help with corners first mod

[GuyWithFunnyHat]

I have recently been looking into corners first methods, the variation I was thinking of adopting myself goes like this

stage 1. corners, (guimond or ortega 2x2 method)
stage 2. 3 edges on each side (mostly intuitive)
stage 3. last 6 edges (I do this as the last step of the roux method described here http://grrroux.free.fr/method/Step_4.html)

stage 2 is the biggest time waster, and I'm looking for some way to cut the move count down. I am currently using the moves from these two pages: http://rubikscube.info/beginner-2.php and
http://rubikscube.info/waterman/stage1.php
for the edge insertions.

I am thinking that a logical way to cut down on moves would be to insert edges two at a time, instead of one at a time. for example, if the sides we are working with are blue and green, we would insert the blue-red edge and the green-red edge together, the blue-white and the green-white together, etc.this would be something like making the move from beginners method to fridrich F2L.


here's an example solve.

scramble: F U' D' L' F' U2 L U2 B2 U2 B F D' F2 D' U2 L' U F' R2 B R L' U' B (I scramble with blue on top and usually start with green, yay for not conforming!)

corners.
(f) (u') U (u) L2 D' R U' D2 R2 U' R2 (orient corners)
(u') R' U R' B2 R U' R (a rather odd alg)
M (align centers)

3 bottom edges:
(u') M' U2 M
(u) M' U2 M
(u) U' M' U' M

3 "top" edges
(u2) (f) R2 U' M U
R U M U'
r U M' U'
R

last 6 edges:
U M' U M' U M U (M M2) U M U2 M U M U2 M U2 M' I didn't notice the cancellation..., and I only learned that step yesterday, so I have no doubt that there is a better way to solve those edges.

hmm, i don't usually do that many cube rotations. oh well.

with the cancellation it comes out to 58 moves in STM. very similar to fridrich...
according to Roux, the last 6 edges can be done in 15 moves average, so mastering that step will cut down the move count a bit. Any other suggestions on how to optimize this method, especially the first 6 edges, would be very much appreciated.

 

[rubiksfriend]

I would recommend learning Waterman. It can be crazy fast if you can build the face quickly and memorize 100 sequences (about the same as Fridrich). Plus, nobody uses it!

 

[Swordsman Kirby]

I would recommend learning Waterman. It can be crazy fast if you can build the face quickly and memorize 100 sequences (about the same as Fridrich). Plus, nobody uses it!

I only recalled Fridrich having 57+21=78 algorithms.

 

[rubiksfriend]

What about awkward F2L cases? Plus, almost all of the sequences involve only right-hand and left-index turns. Very fast, but you must be able to separate the cases. (Similarity).

 

[Swordsman Kirby]

What about awkward F2L cases? Plus, almost all of the sequences involve only right-hand and left-index turns. Very fast, but you must be able to separate the cases. (Similarity).

Huh? I didn't learn any algorithms for F2L.

 

[deadalnix]

Many waterman's cases are commutotors and so easye to learn. Realy easier than oll/pll if you understand how use commutators.

 

[DonQuixote]

Any other suggestions on how to optimize this method, especially the first 6 edges, would be very much appreciated.

Well, there is a reason why you solve U corners third in Roux: you'll hinder yourself with what you've already solved if you solve corners first. So I don't think randomly mixing different methods is a good idea.
Do your "stage 1" sequences preserve any of the 12 edges? If yes, you could try to solve those first, otherwise I'd recommend something like Waterman, Roux oder Heise rather than blindly putting different speedcubing approaches together.
It's like F2L with Roux and then PLL+OLL: it's not a good idea.

Kind Regards
Martin

 

[GuyWithFunnyHat]

thanks for the replies guys. I was really trying to avoid learning waterman due to the large number of algs, and the limited availability of resources, as far as I know its just http://rubikscube.info/ that has any info at all on the subject (are there any others?). also a lot of the algs on that site aren't very finger friendly. I guess I could find my own with cube explorer or some such thing...I'll probably end up with waterman anyway, since it has already been optimized.

@ DonQuixotie:
I'm not really sure what you mean by

... blindly putting different speedcubing approaches together.
It's like F2L with Roux and then PLL+OLL: it's not a good idea.

it seems like a pretty generic corners first method to me except for the last bit where i do roux. I would argue that it is a bit more like doing petrus F2L then OLL+PLL.

I can definitely see your point about the top corners interfering with the solve though. I think that's why waterman decided to do those after the first layer. Solving a couple edges during the corner part also seems like a good idea, I'll look into that.

 

[rubiksfriend]

Not finger friendly? Half the method uses only your right hand! Plus, most of the algorithms for the first half of the third step are really easy (solve 2 redges, solve 1 in the ring 1 in R-layer). Plus, the site has all you need to know, but first try to understand what you're trying to accomplish rather than blind memorization. Roux is also a very good choice. Speaking of algorithms, Waterman says that he tried to avoid using the table and would occasionally solve 3 ledges, opposite corners and the remaining ledge with a redge to cut back on the memo.

EDIT: You're referring to the solving of the remaining corners. Silly me. Just use ones from other sites (Razoux Schultz has CLL, most of Roux's work).
Also, here are the steps of the method. It's actually very optimized if you think about it, because you can deal efficiently with about 90+% of the cases.
1) Solve one face on L.
2) Solve the opposite corners.
3) **Solve two redges and place one redge in the R-layer OR solve three redges**
4) Solve the remaining redges and orient midges simultaneously.
5) Place midges.
It's actually quite good. However, you can solve the last ledge with a redge like beginner methods, then use sequences for the rest. That way, you have to memorize less and can learn the remaining cases gradually. First, however, learn how to solve two redges in the middle ring and solve one redge in the ring and one in the R-layer. Then, you can be very fast.