[WCA Regulations 2014] Mean of N for FMC

[Sebastien]

That's the only point I could agree with. But it's not as bad as it might seem. You need one judge for each competitor and one mechanical counter. 10 minutes (the upper bound of my proposed time limit) is shorter than your average Multi/Big Cubes BLD attempt (where you theoretically also have look at your competitor's cube during memo). By using cutoffs (Best of 2 / Average of 5 (< 60 moves)) like we already do you can also quickly reduce the amount of required judges.

Makes about an hour per group. Sounds very much impractical to me.

Not sure if you think I was implying that, but as I've already said, competitions would still be based on singles in the same way. There would just be a new WR and a new set of world rankings. It is already an interesting statistic, but is there any reason not to make it an official interesting statistic?

But is there a reason to do so? The single ranking displays skill well in BLD, so there is no other ranking needed.

However, I don't believe mean of 3 is practical either due to time restrictions. Some comps may have it, but entire countries will go without it indefinitely.

I fully agree, even though I'd be happy if a Mean of 3 would be allowed, even it would be used not very often, as long as major important competitions like EC or WC would use this format to determine the winner.

Another option would be a "Mean of 2", which has again other issues.

Either way, I'd be happy with one of both options to be implemented.

Are we honestly still bothered by perceived "luck"?

Can you precise this?

 

[Kirjava]

I understand wanting to minimise 'luck' by using different formats, but luck is an intrinsic aspect of any event and cannot be eliminated. Is adding a format that most people do not have access to and takes an inordinate amount of time to execute really worth it to attempt to decrease it?

Some events are inherently more difficult to implement than others - FMC and multi are the worst.

 

[Sebastien]

Some events are inherently more difficult to implement than others - FMC and multi are the worst.

That's the issue. If it would be different, we would have an average for FMC for a long time already. Now FMC is the only event where luck is an intrinsic aspect, which is not ranked by some kind of average. This is obviously bad, so I and others try to seek for a way to fix the event.

 

[Kirjava]

Now FMC is the only event where luck is an intrinsic aspect, which is not ranked by some kind of average.

BLD?

 

[Sebastien]

just no.

http://www.speedsolving.com/forum/s...n-of-N-for-FMC&p=846680&viewfull=1#post846680

 

[Kirjava]

Sorry, are you trying to claim that luck has no effect on BLD whatsoever?

 

[Sebastien]

I claim that luck is no factor in BLD in a way, that makes ranking BLD by singles inadequate.

 

[Kirjava]

Possibly, but BLD is still an event where luck is an intrinsic aspect, which is not ranked by some kind of average.

 

[qqwref]

http://en.wikipedia.org/wiki/False_dilemma

Of course BLD involves luck. Maybe FMC involves more luck - maybe not - but they're very different things and neither are consistent.

 

[Cubenovice]

In BLD there is luck as in "some pieces already solved" or "lol, these three edges are a U perm from solved" but there is no luck as in "skipping a step".
For the same scramble all BLD solvers will have the same cycles to go through.

Another way to look at it::
In BLD you will not unexpectedly find yourself with a solved cube after having solved all corners. Your solving steps did not affect edges (ib4 parity...)
In a very lucky FMC solve this *could* happen.
I now TMOY would love this

 

[Noahaha]

[WCA Regulations 2012] Mean of N for FMC

I didn't mean to start a BLD vs FMC argument here. I think both events would benefit greatly from ranked mo3s. I should have said that BLD mo3s would be easier to implement given that all that would happen would be a new page on the WCA site.

I think the argument that mo3s are not needed in BLD is fairly invalid unless you discount the necessity of mo3s in 6x6 and 7x7... the single rankings are plenty to rank people in those events but it's more FUN to have both. Do we really have to have a 100% utilitarian purpose for everything? Is it not enough for something to be completely harmless, no work to implement and desired by many people? I'm sure I'm not the only one who wants this.

My personal reason for wanting it is so that competing in 3BLD is more fun. I go into competitions hoping for one nice scramble. At my last competition I got one, but at others I have not. Adding ranked mo3s would give me another goal that can actually be attained with any set of scrambles.

