL of a given page directly from the history.
2. during update of the data, the site display lots of script errors and warnings.
I see little reason not to allow some flexibility and having servers at slightly different times, as long as no servers start after initial results and IGCs are published (so, say a 1 or 2 hour time span between first and last servers).
With the "only one result" per race thing, I agree. This would for sure increase participation.
Not so sure if I want to keep the "country cup" or if it could not be replaced by teams that people can create/join as they see fit.
I have an alternative idea: teams should be randomly set up by the LLC organisers, taking only those pilots that gave availability for that. Each team should have the same number of members and prizes should be given to the best teams at the end of the season too.
This way pilots would be fostered in co-operating, sharing, supporting each other in the teams... I think this would make LLC even more exciting! :)

I have nothing against keeping the country cup (although it doesn't say much to me). I would simply add - in this case - a team for those who don't feel to belong to a single country. A kind of "united countries" team, but with fixed members.
There are other things, but I think this is long enough already. Let’s hear your input
That was my personal 2 cents. I appreciate very much the fact that you keep LLC a "participative" collective exercise. :)[/quote]
Image

User avatar
Vertigo
Posts: 1080
Joined: Fri Sep 16, 2005 11:17 am
Location: Belgium

Re: LLC 2007

Post by Vertigo » Sat May 06, 2006 9:54 am

mac wrote: I agree with the principle, but:
1. I don't understand what a "flight with no start" means.
2. I have some reservation in case of early disconnected flight: I personally preferred in the past to quit the race and re-start on another "late start" server, as playing online is so much more fun than spending 2 hours on your own...
Flight with no start, means a flight that didn't fly through the start zone after the opening of the line. So early disconnection (before race start) are not a problem. After you have crossed the line, the flight counts, no matter what happens. If you do get disconnected, continuing offline will be your only option, otherwise it will open the door to abuse (bad start, lets try again on a different server). This is no different as it is now really.
I disagree, as I believe is up to the organising committee to guarantee a systematic and reliable ranking. A self management based system would open the door both to mistakes and cheats, and both would be a fireback on the credibility and prestige of the competition.
As I said, we would also check IGCs. Main difference being we publish results prior to checking them, assuming them to be correct.
I'm not an expert in scoring system and I believe I only raced on dayfactor 1 competitions. Anyhow a possible alternative solution could be to adopt a scoring system that take into considearation also day factors. Something like for example "result of the month=best 3 tasks with dayfactor 1 + best 3 tasks with dayfactor=2".


Nah, I like to stay close to FAI scoring, and I am not going to invent something like this.
I agree only partially as frequent crashes can be related to poor soaring skills: a beginner could take 8 competitions/month, crash 4 times and arriving 120th for the other 4... why to take away from him the satisfaction of benefitting of those few points?


This is a competition, it does require ability to fly. everyone is welcome of course, but if you can not even land safely or turn without crashing into mountains, you shouldn't expect any points. Besides I'm pretty sure >95% of the crashes we saw where due to taking avoidable risks (flutter at start, too close to mountains, not deciding to land out, but just carry on hoping for a miracle, etc....

If we assign more place scores than now, this penalty will not detract the lesser gods among us, it will just ensure the lesser gods that fly safely and land out properly have an advantage over the ones that take insane risks. Sounds fair and reasonable to me.
I understand the concern but I disagree on the solution, because there are conditions (certain kinds of PC crashes) that make impossible for the pilot to demonstrate (s)he didn't disconnected because of an imminent glider crash.
It might be possible this happens, but with the improved lasttrack.ftr feature, it should be *extremely* rare. IF its not, your PC is not fit to fly online condor competitions, and no rule will help out. Besides there is always the jury to decide differently. We have taken people's word for it in the past, but I don't want to turn that into a rule.
Also, if my computer crashes and a server is stil open, I would definitively try to race again, rather then let my computer decide my ranking...
You'd be welcome too, but you wouldn't get any points for the second attempt. This is a must because you can not prove your pc crashed, and we want to avoid people trying several times.

