WCA Regulations 2008 draft open for public feedback

[Ron]

WCA Regulations 2008 draft open for public feedback.
http://www.worldcubeassociation.org/forum/viewtopic.php?f=4&t=421

 

[pjk]

Great, looks good. I like this:

9f16) For the 3x3x3 Cube: Multiple Blindfolded event the order in the results is based on number of puzzles solved minus the number of puzzles not solved (higher is better). If the result is 0 or lower, the solve is disqualified. If competitors have the same result, then the order is based on total time (lower is better).

I have a question about this:

B2d1) The competitor must remove the cover from the puzzle after starting the timer..

What does this mean?

Also, on a sidenote, is there a limit to the number of competitions that can occur in a location within a given period? If you have a group of 15 guys at a school, what is stopping creating a competition each weekend?

 

[Harris Chan]

Also, on a sidenote, is there a limit to the number of competitions that can occur in a location within a given period? If you have a group of 15 guys at a school, what is stopping creating a competition each weekend?

I heard that there is a 4 competition-per-year-per-state/provinces rule.

 

[Ron]

Please post your feedback on the WCA forum, so that everyone can see it.
It is better to keep the discussion centralized.

Thanks,

Ron

 

[AvGalen]

I would like to post there Ron, but I cannot register.
On http://www.worldcubeassociation.org...&coppa=0&sid=81dc40bf47eb5e89676afee4218d1db5 I click on the "I agree to these terms" button, but only a screen refresh occurs.

So far the new rules seem to make sense, although 8 minutes per cube for multi-blind seems a bit steep. The Dutch record has been 2/2 in 27:41, 26:20, 19:47 and 18:08. According to H1b1) When the total time is reached, the attempt is stopped and the number of solved and not solved puzzles is counted. That would have meant 1/2 in most cases. According to 9f16) For the 3x3x3 Cube: Multiple Blindfolded event the order in the results is based on number of puzzles solved minus the number of puzzles not solved (higher is better). If the result is 0 or lower, the solve is disqualified.
If competitors have the same result, then the order is based on total time (lower is better). the result would have been 1-1=0 so it would have been disqualified. I think this rule should be changed to "if the result is < 0".
Also, people that don't solve entire cubes at once (like Clément Gallet and me, we orient every cube first) would be at a disadvantage with these rules. I don't think the method of solving should have an impact on the results.

Finally, if Cube Explorer is required anyway I would like it if FMC-scrambles would be provided in optimal form

 

[ExoCorsair]

Finally, if Cube Explorer is required anyway I would like it if FMC-scrambles would be provided in optimal form

I wholeheartedly concur.

 

[Dene]

I like the new OH rule, in that observation can be done with both hands, definitely better. I wonder about the 3x3x3 scramble, though. There is no longer a limit to the amount of moves done, as long is it is in a "random position"? I could do U2, R' B L, and it would look moderately "random" to an inexperienced cuber/someone who doesn't cube at all. This is far too general.

 

[DennisStrehlau]

8 minutes per cube in multi...10 would be better, but its okay i think...

Greetings...Dennis

 

[Lucas Garron]

I like the new OH rule, in that observation can be done with both hands, definitely better. I wonder about the 3x3x3 scramble, though. There is no longer a limit to the amount of moves done, as long is it is in a "random position"? I could do U2, R' B L, and it would look moderately "random" to an inexperienced cuber/someone who doesn't cube at all. This is far too general.

What? It's been shown that the current scrambling system produces cubes that are not scrambled random enough. Yes, a 4-move scramble could show up, but 1) it's actually less likely now, and 2) the event's judge would probably remove that scramble.
And the directions are now very specific, while previous definitions were vague and general (terms were even used and not defined).
Random-state scrambling is really the est we can do, we just need to handle it appropriately.

 

[Dene]

But is "random" not one of those undefined words? The cube in its solved state could be seen as "random".

 

[tim]

But is "random" not one of those undefined words? The cube in its solved state could be seen as "random".

Random means uhm.. random. A scramble wasn't really random. The scramble itself was, but not the cube, which was scrambled with that scramble.

 

[Bryan]

Also, on a sidenote, is there a limit to the number of competitions that can occur in a location within a given period? If you have a group of 15 guys at a school, what is stopping creating a competition each weekend?

I heard that there is a 4 competition-per-year-per-state/provinces rule.

Well, all competitions need to have WCA approval, so I'm sure they would just stop approving them.

