1 Hour FMC is a Luck Based Event

[masterofthebass]

both 27s were PLL skips by mediocre FMCers and bambozzled you and I out of the podium. another example is from armonk where bob beat my 39 with a 36 because he had the 6 move OLL and a PLL skip. since those results place him at the top of that competition then that means he was the best FMCer there which certainly isn't true.

First of all, if you think I am a better FMCer than David Woner and that he is "mediocre", you are sadly mistaken. You have seemed to overlook his 27 at USNats 2010 as well.

Second, perhaps if you went to more competitions with FMC, you'd have a higher chance at not getting undercut by 1 or 2 lucky solves. Look at the Worlds 2011 podium. The people in the top 5 are people who are genuinely good at FMC. You just seem to have gotten a little but unlucky with some people beating you, but a 39 solve is really nothing to be proud of when you compare it to people like Sebastian who are consistently sub 30.

 

[ErikJ]

but a 39 solve is really nothing to be proud of when you compare it to people like Sebastian who are consistently sub 30.

was sebastian sub 30 was 4 years ago? that's when I got the 39 hahaha

This has been discussed before. What I got out of those discussions was that making competitors do *more* solves in an hour only forces a good FMC expert to spread out her/his effort too much, while not really reducing the potential for luck to determine the result. It's no longer about finding a good solution, but gaming the scrambles to find the best way to spend your time on them.

Also, requiring more solves to be written down (e.g. all 3 for mean of 3) is a lot of pressure, and more room for accident. If not all the scrambles need to be solved (e.g. median, average of best 2, etc.), it's back to luck on which scrambles the competitor chose to focus on.

I'm not saying that it has to be a mean of 3. it could be average of 2. even keeping it one solve but lowering the time limit to 30 minutes might reduce luck. I really don't know what's best for the event. but one solve per hour really is not a good way of doing it. Let's just try some different things and see if we can make it better.

 

[masterofthebass]

was sebastian sub 30 was 4 years ago? that's when I got the 39 hahaha

39 wasn't in the top 100 solves in at the end of 08. You still had people like Guus and Gzregorz and Mirek and Zbigniew getting solves ~31 "consistently". There's no reason to be so upset because Bob got 1 lucky solve and beat you with it.

 

[Tim Reynolds]

Strongly disagree.

Comments: You're putting too much faith in the 3x3 average rankings. Getting two PLL skips in an average is not unreasonable; I am sure there are people whose best average was extremely lucky and thus are not a correct indication of skill. The same is true in the other direction; there are certainly people who have not gotten good averages because of the particular format we use, and thus are ranked worse than they actually are. If you think we should look for a better system for the WCA rankings besides just taking each person's best time, I agree, but I haven't seen any good systems yet. What happens when someone gets lucky twice in one competition? Then they're "the best in the world", and nothing is fixed. In other events where single solves can be affected greatly by luck, the average rankings still aren't foolproof. Just ask Dan about the 2x2 average rankings.

Individual competitions don't need to replicate the WCA rankings. Just because someone beat you at a competition doesn't mean they're better than you; conversely, just because you came in last at Cornell Spring 2008 with a DNF doesn't mean you were the worst person there.

DNF average-of-2s are going to be very common, and people who are good at FMC are going to get them. This will show them as the "worst" people there, while people who do two CFOP solves will beat them by not DNFing.

I've got more to say, but I'll leave it at that for now.

 

[irontwig]

I think 1 hour is a pretty good balance between being long enough for "advanced" techniques and not being screwed if you made a bad choice at the start (like linear solving) and short enough to not allow for too much just fudging around hoping to get lucky (like one week/unlimited time solves).

 

[clement]

Go to France
http://worldcubeassociation.org/res...Results&competitionId=PuydeDomeOpen2011#333fm
http://worldcubeassociation.org/res...etitionId=ClermontOpen2009&allResults=1#333fm

 

[Gaétan Guimond]

The 2x2 is the best example of that skill and knowledge isn't the speed record. It makes him look bad the rules of the competition and gives a controversy in the way it sets the rules for a record. It feels like cheating when a competitor at the chance to resolve it with half the average number of movement to solve a 2x2 (fmc).

Already no human has yet proven to be able to find the fewest move on each of scramble of 2x2. Its says it all!.

