Welcome to the Speedsolving.com, home of the web's largest puzzle community! You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

What I don't like with the basic idea is that 2/3 of the time, the 4c are no better than a U-perm (in movecount). But actually, there's a way to force nice cases. (Let's call it Phased COLL for now)

Here's the idea: you may find that the best 4c cases occur when edges that should be opposite, are opposite:
good: DF/DB on UR/UL or UF/UB.
bad: DF/DB on UR/UF or UR/UB.

So why not force opposite edges to be opposite while solving the COLL? (Or on the contrary, force opposite edges to be adjacent?)
Granted, you need many more algs. There are 3 cases per COLL, so 100 to 120 algs, 40 of which are COLL. But ZBLL is 12 cases per COLL, so you need 4 times less algs, and still get all the advantages of ZBLL.

Why? Let's have a look at the 4c you may get. AUF for 0.75 moves on average, insert the UL/UR with M2 (1 move), then you get (each option is equally likely):
- DF in DF, that's a skip (with AUF, total 2.75 move after ZBLL - I am including EVERY move after the ZBLL, including the insertion of UL/UR and AUFs)
- DF in UB (AUF, M' U2 M2 U2 M' U2, total 8.75 moves). Algs like E2 M2 U M' E2 M' or M2 u2 U' M' E2 M' solve it in 6.75 moves.
- DF in UF (AUF, M' U2 M2 U2 M', total 7.75 moves)
- DF in DB (AUF, M2 U2 M2, total 5.75 moves)

Furthermore, the recognition in this step is easy: you just have to look at whether the FU (or DU) sticker matches the F center. And there is little ambiguity in which alg to apply (up to symmetry).

If you decide to phase so that opposite edges are NOT adjacent, you only need two algs per COLL (33% less!), and you may get:
- A 3-cycle with both D edges (7.75 moves)
- A 3-cycle with both U edges (6.75 moves)
- no skip
It gives an average of 7.25 moves, 1.5 more move, with less skips but with a less nasty case and a lower alg count. The recognition is just as good.

Let's sum up.
- On average, you need 5.75 moves, very possibly less after the phased COLL, compared to 0.75 moves for the AUF in ZBLL. Remember that phased COLL is a subset of ZBLL, so you actually get just 5 more moves than ZBLL. Since you may chose shorter ZBLL, and gain a few moves at the start, the extra movecount might be even lower.
- The recognition of phased COLL is faster and easier than ZBLL (it's COLL recognition, plus looking at where the D stickers are on the U face!), so you spend less time there.
- The 4c cases, being the easiest, are even faster than in standard NPL.

Pros of phasing COLL:
+ 4 (or 6) times less algs than ZBLL for less than 5 (or 6.5) extra moves on average (but recognition might be faster, any expert here?), Compared to approx +8 moves for COLL+EPLL.
+ stepping stone to learning ZBLL
+ recognition is incredible
+ The ending is braindead.
+ Skip 25% to 50% of the time (depending on whether you count the DF/DB case as a skip)

Cons of phasing COLL:
- still around 3 times more algs than COLL.

Looks like an interesting method, although I don't quite understand how you do the last step with mismatching centres.
Also, can someone point to a tutorial on orienting edges during f2l? I am not certain of how to consistently do this.

Looks like an interesting method, although I don't quite understand how you do the last step with mismatching centres.
Also, can someone point to a tutorial on orienting edges during f2l? I am not certain of how to consistently do this.

ZZ is by far the most reliable way to orientate edges, and works better with this method than CFOP imo. It has already been posted on this thread how to do it. It's slightly better than ZZ with COLL/EPLL, but not as good as with ZBLL, so you can decide.

So, there shouldn't be many algorithms to learn with this method.

Firstly, you can and probably should use 2-look corners (aka 2-look COLL/CMLL/CxLL). This means you only need to learn 7 OCLL cases (~4 algorithms tops) and 2 CPLL algorithms.

For EOLL, I would recommend learning 3 corner-preserving algorithms. This would allow you to eventually do 1-look compound (C)OLL with this method. (In other words, you could plan your EOLL alg and COLL alg at the same time and execute both algs one after another to arrive at step 4c with only 1 look.)

The other option here would be to orient edges at the very beginning of your solve, treating this method as a ZZ variant moreso than a CFOP variant. This carries a steeper learning curve though.

To get better at 4c (and 3-cycle cases), look up a guide on or watch a video about BU recognition. There are dozens of them. Find one that makes sense to you.

Ultimately, 4c only requires you to know 2 algorithms in my opinion: U2 M2 U2 and E2 M E2 M'.

I guess I'm saying: Don't overthink it. You could still get amazingly fast with this method using only intuitive F2L and ~10 algorithms.

This is a very underrated method in my opinion. Simply because 4c averages 2 moves less than EPLL.