2009: 2x2x2 scrambles using random state
MadsMohr (2008-11-12 08:42:25 +0000)
I think that we should use random state scrambles for the 2x2x2, just like we do on the 3x3x3. You could just apply the 3x3x3 scrambles or ask Herbert Kociemba to update his WCA scrabmler in Cube Explorer.
Erik (2008-11-12 10:58:29 +0000)
I'm in favor of this, it would also speed up the scrambling procedure since you only have to perform 8 or 9 moves on average
![Smile :)]({SMILIES_PATH}/icon_smile.gif)
PatrickJameson (2008-11-12 11:34:36 +0000)
The one thing that I have thought of with this is the possibility of cheating. If the scramble is 6 moves, people could easily get one glance at the scramble and practice it. I'm not sure if I'm in favor of this or not.
Edit: Also, the scrambles are already very easy to follow with 3gen
Kenneth Gustavsson (2008-11-12 12:30:39 +0000)
I'm against it becuse it reveals a short solution that is easy to remember, that does not happen on 3x3x3, 19 (normally) turns is way to much.
Gustav Fredell once showed (by scripting a huge number of examples into Acube) that 15-16 turns gives the same result as 25 turns in 99% of the cases for 2x2x2.
Why not move down to 16 turn scrambles instead of random position?
Erik (2008-11-12 12:56:01 +0000)
How does it reveal a short solution? Competitors don't see the scramble sheets during the rounds.
MadsMohr (2008-11-12 13:02:25 +0000)
[quote="Kenneth Gustavsson":1xu0v0vr]I'm against it becuse it reveals a short solution that is easy to remember, that does not happen on 3x3x3, 19 (normally) turns is way to much.
Why not move down to 16 turn scrambles instead of random position?[/quote:1xu0v0vr]
This makes no sense. How would you remember a solution that you are not permitted to see before solving?
The length of the scramble does not guarantee a random state, that's why we should use random state and generate the scramble for that state. Note that you wont guarantee non-lucky scrambles when using random state, but it should remove bias included in the scramble algorithm.
Tim (2008-11-12 15:45:31 +0000)
If competitors are able to see scramble sheets, that is a large issue. Similarly, should we get rid of the scramble pictures on the scramble sheets? One glance at them (on 2x2 for example) could tell you a lot about the solve. Competition organizers should be careful about not letting competitors see the scrambles. But we shouldn't be making scramblers do 3 times the necessary number of moves just in case the scrambles might be publicly visible during the round.
PatrickJameson (2008-11-12 16:16:06 +0000)
[quote="MadsMohr":47u7469a]This makes no sense. How would you remember a solution that you are not permitted to see before solving?[/quote:47u7469a]
Of course you're not 'supposed' to see the scramble, but how most competitions are run now it is possible to peak at the scrambles. Maybe if we enforce the 'competitor area' rule in the regulations better we can do this. As the competitions are now though, it doesn't seem very rational.
Shelley (2008-11-12 16:55:36 +0000)
COMPETITORS SHOULD NEVER BE ALLOWED TO SEE SCRAMBLES. If competitors who have yet to compete have access to scrambles you are doing something wrong. That should never be an issue.
Kenneth Gustavsson (2008-11-12 19:54:42 +0000)
Allowed?, of corse not!
But did you hear about cheaters??
I someone like that has got the chance it's far easier to remeber a 8 turn set-up than a 16 turn scramble.
MadsMohr (2008-11-12 21:44:55 +0000)
[quote="Kenneth Gustavsson":11ffcv8d]Allowed?, of corse not!
But did you hear about cheaters??
I someone like that has got the chance it's far easier to remeber a 8 turn set-up than a 16 turn scramble.[/quote:11ffcv8d]
We should enforce the current rules regarding access to scramblers and competitors area more strictly and make the scrambles easier to apply and of better quality. Changing rules to fit cheaters are the wrong way to go and if it really is that easy to cheat then perhaps we should go to the extreme and retire the event? (i don't want that)
BryanLogan (2008-11-12 22:10:40 +0000)
I don't think the "competitor's area" would be helpful here, because people who have not been called up can't be in that area. Basically, you just have to state that no one can be in the scrambler's area.
