He actually had a revolutionary new method!
Ever saw how he did a LOT of seemingly useless moves before starting the the cross/F2L phase in CFOP? Either that, or he had instant x-crosses [double even]! None of us understood it, and we came to the conclusion that he:
- did a solve backwards
(random easy PLL, random easy OLL, or skips if he wanted, break some F2L pairs, break cross, insert random moves)
- obtained a scrambled cube state by doing that,
- and generated the scramble of that state by using Cube Explorer, and reverse it
(which explains the domino effect at the start of his scrambles, similar to the last few steps of CE solves)
- and memorize the inverse of the backwards solve, and possibly rehearse it
However, this means that he supposedly memorized, for his "average of 12" video, 12 inverted solves, with totally random moves at the start, and was able to perform all of this at full speed, in front of a camera, without making a mistake or forgetting any of it.
That alone already sounds questionable.
Anyone knows the pressure when cubing with a timer is different than without. What more if you know you're being taped. AND if it was supposedly a fake solve. And If you memorized TWELVE fake solves. And coupled with the fact that you supposedly suck, I doubt you would be able to pull it off.
And look at his wristing technique. While he is a wrister, his wristing speed cannot be attained by an amateur.
Now, look at this way instead.
If you, a CFOP user, and you don't do BLD or Petrus/Roux/ZZ, or anything with EO in particular, saw a ZZ solve, wouldn't the EOLine step to you seem totally bogus moves at the start as well? Because you don't know the concept of EO, thus you can't see what is happening.
What if Yish knew a new concept, like EO, that we couldn't comprehend, and thus his moves seemed like nonsense to us?
Well, I discovered that there exists a concept that Yish knew about!
I found a link in the sgcubers forum, that they were trying to keep under wraps, which might explain why Yish didn't do a low move count explained solve, despite looking fake in the process.
It's called "edge-corner pairing correlation" (ECPC). And unlike EO which only deals with orientation, it fixes BOTH orientation, and potential corner/edge permutation, such that after fixing ECPC, F2L pairs seem to just "magically form" one after the other, ala Yish solves.
Is this even possible?
If you'll look at the solves generated by the program CubeTeacher here, it almost always gets an OLL or PLL skip. How?? By choosing the proper F2L ordering such that it gets one.
The program itself has proven it possible, but is it even humanly feasible to be able to predict which one to use?
Well, EO has the magical effect of leaving the F2L in an RUL solvable state right?
ECPC works the same way, except it includes potential permutation, and is thus more powerful.
Look at Yish's instant xcross and double xcrosses. An average CFOP user struggles to form an xcross everytime, yet he does it with ease.
You know how when a CFOP user looks at the Petrus tutorial, they watch step 4b and say "HEY! F2L cases!" and can instantly comprehend it, when Petrus learners struggle with it?
Like how we struggle with the xcross and he doesn't. If you are aware of the cube's current ECPC state, you can easily see the optimal xcrosses everywhere, as easy as seeing the F2L case in Petrus step 4b.
In fact, they've managed to reduce most xcross block-building cases to mere algs due to this. Guide here.
If you are insterested in learning Yish's "edge-corner pairing correlation" concept, you can read up on it at that link, but be warned, knowledge in calculus is a must.
Ever saw how he did a LOT of seemingly useless moves before starting the the cross/F2L phase in CFOP? Either that, or he had instant x-crosses [double even]! None of us understood it, and we came to the conclusion that he:
- did a solve backwards
(random easy PLL, random easy OLL, or skips if he wanted, break some F2L pairs, break cross, insert random moves)
- obtained a scrambled cube state by doing that,
- and generated the scramble of that state by using Cube Explorer, and reverse it
(which explains the domino effect at the start of his scrambles, similar to the last few steps of CE solves)
- and memorize the inverse of the backwards solve, and possibly rehearse it
However, this means that he supposedly memorized, for his "average of 12" video, 12 inverted solves, with totally random moves at the start, and was able to perform all of this at full speed, in front of a camera, without making a mistake or forgetting any of it.
That alone already sounds questionable.
Anyone knows the pressure when cubing with a timer is different than without. What more if you know you're being taped. AND if it was supposedly a fake solve. And If you memorized TWELVE fake solves. And coupled with the fact that you supposedly suck, I doubt you would be able to pull it off.
And look at his wristing technique. While he is a wrister, his wristing speed cannot be attained by an amateur.
Now, look at this way instead.
If you, a CFOP user, and you don't do BLD or Petrus/Roux/ZZ, or anything with EO in particular, saw a ZZ solve, wouldn't the EOLine step to you seem totally bogus moves at the start as well? Because you don't know the concept of EO, thus you can't see what is happening.
What if Yish knew a new concept, like EO, that we couldn't comprehend, and thus his moves seemed like nonsense to us?
Well, I discovered that there exists a concept that Yish knew about!
I found a link in the sgcubers forum, that they were trying to keep under wraps, which might explain why Yish didn't do a low move count explained solve, despite looking fake in the process.
It's called "edge-corner pairing correlation" (ECPC). And unlike EO which only deals with orientation, it fixes BOTH orientation, and potential corner/edge permutation, such that after fixing ECPC, F2L pairs seem to just "magically form" one after the other, ala Yish solves.
Is this even possible?
If you'll look at the solves generated by the program CubeTeacher here, it almost always gets an OLL or PLL skip. How?? By choosing the proper F2L ordering such that it gets one.
The program itself has proven it possible, but is it even humanly feasible to be able to predict which one to use?
Well, EO has the magical effect of leaving the F2L in an RUL solvable state right?
ECPC works the same way, except it includes potential permutation, and is thus more powerful.
Look at Yish's instant xcross and double xcrosses. An average CFOP user struggles to form an xcross everytime, yet he does it with ease.
You know how when a CFOP user looks at the Petrus tutorial, they watch step 4b and say "HEY! F2L cases!" and can instantly comprehend it, when Petrus learners struggle with it?
Like how we struggle with the xcross and he doesn't. If you are aware of the cube's current ECPC state, you can easily see the optimal xcrosses everywhere, as easy as seeing the F2L case in Petrus step 4b.
In fact, they've managed to reduce most xcross block-building cases to mere algs due to this. Guide here.
If you are insterested in learning Yish's "edge-corner pairing correlation" concept, you can read up on it at that link, but be warned, knowledge in calculus is a must.