 

[okayama]

Remark: if the format is best of 3 to decide the winner of the competition, you cannot get any mean result.

For example, let us assume that 3x3 Feet is held with best of 3 format.
In this case, competitors do not get average (see such an example),
even though mean of 3 can be calculated from the results.
In order to get average results, the organizer must choose "mean of 3"
in the competition.

Therefore, as long as "best of 3" is used for 3x3 BLD to decide the winner,
you cannot get any average record, even if "mean of 3" is officially approved.

 

[qqwref]

I consider that a problem with the way things are set up, not a universal constant that cannot be changed. There is no practical reason why we cannot record an average while ranking people by the best single solve.

 

[okayama]

There is no practical reason why we cannot record an average while ranking people by the best single solve.

I agree. I don't know why the current system does not calculate average in that situation.
(But this seems off-topic... sorry for my interruption)

 

[ErikJ]

I posted a thread about this quite a while ago and it didn't get much attention so I'm glad to see some really talented FMCers going after the issue. I approve :tu

anyway, here are my thoughts...

From my experiences in WCA and "no time limit" FMCs, I think the problem is the time limit. When an FMCer is given more time they will have more opportunities to find the lucky cases. obviously the best way to prevent them from trying to find lucky cases is to make the event a linear FMC. This however is not possible in competition for reasons I don't think I should have to explain. Linear style solving would be the extreme way to fix FMC but since it is not possible, a compromise must be made:

In my opinion, the best way to fix FMC is to decrease the time limit thus making it feasible to do more than one attempt. I believe adjusting the time limit (up or down) involves a trade off. When a solver is given less time, they have a smaller probability for finding the lucky cases (ex: solving F2L pairs in different orders) but they will also have less time to explore complicated techniques involving insertions and inverse scrambles.

I propose an FMC event in which solvers are given at least two scrambles and must solve all of them within one hour. This adds an interesting concept of time management while adding the ability to take a mean and not increase the duration of the event. Since there will be less time for solving, It will be harder to achieve the move counts we are used to seeing on the podium. This brings up some fundamental changes that would really help FMC:

1. Raise the skill ceiling

2. Widen in skill gap

Quick definitions:

Skill Ceiling: the best score someone can possibly achieve

Skill Gap: the difference in scores between noobs and pros


In the current FMC format, it isn't hard for the average CFOP solver (lesser skilled) to get 40s or even high 30s while the best of the best FMCers can get mid 20s to low 30s. The problem with letting lesser skilled solvers get scores that close to the top solvers is that they are only a PLL skip away from a podium score. If the time per solve is decreased the scores of all solvers will become worse. This in effect raises the skill ceiling by making it require more skill to achieve low move counts. The other effect the new time constraint will have on scores is the move counts of the lesser skilled solvers will increase much more than those of the truly skilled FMC solvers. Thus, the skill gap is widened, which means a PLL skip wont become a podium solution. these concepts are evident in the results of online linear FMCs (assuming the results are legitimate).


sorry for all the text. I hope I made some good points. have a nice day.

 

[mycube]

I think the change of the event FMC in your way would change it much more to a luck challenge.

From my experiences in WCA and "no time limit" FMCs, I think the problem is the time limit. When an FMCer is given more time they will have more opportunities to find the lucky cases. obviously the best way to prevent them from trying to find lucky cases is to make the event a linear FMC. This however is not possible in competition for reasons I don't think I should have to explain. Linear style solving would be the extreme way to fix FMC but since it is not possible, a compromise must be made:

In my opinion, the best way to fix FMC is to decrease the time limit thus making it feasible to do more than one attempt. I believe adjusting the time limit (up or down) involves a trade off. When a solver is given less time, they have a smaller probability for finding the lucky cases (ex: solving F2L pairs in different orders) but they will also have less time to explore complicated techniques involving insertions and inverse scrambles.

Not just the FMCer have the chance to try move opportunities but also everyone else. The more time they have the more they can find a PLL skip and a shorter solution. And I think the lucky solves from non-FMCer were found in less than half an hour so a decrease of time would change just nothing.