In my competitions, the scrambling tables have been located behind the timers. Actually, the setup I have at the Minnesota Open is quite nice. The scrambling table is behind a curtain behind the timers, so it's not even visible to competitors.
PatrickJameson (2008-11-12 22:11:25 +0000)
[quote="MadsMohr":n969ad3x]
We should enforce the current rules regarding access to scramblers and competitors area more strictly and make the scrambles easier to apply and of better quality.[/quote:n969ad3x]
I don't think there's anything wrong about the scrambles we have now. 8 move scrambles wont save [i:n969ad3x]that[/i:n969ad3x] much time compared to the 3 gen scrambles we have now. We should enforce the competitor areas more though.
Maybe we should have a [i:n969ad3x]scramblers[/i:n969ad3x] area instead. Only judges and scramblers are aloud to be in the area. We might be able to enforce it better.
MadsMohr (2008-11-13 10:11:33 +0000)
The scrambles doesn't need to be bad in order to get better
![Wink ;-)]({SMILIES_PATH}/icon_wink.gif)
TomZ (2008-11-13 11:30:06 +0000)
I think that the scrambles shouldn't be optimal, near optimal is fine. I reckon if a 4-move scramble appears, it would cause a lot of commotion. So I think that the minimum for a 2x2 scramble should be 9 moves, regardless of optimality.
MadsMohr (2008-11-13 11:46:39 +0000)
[quote="TomZ":2xsy0aqu]I think that the scrambles shouldn't be optimal, near optimal is fine. I reckon if a 4-move scramble appears, it would cause a lot of commotion. So I think that the minimum for a 2x2 scramble should be 9 moves, regardless of optimality.[/quote:2xsy0aqu]
You have a good point there. Seems like a very good compromise to me.
Kenneth Gustavsson (2008-11-28 01:45:43 +0000)
Even if I said it may not be a good idea earlier I will actually try this at the Cube Day to save some time. We got 2x2x2 OH and that is not a WCA event so it will not violate any rules. This way we will have it tested, see what people thinks about it in real action =)
StefanPochmann (2008-11-28 09:39:30 +0000)
[quote="Kenneth Gustavsson":3hn95n79]If someone like that has got the chance it's far easier to remeber a 8 turn set-up than a 16 turn scramble.[/quote:3hn95n79]
Yeah so let's play it safe and use 100 turns.
[quote="TomZ":3hn95n79]I reckon if a 4-move scramble appears, it would cause a lot of commotion. So I think that the minimum for a 2x2 scramble should be 9 moves, regardless of optimality.[/quote:3hn95n79]
Interesting point. Such a short scramble could even be noticed from large distance if people can see the judges (a careful observer wouldn't even need to see the cubes, time of the judges picking up a cube until laying it down could suffice). I believe it's possible and fairly easy to produce non-canceling 11 moves scrambles for all cube cases, so that all 2x2 scrambles are exactly the same length.
jbcm627 (2008-11-28 18:59:36 +0000)
[quote="StefanPochmann":pfvp1abh]I believe it's possible and fairly easy to produce non-canceling 11 moves scrambles for all cube cases, so that all 2x2 scrambles are exactly the same length.[/quote:pfvp1abh]
Could this be added as a feature to Tim's [url=http://www.worldcubeassociation.org/forum/viewtopic.php?f=4&t=446:pfvp1abh]already existing program[/url:pfvp1abh]?
Tim (2008-11-29 06:31:05 +0000)
Uh I could try...honestly I haven't looked at the source code in months but I'll work on it.
TomZ (2008-11-30 12:33:58 +0000)
I have altered Jaap's 2x2x2 simulator to take a random position and make a generator for it. The generator has at least 9 moves, at most 11.
The scrambler: http://zandenonline.nl/22scrambler/
Source: http://zandenonline.nl/22scrambler/source.zip
It's written in JavaScript, so it is portable. Currently it randomizes the cube by doing 100 random moves - is this random enough?