I would also say: a free/open CMS, so that other members of the community could also re-use the code for organising other contests...
Much of the code will be intimately tied with the LLC rules and concept. The parts that aren't (registration, result parsing, perhaps day scoring) will be "open". Registration will probably be shared with Speedbattle, current dayscoring code (in excel) is already available for anyone to use.I'm a bit reluctant to go much further than that. Wiktor Kozlik has already created an open condor contest platform, and I don't think there is much point in having several LLC clones. Maybe one "clone", approved by us, for different time zone would be a good idea though.
Image

EA
Posts: 278
Joined: Mon Feb 20, 2006 1:13 am
Location: EBSH

Re: LLC 2007

Post by EA » Sat May 06, 2006 11:00 am

Vertigo wrote:1) sub 1000 point days.
This has been debated extensively, and I know SL disagrees, but when not counting each day, I still think each day should have equal weight. I therefore propose to drop the dayfactor (set it to 1), but keep the speed/distance points system. Another alternative I might find acceptable is a place scoring system for days (as well as months).
I agree with SL...
Vertigo wrote:2) The “50% of all day resultsâ€

User avatar
Panther
Posts: 110
Joined: Sun Mar 19, 2006 3:26 pm

Post by Panther » Sat May 06, 2006 1:34 pm

Hi Vertigo,

I think your suggestions are well-thought and very beneficial for LLC, I agree with you for most of the points (the only point I disagree is cloud penalty - I would keep them, even if it means a high risk because crashes are punished).

Especially I strongly encourage you to implement the proposed system of penalizing crashes, especially with the rule 'disconnections are crashes unless proven otherwise' which should defeat cheating in most cases. The disadvantage which could in some rare cases possibly be caused for single pilots by rarely possible 'unfair penalties' due to system crashes will be much, much less than the disadvantage caused for the whole community by either cheating if disconnections are not punished or by the currently active system which doesn't punish crashes at all and encourages pilots to take risks which would simply be deadly IRL.

In addition, I would also fully punish midair collisions ('deactivate Q-key') - although in most cases they are caused mainly by one of the colliding pilots who flew like a madman, IRL it is the responsibility of everyone to take care that no collisions occur. In LLC, if pilots are e.g. extremely ruthless while thermaling, penalties should be given based on eyewitness reports.

A more flexible time schedule would also be very good - I for myself find 20:30 the 'earliest possibility' on Sundays and I never had the possibility to join Thursday 20:00 as this is simply too early for me (guess I work too much ;-) ).

From my point of view, I'd prefer to score less than 50% of the monthly tasks. I am working and I have family, so for me it is in most cases simply not possible to participate 50% or even more of the tasks. I think e.g 40% or 'one task less than the half of all for the month' would be a very good compromise to apply no disadvantage to 'frequent flyers', but also support those who love LLC but don't have much spare time.
Image

User avatar
Tima (TSD)
Posts: 1608
Joined: Mon Dec 12, 2005 9:08 am
Location: St.Petersburg, Russia
Contact:

Post by Tima (TSD) » Sat May 06, 2006 2:11 pm

I agree with day factor always = 1. The reason behind is that participants have very different skills which is not the case IRL competitions. So day factor doesn't act here as it is does in the FAI rules.

I strongly support max penalties for cloud flying like IRL. Reasons are already explained by previous speakers.

I don't agree with penalizing of mid-air collisions. Besides the very limited possibility to monitor traffic around you there are also could happen "random" movements of gliders just due to latency on their and/or your internet connection. So even if you flying safe using what you see around, such movements can cause the crash.
Image

TimKuijpers
Posts: 3214
Joined: Sun Apr 24, 2005 12:33 am
Location: The Netherlands

Post by TimKuijpers » Sat May 06, 2006 3:47 pm

My opinionabout the 50% rule.
why 50%? I could understand it would be because some pilots only can fly on thursday/sunday
That way they always have 50%...
not in this case, because the months contain sometimes more thursdays then other months.
If you keep the 50% rule into the system, I would suggest to make a semi-month.
A semi-month would contain as much sundays as it contains thursdays.
This also means that you can fly a task in january which counts for the month february, but it would make much more sense to me.

If 50% is just choosen because some pilots can miss some races, I would prefer to make it 75% and if it comes out at 9 days, it should be 6 and not 7.
This will make the flying also a lot more serious.
With only 50% encounted you can take high risks and hoping it works out right and therby get a much higher speed.
So flying for more then 50% make the flying style more likely to RL
Think positive, flaps negative.

wiktor256
Posts: 72
Joined: Fri Jan 13, 2006 6:32 pm

Post by wiktor256 » Sat May 06, 2006 10:20 pm

I just had an idea about the problem of people not being able to participate in all the races in a month and therefore the need to give them a fair chance by using the 50% rule.
I don't want to start another lenghty discussion on this topic, so if that idea was already discussed, please ignore this message.

