WW: New speedcubing method; Zb variant

[cube980]

I was thinking about learning Zb in the future. I was intimidated by the amount of algorithms. I thought about reducing the alg amount and thought of using corners instead of edges. I found the winter varient (f2ll) which oreints the corners while inserting the last f2l pair and developed WW.

WW explanation on the wiki
"WW(Whittaker-Winter named after me the inventor and Lucas Winter) is an experimental 3x3 method that is a variation of the zb method. that orients the corners when inserting the last f2l pair (Winter variation). Then solves edge orientation and pll in one step. Orienting the corners first rather than orienting edges(like in zb) has the advantage of using far less algorithms(using only 90 total compared to zb's 300). It is unknown if and very doudtful if it has ever been used before.

To solve using WW first solve cross and 3 f2l slots like in fridrich(or any way one wishes). Next use the Winter variation to insert the last f2l pair and orient last layer corners. Finnaly solve eoll and pll in one algorithm.

Pros Uses a fairly decent amount of algorithms(21 of which are pll algs). As fast(or faster; better reconization) as zb using much less algorithms. Fair reconization.

Cons It is unknown if the downfall of this method is having bad algorithms,or bad look ahead. Never been used. "

I will soon be making a list of the algorithms.
So what do you think(is it good for speedcubing?,etc.)

 

[hic0057]

It could work or be a total flop. If it does turn out to be a good method I might try it.

 

[Tim Major]

http://www.speedsolving.com/forum/showthread.php?p=276502#post276502
Very similar, except yours is more of a VH-F2L approach to the corners, whereas mine was a ZB-F2L.
I can't imagine this being too efficient. For one, you are effectively doing the LL in 2 steps anyway. The WV algs, then the finish. This would be like doing VH-F2L, or an edge flipping OLL such as F R U R' U' F', then finishing.
VH-F2L can only really be used as a stepping stone to ZB-F2L.
Most of the stuff in the thread I linked to is relevant to your method, so check that out.

 

[cube980]

The main reason of this method is reducing the algorithms from zb as much as posible while still being very effective. I now relize this is alot like crorner Vh-f2l. Now I have dont realy know if it will be very effective but im looking into it. I hope to have algoriths up soon(within a week).

 

[Weston]

Recognizing and executing the winter variation alg after you make the F2L pair adds an extra step so its still kind of the same amout of steps as Fridrich.

 

[qqwref]

Mm, but because of the number of cases of corner orientation, it isn't really possible to do F2L and CO in one step. So you are doing last slot + last layer in 3 looks, just like F2L/OLL/PLL. One of the big advantages of ZBF2L/ZBLL is that it is only 2 steps. Maybe you'll save moves compared to the normal approach, so it might be faster, but you won't save time for any other reason.


Computation of # of algs follows. F2L cases can be considered like this:
- 17(*2) asymmetrical F2L cases with at least one piece in U
- 2 symmetrical F2L cases with at least one piece in U
- 2(*2) asymmetrical F2L cases with no pieces in U
- 2 symmetrical F2L cases with no pieces in U (counting solved).

For ZBF2L, asymmetrical cases with a piece in U have 8 possible EOs; symmetrical cases with a piece in U have 8 EOs (without mirrors, 6 if the edge is solved and 4 if it is flipped); all cases with no pieces in U have 4 EOs if the edge is solved and 2 if it is flipped. I count 306 cases if you include mirrors, and 158 if you don't. One of each of these is the solved case.

OK, now for the full F2L + CO variant you suggest. Asymmetrical cases without a piece in U have 27 possible COs; symmetrical cases with a piece in U have 27 COs (without mirrors, 15); asymmetrical cases with no pieces in U have 8 COs; symmetrical cases with no pieces in U have 8 COs (without mirrors, 7). I count 1020 cases if you include mirrors and 519 if you don't. Again, one of each of these is the solved case.

So yeah, you definitely can't do it in one look. It's 3 times as many cases as ZBF2L, and that's already a pretty tough alg set to learn.

 

[Erik]

Already tried this in 2007:

Disadvantages:
- long algorithms for last pair + corner orientation
- LOTS of algorithms for last pair + corner orienation
- getting used to an ELL sorta recognition (possibly faster than ZB)
- attempting to reduce the amount of F2L+CO cases is a total fail, first pairing up and then working towards a WV case is slow and sometimes takes even more moves than normal OLL/PLL

Advantages:
- can be very short in some cases (though on average the algs suck)
- cool

 

[Kirjava]

Yeah, you should at least make the last layer 1look. The algs for what you propose already exist.

I do this when I get a CO skip for LL ^_^ (Well, it's like.. an OLLCP/KCLL subset).

 

[cube980]

wow, this is probaply just a flop I gress I forgot to account for the extra look in wv.

 

[joey]

Same as http://www.speedsolving.com/forum/showthread.php?t=19320

 

[RyanReese09]

I thought about this a month ago after nationals, in the end, the above cons posted made me not stick with it (they are really good cons )

 

[keemy]

btw if you didn't know WV sucks for most cases when I thought of this some alternatives were a psudo-winter where you ignored EO (and learning 2 for each case to avoid the 4 edge flip algs as they all suck) and something like insert last edge + CLS, both are better than pure WV probably but w/e too lazy to learn method.

 

[bluedasher]

I imagine it would be very hard to recognize the case you have since their are three possible orientations of the corners and four possible corner permutations. Not to mention the fact that LL pieces may very well be in the last slot. By the time you recognize the slow, crappy algorithm you might as well have done OLL and PLL.

 

[hic0057]

Another idea you could try is a step cross between vh and Oll. So when you insert the last f2l you orientate the last layer then you just permute it.

 

[Erik]

Inserting last pair plus orienting corners only = 29 cases, 29 more cases when you include the mirror (L'UL)
Full OLL:
Now for 0 edges oriented: 1 possibility
Now for all edges oriented: 1 possibility
For 2 edges oriented:
A&B, A&C, A&D, B&C, B&D, D&C
In total: 8 * 29 = 232 cases 464 cases including mirrors.

Then again, the gain is almost 0 because making the pair and making it ready for insertion is mostly not even shorter than solving the pair.............

 

[hic0057]

Inserting last pair plus orienting corners only = 29 cases, 29 more cases when you include the mirror (L'UL)
Full OLL:
Now for 0 edges oriented: 1 possibility
Now for all edges oriented: 1 possibility
For 2 edges oriented:
A&B, A&C, A&D, B&C, B&D, D&C
In total: 8 * 29 = 232 cases 464 cases including mirrors.

Then again, the gain is almost 0 because making the pair and making it ready for insertion is mostly not even shorter than solving the pair.............

I don't really understand what you are saying but the method could be good for a couple of special cases when you insert the last pair using RU'R' or L'UL so and all the edges are orientated so you can use this step to skip Oll.

I could try this step after I learn't OLL.

 

[oll+phase+sync]

What about "Winter Positioning" only position not orient the LL Corners while slotting in the last ce pair. There are just 6 cases (3,6,7,7,8,9 moves) if I recall correctly. After this one could use/optimize petrus Step6+Step7 index which has 49 algs http://lar5.com/cube/xMain.html

Obviously you need oriented edges, for this.

 

[Kirjava]

hurr durr CPLS

 

[StachuK1992]

hurr durr CPLS

Some lolRoux variant I came up with on the toilet:
First block.
Second block-DFR
CPLS (CPLL and finish second block)
SuneOLL to do OCLL and L6E0
then just continue from Roux step 4b.

Pretty crappy method, eh?