Faster Scrambling Idea?

[qqwref]

ROF2L

who can sub 15 a 4x4 first

I'm not sure anyone can turn a 4x4 that fast.

I call it for now, with my crappy time:
39.50

BEAT ME NOW.

I got a 23.77. But it was a good solve.

 

[Ton]

4x4- 2 scrambles, 1st outer tier and then 1st inner tier

Than solve it like a 2x2 ( for the centers+edges)- so you skip the edges, so only use Rw, Lw etc, than solve it like a 3x3x3 ( no parities)

I'd like to see someone do that, that's essentially the equivalent of exactly reversing a scramble on any solve, which would require an absolute Genius to do or someone with an EXTREMELY complex method in my opinion. If this still seems like a problem, then the scramble length can be extended, as still the method of scrambling still in my opinion is much more efficient and faster than what is in use officially today.

Well no, If you can solve a 2x2, and a 3x3 , just try it

 

[Stefan]

Um, there even used to be UWR lists for this.

Search for "Solving like" on this page:
http://web.archive.org/web/20041210111946/www.speedcubing.com/records/recs_fun.html

There were four lists, for 4x4 and 5x5 and both inwards and outwards.

 

[Swordsman Kirby]

Rubik's Cube 3x3x3: Solving using only permutations

Huh... doesn't seem to difficult to me?

 

[whauk]

i get sub40 4x4 averages with this scrmbling type. and if i take lucky/easy scrambles i can do 4x4 sub10!!!!!!

 

[Ton]

Try this (only for a 4x4)

 

[blah]

i can do 4x4 sub10!!!!!!

Uh huh. (message too short)

 

[Tyson]

Haha, I enjoyed this thread. I sat there reading "I'm pretty sure there's a problem" and was glad I was correct as it revealed itself in the thread. But that doesn't mean we can't think about this some more. I think the original idea has some merit, if perhaps only in execution and not practicality.

So... let's ignore this problem for a moment where the cube is easier to solve. Don't you guys think that, for example, on a 5x5, that 35 double-layer turns followed by 35 outer-layer turns is faster than a 60 move sequence?

Maybe not... *IF* you're a big cube solver. But for someone such as myself, I think it might be faster.

Of course, the pure way to do this is just a random state generator, but I don't know how good computer programs are for solving efficient generating algorithms for big cubes.

 

[Dene]

I think this is a bad idea. If the problem is that scrambles are too long - THERE IS A GOOD REASON FOR THAT.

 

[Stefan]

I don't know how good computer programs are for solving efficient generating algorithms for big cubes.

They aren't.

After the big cube scramblers were announced on the WCA forum recently, I started writing about some ideas I had. Didn't finish it back then, but now I polished it a bit so here it is:

I have a suggestion for alternative 7x7x7 scrambling, though I admit I have no idea about the quality of the scrambles compared to the "normal" method:

Produce five 3x3x3 scrambles using Cube Explorer. Apply them to the 7x7x7, for the first 3x3x3 scramble making single layer turns, for the second 3x3x3 scramble double, then triple, then double, then single. So 1-2-3-2-1. Could be 3-2-1-2-3 instead. No idea what's better.

It would make the notation easier and Cube Explorer's 3x3x3 scrambles do almost maximize the "chaos per turn" at least for the 3x3x3, so it could be high quality scrambling for the 7x7x7 as well.

Actually... here's another idea: Get those five 3x3x3 scrambles from cube explorer, make sure they don't cancel. You get a sequence of about 100 moves. Now randomly assign a turn width (how many layers) to each turn. That's the same as the "normal" approach, except it might have a higher "chaos per turn" rate, similar to us using Cube Explorer for the 3x3x3 now, creating *better* scrambles with *fewer* moves.