I propose an FMC event in which solvers are given at least two scrambles and must solve all of them within one hour. This adds an interesting concept of time management while adding the ability to take a mean and not increase the duration of the event. Since there will be less time for solving, It will be harder to achieve the move counts we are used to seeing on the podium. This brings up some fundamental changes that would really help FMC:

1. Raise the skill ceiling

2. Widen in skill gap

Quick definitions:

Skill Ceiling: the best score someone can possibly achieve

Skill Gap: the difference in scores between noobs and pros

In my opinion this just changes the movecount of FMC upward because the most FMCer need for their solutions for one scramble the full hour. And why should we change the way of doing FMC fundamental? We just need a way to do it like we do it now with a better balanced result system.
And I think, your idea just would give the luck solutions more importance. Many people would try to do a first, short and lucky solution and use the remaining time to find a really good solution. I can't imagine to do two real FMC Solutions within an hour which will be sub30 without a lot of luck.
The skill gap would not appear because everyone just can search for two PLL-Skip (or whatever) solutions to have a good mean. Luck would be a much more important aspect

In the current FMC format, it isn't hard for the average CFOP solver (lesser skilled) to get 40s or even high 30s while the best of the best FMCers can get mid 20s to low 30s. The problem with letting lesser skilled solvers get scores that close to the top solvers is that they are only a PLL skip away from a podium score. If the time per solve is decreased the scores of all solvers will become worse. This in effect raises the skill ceiling by making it require more skill to achieve low move counts. The other effect the new time constraint will have on scores is the move counts of the lesser skilled solvers will increase much more than those of the truly skilled FMC solvers. Thus, the skill gap is widened, which means a PLL skip wont become a podium solution. these concepts are evident in the results of online linear FMCs (assuming the results are legitimate).

No, the movecount of the lower skilles FMCer would not change in this way. It would be still nearly the same.
Yes, this is true. You really need more skill to get the same results such as in an hour. But how many of the good FMCer are able to do this twice in an hour?

 

[Sebastien]

I'm intending to push towards "Mean of 3" being a possible format for FMC in 2014's regulations, just like for 6x6x6, 7x7x7 and feet. Nowadays ~30% of the competitions hosting FMC have 2 attempts or more. I imagine that the number of competitions having 3 attempts would increase if this gives the possibility to the competitors to get a result for one more ranking. Especially major competitions should use a Mean to etermine the winner of FMC.

Any thoughts/comments?

 

[Mike Hughey]

I would love that too. It certainly would have been nice to have had that at the World Championship this year. (Although fairness dictates that I mention you would have probably been world champion this year with that format! 27.5 mean of 2 going into the third, with your historical performance, would have been very hard to beat!)

 

[kinch2002]

In summary, yes please!

I would certainly like to see a mean format in major championships. I don't mind in smaller competitions, because I don't mind if somebody gets lucky and wins - that's just pretty cool. But in a major championship, I go in feeling like I need to get some luck in order to beat the inevitable few who get really really lucky (given there's hundreds of people doing it). A mean of 3 format
would leave me feeling a lot more satisfied that the winners deserve their place on the podium. Even a mean of 2 would make me considerably happier.

As a side note, I couldn't even be bothered to do my insertion on the second attempt at Worlds (hence the DNF), because I knew I was highly unlikely to beat my 29 in the first round, and even that wouldn't get me on the podium. If it was a mean of 2/3, I certainly would have finished and gotten a 30/31 mover, keeping me in the hunt for medals.

 

[okayama]

I'm intending to push towards "Mean of 3" being a possible format for FMC in 2014's regulations, just like for 6x6x6, 7x7x7 and feet. Nowadays ~30% of the competitions hosting FMC have 2 attempts or more. I imagine that the number of competitions having 3 attempts would increase if this gives the possibility to the competitors to get a result for one more ranking. Especially major competitions should use a Mean to etermine the winner of FMC.

Any thoughts/comments?

Great, but how about Median of 3 if possible?
For example:

• Which competitor should win: (a) 28, 29, DNF, (b) 30, 40, 50
• Which competitor should win: (a) 28, 29, 31, (b) 21, 32, 34

I hope (a) wins with the result 29.
Related comments for the first example: cuBerBruce and irontwig