Does anyone think a lower or higher lower boundary should be used for the generator?
[quote="Scrambler Requirements according to Ron":35tbhvz6]Requirements:
- fast generation of random positions and the required move sequences to get there
- not too lengthy move sequences (<=70 moves [note: this quote was taken from a topic about SQ1 scrambles]) and not significantly longer than current sequences
- image of each generated random positions, with print option
- preferrably available on all platforms
- preferrably one program that handles all puzzles
- offline use[/quote:35tbhvz6]
Edit: By request I have also uploaded optimal and 11 move minimum variations:
- 11 moves minimum: http://zandenonline.nl/22scrambler/11moves/
- 0 move minimum: http://zandenonline.nl/22scrambler/optimal/
StefanPochmann (2008-12-07 17:54:33 +0000)
Looks good, could this become the official scrambler?
Herbert also mentioned he could build this into his Cube Explorer, though I'm not sure exactly how and whether he'd do it before knowing whether it would be used.
Btw, link to the spin-off thread about this in the speedsolving.com forum:
http://www.speedsolving.com/forum/showthread.php?t=7605
TomZ (2008-12-07 18:48:59 +0000)
Reply by me on the SpeedSolving.com forum:
[quote:3w2kk0gn]I am not sure on the 11 move limit. I think if a scramble can be generated in N moves, it must also be generatable in N+M moves.
There would be (3*3)*(2*3)^10 different 11 move scrambles, which means that if the scrambles were distributed evenly, each scramble would have about 150 generators.
Of course this isn't a proof, but there is an easy workaround: set the upper bound to 20, so that if no 11 move generator exists, it will find a generator with more moves. That would take care of the problem. And as it quits searching once a generator is found, it wouldn't present a 12 move generator if a 11 move could be found.
This one begins the search at 11 moves, and ends at 20: http://zandenonline.nl/22scrambler/20upper/[/quote:3w2kk0gn]
Ron (2008-12-21 15:49:45 +0000)
Hi Tom,
Thanks for creating this solution for 2x2 scrambling.
Would you be able to convert your solution to the standard format of the scramble program?
Now the program uses several frames.
Ron
TomZ (2008-12-28 14:53:23 +0000)
Ron,
I think I can, but I don't think I can finish it before the new regulations.
Tom
Ron (2009-01-03 15:30:07 +0000)
Tom,
That is OK, we can change it later during 2009.
Ron
Bob (2009-01-05 01:00:09 +0000)
[quote="Ron":4lh2z0p9]That is OK, we can change it later during 2009.[/quote:4lh2z0p9]
Why not use Tom's for now, and he will eventually change it some time soon. When he does, we use the updated one. This will save 14 moves per scramble. Even scrambling at 2.5 tps, this saves almost 6 seconds per cube, which amounts to nearly 30 seconds per competitor. In a competition with 40 competitors, this saves 20 minutes of scrambling.
Bob (2009-01-25 21:09:31 +0000)
Would it be a problem to use this new scrambler for 2009 even though the format is not yet ideal? If it has been shown to provide a random state, I think it is at least better than the scrambler we use now.
blade740 (2009-01-26 21:59:57 +0000)
Cube Explorer is not the "correct format" and yet it is used because it provides a better scramble. I see no reason not to use Tom's scrambler as well.
Ron (2009-02-04 21:51:13 +0000)
For Regulations 2009 Draft 3 I added the OPTION of a Random position generator for 2x2x2.
Cube Explorer and Tom's solution are approved for that purpose.
For the time being I would not like to recommend the latter in the regulations.
Once the solution is finished I will change it in the regulations.
Thanks,
Ron
TomZ (2009-02-06 14:57:22 +0000)
Dear Ron,
I have adapted my scrambler to use the standard layout: http://zandenonline.nl/22scrambler.html
(Keep in mind that loading the page takes a while, as the browser needs to generate all the scrambles. It appears Google Chrome is the fastest at this.)
Ron (2009-02-06 15:05:57 +0000)
Great work Tom and in time for Regulations 2009!
Thanks,
Ron