GG

 

[ErikJ]

Go to France
http://worldcubeassociation.org/res...Results&competitionId=PuydeDomeOpen2011#333fm
http://worldcubeassociation.org/res...etitionId=ClermontOpen2009&allResults=1#333fm

look at the history of both people who took first place. I don't know how good that actually are but from their previous results it looks like they got lucky and had better results than the solvers bellow them. there are some very skilled fmcers in those results but they got out lucked.

the issue with fmc is that getting lucky helps you too much. let's talk about PLL skips: let's say that dan and rowe are both having great days. rowe gets five 8.00 second solves. dan gets three 9.00 second solves and two PLL skips which were 7.00. rowe's average would be 8 seconds and dan's average would be 8.33. getting two PLL skips is very lucky but it only improved his average by about 9%.

now lets say someone tends to be around 40 moves in FMC. all they have to do is get one PLL skip and their solve will be reduced down by at least 10 moves. just by getting lucky once, their result is improved by 25%.

this is a classic case of a small skill gap. a skill gap is just the distance in results between noobs and pros. 3x3 has a rather large skill gap which is why there is a huge range of times. but FMC only tends to range from 25-60 moves and when you factor in how much getting lucky can help, the skill gap becomes almost non existent.

you have to admit that FMC is the worst event when it comes to being rewarded for getting lucky. even worse than 2x2 because at least 2x2 has an average. I don't see any harm in trying a few different formats online.

 

[ThomasJE]

I like your idea of three scrambles in an hour, but I think it should only be two. The world record is 22 twice, and most winning sequences are around 25-35. It could be quite clear that the 22's was a bit lucky. Because of there being two scrambles, lucky solutions are filtered out a bit. Here's an example:

Person A: 1st solution = 21 moves | 2nd solution = 35 moves | Mean of 2 = 28 moves
Person B: 1st solution = 26 moves | 2nd solution = 28 moves | Mean of 2 = 27 moves

Person A gets a new single WR for solve 1, and messes up the 2nd.
Person B is much more consistent, and has a better mean.

This way, it favours consistant people, it filters out lucky solutions more, time management is a bigger factor and it becomes less than a 2x2 single - if the solution is short, then you get a very good time.

DNF average-of-2s are going to be very common, and people who are good at FMC are going to get them. This will show them as the "worst" people there, while people who do two CFOP solves will beat them by not DNFing.

I have a way of getting around that a little, but not totally. The order of results could be as follows:
1) Mean of 2 (lower is better)
2) If only one solve is completed, then the one solve decides the order (lower is better)

Here's an example:
Person A: 1st solution = DNF | 2nd solution = 21 moves | Mean of 2 = DNF (21 single)
Person B: 1st solution = 26 moves | 2nd solution = 28 moves | Mean of 2 = 27 moves
Person C: 1st solution = 29 moves | 2nd solution = DNF | Mean of 2 = DNF (29 single)
Person D: 1st solution = 24 moves | 2nd solution = 29 moves | Mean of 2 = 26.5 moves

So the order would be this:
Person D
Person B
Person A
Person C

To avoid the DNF, you could just write a 40 move solution for both in 10 mins, so you don't DNF, then you can get the moves down. I don't FMC, but that's just an idea.

 

[bamilan]

Basically your method is a mean of 2.

 

[aronpm]

I don't FMC, but that's just an idea.

This basically undermines everything that you just said, because you clearly don't understand it.

 

[Pyjam]

now lets say someone tends to be around 40 moves in FMC. all they have to do is get one PLL skip and their solve will be reduced down by at least 10 moves. just by getting lucky once, their result is improved by 25%.

Even if you get PLL-skip every time, Tomoaki Okayama will beat you every time.
There's no place for a PLL in a good FMC solution.

 

[Escher]

I'm sorry but why is this even a question?

Is cubing really that young that we still moan about 'luck'?

It's an almost inescapable fact that if we have systems to interpret cube data then some situations within that system will be easier than others. It's up to the individual to maximise the number of 'cases' they know for a system, or the number of systems they know. Deriding the ranking system because it does not 'reflect the best' is ridiculous in a game like this, with the limitations we have in testing ability.

I think it would be nice to develop a weighted ranking system or point-scoring system based on a) ranking achievements and b) consistency, but imo in order to support that in a way that would be useful we would need a much larger competitor pool (i.e. it should be common for small countries to have >12 competitions per year).

 

[irontwig]

Have heard anything from Fredlund? If not you can just try to make a competition thread, I for one is a interested in trying two scrambles in one hour.

 

[MaeLSTRoM]

...if the solution is short, then you get a very good time...

The people who are good at FMC spend a long time trying to get a short solution, hence why the hour is needed.
More time = More thinkting about insertions/cancellations = better solve.