Harris, in the WCA forum I asked about this, and Ron's response was basically "If you're going to have many competitions, try to move them around a bit so many people can participate." But he didn't say there was an official limit. I believe you even responded to that post.

As for "random" position, I believe what Cube Explorer does is the equivilant of taking apart the cube, putting it together randomly (and knowing to avoid parity) and then determine the algorithm to get to that position. It's not just "mix it up randomly".

 

[cmhardw]

But is "random" not one of those undefined words? The cube in its solved state could be seen as "random".

True, but we're assuming that Cube Explorer operates on close to a uniform distribution when producing all of the possible states.

So yes, the solved state is a random state. This makes perfect sense.

However, the probability (we hope) of Cube Explorer producing the solved state is 1/43252003274489856000

So I don't see the problem?

Chris

 

[AvGalen]

The solved state is indeed one of those random states. And if it would ever be the official "scramble" I will allow it (after checking with the WCA-delegate). However, you should really think about the chances that this would happen. Please read http://kociemba.org/performance.htm for details, but to summarize:

moves Probability
0 2.31203 *10-20
1 4.16166*10-19
2 5.61824*10-18
3 7.49098*10-17
4 9.99699*10-16
5 1.32921*10-14
6 1.76141*10-13
7 2.3306*10-12
8 3.08042*10-11
9 4.06836*10-10
10 5.36965*10-9
11 7.08242×10-8
12 ~9.3*10-7
13 ~0.000012
14 ~0.00016
15 ~0.0021
16 ~0.025
17 ~0.26
18 ~0.69
19 ~0.03
20 P > 0
21..26 unknown, but presumably 0
>=27 0

Basically that means you will almost always get a scramble that will require between 16 and 19 moves
Getting a complete last layer skip > 1000's times more likely to happen compared to getting a <= 11 moves scramble.
And if you would get an 11 move scramble you would have a hard time finding that optimal solution in 15 seconds. Even in 1 hour it would be tough.
There was a 4 move scramble on 2x2x2 and only one person noticed it!

 

[Stefan]

There was a 4 move scramble on 2x2x2 and only one person noticed it!

And I believe even he noticed it only afterwards.

 

[Stefan]

Oh and the solved state is *not* a random state per se. The term "random state" is just an abbreviation for something like "randomly produced state" and thus only makes sense in the context of a producer, for example Cube Explorer.

I am however not sure that Cube Explorer is nearly perfectly random. It's natural for programmers to encode the solved state with the number zero. Now if Cube Explorer generates a number and interprets it as a cube state, and its random number generator uses lets say just 16 bits, then there are at most 65536 numbers it can generate, and the number representing the solved state would (probably, I believe) be among them. In other words, we might have a 1/65536 chance for it to give us the solved state, instead of the desired 1/(4e19).

I somewhat remember we had a discussion and analysis about the quality of Cube Explorer scrambles a few months ago. Possibly at the yahoo group, though. Does anyone remember?

 

[Bryan]

Finally, if Cube Explorer is required anyway I would like it if FMC-scrambles would be provided in optimal form

Wouldn't FMC scrambles be the candidate for not being provided in optimal form? What happens if a competitor comes along, and calculates out the optimal path, which just happens to be the reverse of the scramble? Or someone uses the scramble, but inserts an identity or a slight scramble in the middle, and then fixes the slight scramble at the end. Seems like you're giving the competitor the answer and it would be very difficult for a judge to recognize if they're cheating or not.

 

[alexc]

8 minutes per cube is a little extreme. I think 10 or 11 is more reasonable. For multi I mean

 

[Pedro]

about the 8 mins/cube rule:

I just went to the results page, and here are the 28 first ranked (the ones who got all cubes correct)

Name - number of cubes - under 8 min/cube?

Dennis - 10 - NO
Istvan - 7 - NO
Rafal - 6 - YES (2 seconds)
Oliver - 6 - NO
Clement - 5 - NO
Constantin - 5 - NO
Stefan - 5 - NO
Rowe - YES
Milan - NO
Bastien - NO
Kai - 3 - YES
Bernett - 3 - NO
Kare - 3 - NO
Eric - 3 - NO
Andy - 2 - YES
Edouard - 2 - YES
Leyan - 2 - YES
Sinpei - 2 - YES
Daniel - 2 - NO
Piotr - 2 - NO
8 more - 2 - NO

so, just 7 out of 28, which is just 25%...

I think 8 minutes is too short...10 would be good

 

[tim]

On the other hand, i think many of these guys would've gotten sub-8 per cube, if they knew that rule.