What if instead of automatically counting the best 50% of the flights, each pilot would have to declare before the race if a flight will count for scoring or not. If a pilot declares that the flight will count for scoring, it will be scored no matter what. This should reduce risky flying and make it more like RL racing.

To explain it better I'll try to list the "rules" below.
1. 50% of the total races in a month counts for pilot's total score. Usually, this will be 4 races out of 8 or 5 races out of 9. The counted number of races in a month must be published before the first race of a month to allow pilots plan their schedules.
2. Each pilot has to declare before the race if the flight will count for monthly scoring. If no declaration is made, the flight will be counted for scoring by default.
3. The first launch of a glider indicates pilot's participation in the race. The race will be scored even if the pilot did not start the task.
4. Flights declared as non-scored for a month will be scored for daily results and potentially affecting other pilots' scores.
5. Each race has a factor 1 to weigh all the races equally.
6. Pilot's declaration about the race counting for monthly ranking remains hidden until after the race. This is to reduce "leaching" based on the logic "Maybe I shouldn't follow NX because he is taking too much risk because his race doesn't count for scoring".

The big question is: what would be the incentive for pilots to participate in a race that will not count for monthly ranking?
I can think of one: by winning the non-scored races you lower scores of competitors and therefore increasing your own chances to do better in a monthly ranking.

{Edited}
I am not sure if a pilot should be able to change his declaration after the task briefing is announced.
VK
Last edited by wiktor256 on Sun May 07, 2006 5:32 am, edited 1 time in total.

EA
Posts: 278
Joined: Mon Feb 20, 2006 1:13 am
Location: EBSH

Post by EA » Sun May 07, 2006 1:48 am

wiktor256 wrote:The big question is: what would be the incentive for pilots to participate in a race that will not count for monthly ranking?
I can think of one: by winning the non-scored races you lower scores of competitors and therefore increasing your own chances to do better in a monthly ranking.
Although I don't agree with your idea, it still has advantages regarding the risk-taking reduction. But I just wanted to remind you why people would fly LLC even if it wasn't for monthly rankings: first, it's in the daily rankings, so if you win, you "proved" you were the best that day (I mean challenge's still there...). Second, LLC are amongst the very few races where you can fly with 31 other gliders around !!! So, don't forget the fun in LLC... As far as I'm concerned, I don't fly LLC to prove I'm the best (btw it would be a nice failure :wink: ), but because it's fun !
Tima (TSD) wrote:I agree with day factor always = 1. The reason behind is that participants have very different skills which is not the case IRL competitions. So day factor doesn't act here as it is does in the FAI rules.
You might be right on that one if you're thinking about continental and world championships (and I insist on "might"...). But regarding nationals, all skill levels can be seen, so IMHO it's no argument to skip the day factor.

My humble opinion, of course ! :wink:

Seb

User avatar
Vertigo
Posts: 1080
Joined: Fri Sep 16, 2005 11:17 am
Location: Belgium

Post by Vertigo » Sun May 07, 2006 11:01 am

Wiktor, I think thats an interesting idea -although I expect few pilots to make such a declaration, so in practice you'd see for most pilots the first 4/5 results would count, rather than the current best 4/5.

Downside of that is that by the end of the month, most pilots would have lower motivation (not flying for points), which OTOH could be exploited by those pilots that did file a declaration, or didnt fly 4 races yet. Might be interesting for tactics indeed.

IF you'd use this, you'd probably have to abolish place point penalties for crashes for non scored flights, otherwise pilots with 4 scored results might be reluctant to fly at all.

Need to think this over, but I like the idea
Image

User avatar
Vertigo
Posts: 1080
Joined: Fri Sep 16, 2005 11:17 am
Location: Belgium

Post by Vertigo » Sun May 07, 2006 11:12 am

TimKuijpers wrote:My opinionabout the 50% rule.
why 50%? I could understand it would be because some pilots only can fly on thursday/sunday
That way they always have 50%...
not in this case, because the months contain sometimes more thursdays then other months.
If you keep the 50% rule into the system, I would suggest to make a semi-month.
A semi-month would contain as much sundays as it contains thursdays.
This also means that you can fly a task in january which counts for the month february, but it would make much more sense to me.

