[Help Thread] ZZ and ZB Discussion

[CubingGenius]

Its not necessary to count all of the edges from all sides. When I use y neutrality, I first count edges on my preferred orientation, which is yellow top blue front. Then to work out the affect of a y rotation, I mentally flip each edge on the E slice with a yellow or white sticker, and each edge on the U or D slice without a yellow or white sticker. That way I need to count a maximum of 8 edges on the second face, but it's often less. I also found it easier to do that than to get used to recognising EO from a different orientation.

It's also possible to do something similar with x rotations. An x rotation only flips edges in the M slice with L/R stickers, and edges in the L/R slices without L/R stickers.

I actually do count them from the same angle. I look at an edge and work whether it is oriented or not on every one.

Like this:

 

[CubingGenius]

I have more information:

I did 6 solves and timed the EO axis recognition and the line recognition.

1. 15.42, 1:18.73, 1:34.16
2. 17.62, 52.42, 1:10.45
3. 11.16, 34.78, 45.94
4. 36.75, 40.69, 1:17.45
5. 35.20, 23.13, 58.34
6. 13.18, 53.61, 1:06.79

I think the first step is getting better, apart from practicing solves, not much I can do there. But I feel like the second step is good with 0-4 misoriented edges, but weaker with 6+ misoriented edges. I think I should train with 6+ misoriented edge scrambles. Do you agree?

 

[pinser]

I just had to get this on the first solve of the session.

11.26 D R2 U' L2 U' R2 B2 R2 B2 U2 F2 L U' L2 B2 F' L2 F2 L2 R'

F2 L2 F D’ L2 D’
U R2 U2 R’ L2 U2 L2 U’ R’ U L
U R U2 R U R2 UR’ U2 R U R’ U
28 HTM

 

[CubingGenius]

I just had to get this on the first solve of the session.

11.26 D R2 U' L2 U' R2 B2 R2 B2 U2 F2 L U' L2 B2 F' L2 F2 L2 R'

F2 L2 F D’ L2 D’
U R2 U2 R’ L2 U2 L2 U’ R’ U L
U R U2 R U R2 UR’ U2 R U R’ U
28 HTM

I also got a LL skip on the first solve of my session yesterday as well that also came to 28 HTM.

13.16

L2 B2 F2 R2 U L2 U L2 U' L2 F' L B2 U R F2 U R2 D' F'

y' R d L R' F L R D'
R2 U2 R U R
U' L' U L U' L' U2 L' U L
U L U L'
U2

Lol.

 

[mDiPalma]

Because some people erroneously believe that phasing takes more moves than TSLE, and the forum search function is too useless to find any information to prove them wrong...

On a flight without a physical cube, I mentally went through all the possible last-slot F2L cases (<RU>) and came up with these statistics for creating+inserting a pair and phasing the LL edges:

To reduce the F2L state to either a pair in the U-layer (any AUF) or an R-U-R' type insert (just that single AUF), 4.667 moves are required.

(As a side note, to reduce the F2L state to a pair in the U-layer (any AUF), 5.573 moves are required.)

Of these, 20/75 are the R-U-R' type, which takes 5.667 moves to insert while phasing, using basic intuition of the 3 unique cases.

52/75 are pair-types that have edges at UR or UB. These all require a 1 move AUF before insertion while phasing.

1/75 has the edge at UL which requires .667 moves to AUF.

2/75 have the edge at UF which requires .333 moves to AUF (one of these is literally taking a LS-skipped pair out of the slot).

Of the pair types, 4.333 moves are required to insert the pair while phasing, using basic intuition of the 3 unique cases.

Combining those statistics and frequencies, we see that 10.067 moves <RU> are required to solve *any* LS case while phasing the LL edges.

And if you aren't dumb about how you deal with a LS-skip, I believe this number drops to 10.004 htm <RU>.

*although i believe i'm pretty (hopefully, by now, perfectly) efficient with <RU>, I may have missed something shorter, so these are upper bounds.

---

Now, according to Phillip Espinoza, TSLE requires 10.37 htm (I think this is <RU>; if not, the <RU> movecount would exceed 10.37). Add .75 moves for AUF and you have 11.12 htm <RU> for TSLE from any initial F2L state.

