Three are over 190,000 cross cases for each colour. Symmetries still won't reduce this to the point where humans could learn it. And as Shadowslice said: would we need this? It's not usually difficult to see a near-optimal cross in inspection.
uhh this is 3x3
What is?
Whoops looks like I mixed up the average PLL movecount (12) with TTLL (14). I stil think the method would be around 45 moves, as I doubt the TSLE step would be 10 moves on average since a lot of cases are <5 moves and the original TSLE is about 9-10 moves on average.
Can you though? If you just have random algorithms memorized and you don't understand why they work, that doesn't make it easier to look ahead to F2L. Intuition is always better for lookahead, because you can understand where the pieces are going and why they're going there. Think about how hard it is to predict PLL before you do OLL. Besides, if you're willing to put the time into learning the impossible amount of algs (190K+, according to TDM; equivalent to learning ZBLL over 385 times) that this method would require, your time would be much better spent working on lookahead.
Also, like others have mentioned, it's not that hard to get an optimal/near optimal intuitive cross with practice. I just pulled up a random scramble on csTimer: F2 R D L' U2 B' D F R F2 D2 L' B2 L' U2 L2 D2 L2 D'
I solved cross on green intuitively like this: x' y R' F2 D F D F2
csTimer's "Solve Cross" feature gave this: x' B' R2 D R D R2
Which is exactly the same thing; I just added a rotation to make it more fingertrick friendly. No algs necessary.
As someone that genned some preliminary algs for this exact substep: yes, it is 10 moves on average. TSLE is not 9-10 moves, it is 10 moves (and that is without AUF even)
12 * 11 * 10 * 9 EP * 2 * 2 * 2 * 2 EO = 190,080 cases, just as @TDM said.
Maybe I could make this like only one edge inserted or I do 999-look last layer
Haha RIP. I for one think it's a nice idea. Honestly the best judge would be to test it out.