Another idea, at least for the 6x6 and 7x7: Turn all layers except the leftmost one. Turn all layers except the leftmost two. Repeat, getting closer to the right side, until you only turn the rightmost layer. So turn along all twisting planes on the x axis. Then rotate the cube as a whole, repeat the whole procedure. One scramble line for the 7x7 could look like "R R' R R2 R2 R'" or equivalently "1 3 1 2 2 3" or still equivalently "U F U D U F". With the last notation I mean where your right thumb ends up after each turn.

But for the current scramblers as well as any alternative suggestions, what we really should have besides people seeing flaws by thinking is scramble quality judger programs to compare them statistically. Something like Daniel did a while ago, but with statistics for correlations and always at least two to independently verify their results.

 

[Mike Hughey]

Another idea, at least for the 6x6 and 7x7: Turn all layers except the leftmost one. Turn all layers except the leftmost two. Repeat, getting closer to the right side, until you only turn the rightmost layer. So turn along all twisting planes on the x axis. Then rotate the cube as a whole, repeat the whole procedure. One scramble line for the 7x7 could look like "R R' R R2 R2 R'" or equivalently "1 3 1 2 2 3" or still equivalently "U F U D U F". With the last notation I mean where your right thumb ends up after each turn.

If I'm understanding you correctly, I think Kenneth already did this at a tournament late last year. I'll have to search for it. This is kind of the way I hand-scramble big cubes anyway. (I just pick an axis and then madly turn the slices without paying attention to how, then rotate the cube and do it again.)

Edit: Found it: http://www.speedsolving.com/forum/showthread.php?p=77680&highlight=scrambling#post77680

I will also mention that I have done a lot of big cube scrambling. With my practice, I really doubt any of these methods would be significantly faster than the official version, for me. But I guess that's due to a) practice, and b) the fact that I'm so slow in general anyway.

 

[EmersonHerrmann]

if you mix it up with inner layers after outer layers...you can solve the inner like a 3x3 (2x2 on 4x4) and then all the edges on the outside will be solved...

 

[tim]

But, did you change the orientation at all before starting to solve? There's a 1/6 chance that between the scrambling table and where you solve that that would happen, i suspect, because you have an idea of what centers etc. go where.

1/6? Think about that number again.

 

[jcuber]

I think he means a 1/6 chance of getting the correct orientation.

 

[pcharles93]

It doesn't even matter. If you scramble one layer inward with each new sequence, you can start with a 2x2 or 3x3 solve and work outward solving a 3x3 with more layers each time. I really wish this guy would try it for himself...

 

[Ellis]

1/6 would be like saying you can only solve a 2x2 one out of six times because it needs to be in the correct orientation before applying any moves.

 

[shelley]

1/6? What does orientation have to do with it? If the last half of the scramble were done as if the 4x4 were a 2x2, you would be able to solve the centers as a 2x2. Seriously, try it for yourself. It won't take long.

If you scrambled inner layers first and then outer layers, it would make it more difficult to solve that way (solve as 3x3 into 2x2). However, this scramble method would still be insufficient as it wouldn't be able to reach all positions.

 

[qqwref]

I also has a point that most people would scramble faster, people who don't solve big cubes. We all seem to be avoiding the obvious point that most people who are scrambling big cubes are doing that instead of competing for a reason.

This is not obvious because it isn't true. In the competitions I go to, when things go according to plan, some poor fellow has to scramble my 5x5 or whatever a couple of times, and then I get to spend the rest of the round scrambling other people's. (I love scrambling bigcubes. Also I'm good at it.)

 

[Tetris Cube]

What if you alternated between inner and outer slice turns? Maybe not after every turn, but with two 3x3 scrambles, apply half of one to the outside, half on the inside, and same for the second one. It wouldn't keep the relativity of the centers and edges, I don't think. Pochmann's idea works too.

 

[qqwref]

mods, CLOSE THIS THREAD please

Unless things have changed drastically around here, mods don't just close threads whenever the OP thinks they're done with it. If people want to discuss let them discuss, if they don't people will stop bumping the thread and it will go out of the first page in a day or so.