If 50% is just choosen because some pilots can miss some races, I would prefer to make it 75% and if it comes out at 9 days, it should be 6 and not 7.
This will make the flying also a lot more serious.
With only 50% encounted you can take high risks and hoping it works out right and therby get a much higher speed.
So flying for more then 50% make the flying style more likely to RL
many reasons why we decided on the 50% rule; people not having enough time to fly 8 or 9 races each month on fixed hours, people with technical problems (especially with older versions of condor, and remember, in the beginning we didn't score IGCs yet); lastly, indeed as you mentioned, inability to regulary fly sunday or thursday.

The downsides of this approach, we are well aware of them, and if you read this thread you may find some propositions to counter this effect ;)

The "semi month" proposial sounds a bit complicated to me, and I see little advantages as I assume most people don't have a time problem each sunday or thursday, but rather several sundays and/or thursdays (or people going on a business trip for a week, etc). 75% of all races also seems too much to me, and would favour people with a lot of free time too much.
Image

User avatar
Vertigo
Posts: 1080
Joined: Fri Sep 16, 2005 11:17 am
Location: Belgium

Post by Vertigo » Sun May 07, 2006 11:18 am

Panther wrote: In addition, I would also fully punish midair collisions ('deactivate Q-key') - although in most cases they are caused mainly by one of the colliding pilots who flew like a madman, IRL it is the responsibility of everyone to take care that no collisions occur. In LLC, if pilots are e.g. extremely ruthless while thermaling, penalties should be given based on eyewitness reports.
We tried this in the beginning, it didn't work. Its impossible to judge who's fault a collission is, especially when you factor in suicidal tow pilots, bad pings making planes jump all over, and last but not least: very limited FOV compared to RL if you don't own a trackir. There already is a small penalty for colliding (lost height, no longer centered in the thermal,..).
Image

User avatar
mac
Posts: 394
Joined: Tue Mar 28, 2006 8:25 pm
Location: Strasbourg, France

Post by mac » Sun May 07, 2006 12:36 pm

wiktor256 wrote:What if instead of automatically counting the best 50% of the flights, each pilot would have to declare before the race if a flight will count for scoring or not. If a pilot declares that the flight will count for scoring, it will be scored no matter what. This should reduce risky flying and make it more like RL racing.
I am neutral/slightly negative towards the idea, but one thing that I would recommend is to keep it simple... In this specific case, I would propose that a flight will be always scored unless declared the contrary beforehand.

My reason for being slightly negative is that any kind of "planned tactics" implies that a pilot can manage freely her/his time a night, and plan the month in advance and thi is not the case for everybody...
Image

User avatar
Vertigo
Posts: 1080
Joined: Fri Sep 16, 2005 11:17 am
Location: Belgium

Post by Vertigo » Sun May 07, 2006 1:30 pm

Been thinking on the scoring some more, and came up with a radical other concept/idea that may solve the 50% problem, the high risk problem, and pretty much all others.. I hope , while complying with the KISS concept:

abolish the month ranking. Each started race counts, up to 1000 points can be won each day. but rather than adding those points, we could average them.

Of course it would be silly to be able to win llc with just one won race (1000p average), so we add 10 or so zero scores to the average.

The more races you fly, the closer you can get to the theoretical maximum score which acts like an asymptote. Fly fewer races and they will count more heavily. Say you flew just 10 races, winning all:
10 wins is 10x1000/20=500p
11 wins is 11x1000/21=523p
You gain 23 points with your day victory

Your 50th race (winnig all) will only give you a very small improvement:
49 wins is 49x1000/59=830,5
50 wins is 50x1000/60=833,3
So you gain 2.8 points

LLC 2007 will consist of ca 50 races again, but I propose to count only 35 or 40 or so (my example used all 50), first 40 races with valid race start

Tons of advantages to this approach:
the more you fly, the more points you can score, but its not linear, so missing a few races won?t hurt all that much. Still most or all your flights will count but its not the pilot with the most time that will win.

Crasing (zero points) will hurt the top pilots much more than the lesser pilots. There is no longer an incentive to take risks, but there is an incentive to participate regulary.

It makes comparison between pilots with many flights and fewer flights more fair, while giving a diminishing advantage to more results..
Image

User avatar
Vertigo
Posts: 1080
Joined: Fri Sep 16, 2005 11:17 am
Location: Belgium

Post by Vertigo » Sun May 07, 2006 1:51 pm

Afterthought: we should still apply a penalty for crashing that is worse than just zero points, otherwise landing out after 30km or crashing would be pretty much the same for high ranking pilots, so maybe give -250 points for a crash
Image

Post Reply