---

Therefore, creating and inserting the final F2L pair while phasing the LL edges is more than 1 move <RU> shorter than TSLE.

 

[AlphaSheep]

Now, according to Phillip Espinoza, TSLE requires 10.37 htm (I think this is <RU>; if not, the <RU> movecount would exceed 10.37). Add .75 moves for AUF and you have 11.12 htm <RU> for TSLE from any initial F2L state.

I believe Philip's estimation already includes AUF.

I generated my own algs with an optimal number of (R U* R' U*) triggers and got an average move count of 10.65 including AUF. You can check my math if you like: (0.75 + 3*3.75 + 24*7.75 + 73*11.75 + 4*15.75)/105. If you generate pure <RU> you can get a bit shorter for some cases (eg the four 15 move cases are all 11 moves optimal in <RU>), but for the vast majority, the trigger combos are already optimal in <RU>, so the average move count won't be much lower.

Your point still stands though. Phasing in 2-gen is still a bit shorter than 2-gen TSLE.

 

[pinser]

Is ZZLL merely a stepping stone to full ZBLL or is it fast enough to use by itself?

 

[CubingGenius]

Is ZZLL merely a stepping stone to full ZBLL or is it fast enough to use by itself?

Simon Kalhofer used it to get the first official sub 10 average with ZZ in the world I believe.

So it would be fast enough to use on its own. It's definitely superior to OCLL/PLL and COLL/EPLL.

 

[GenTheSnail]

Does anyone know what an official 13.79 3x3 2H average would rank me among ZZ solvers?

 

[AlphaSheep]

Does anyone know what an official 13.79 3x3 2H average would rank me among ZZ solvers?

No idea, and also no way of knowing. But I know there are several ZZ solvers with 10.xx averages, so definitely not that high up.

 

[CubingGenius]

No idea, and also no way of knowing. But I know there are several ZZ solvers with 10.xx averages, so definitely not that high up.

How many with sub 9 averages?

 

[genericcuber666]

so if zzll has less moves for the first step, and I already know how to recognize the second step well, because of coll, could someone remind me why i'm supposed to be learning zzct

 

[mDiPalma]

so if zzll has less moves for the first step, and I already know how to recognize the second step well, because of coll, could someone remind me why i'm supposed to be learning zzct

zzll (second step) is also less moves than ttll

now you see why many people question the adequacy of zzct

 

[genericcuber666]

I think that i'm going to learn zzll, phasing is easier, I already know a lot of the algs and the recognition is so much easier. Also currently the algs are better.


I'll learn ct when I average sub 6 and I really want to get lucky and get a wr...

 

[CubingGenius]

I think that i'm going to learn zzll, phasing is easier, I already know a lot of the algs and the recognition is so much easier. Also currently the algs are better.


I'll learn ct when I average sub 6 and I really want to get lucky and get a wr...

Why would you change methods when you average 6 seconds? Why don't you just use ZZ-CT now?

 

[genericcuber666]

Why would you change methods when you average 6 seconds? Why don't you just use ZZ-CT now?

I don't like CT, I've used a fairly good 2 look system and it feels restrictive.

The only way I can see myself wanting to learn CT is if I'm so fast that I just need to be lucky to get a wr/ER

 

[CubingGenius]

I don't like CT, I've used a fairly good 2 look system and it feels restrictive.

The only way I can see myself wanting to learn CT is if I'm so fast that I just need to be lucky to get a wr/ER

But at that point, wouldn't it take an extremely large amount of practice to get back to the times you had beforehand?

Anyway, ZZ-CT's chances of skips aren't very good. I suppose learning OLS would increase the chance to 1/72 LL skip, but even then the chances aren't that good. I think the best way to get a good single would be learning how to force easier F2L cases and avoid bad cases.

 

[genericcuber666]

i've been looking at zzll algs and they don't seem to bad.
Should I use baum-harris recognition?
is it ever worth learning advanced phasing?
could I realistically use zzll for oh?

But at that point, wouldn't it take an extremely large amount of practice to get back to the times you had beforehand?

Anyway, ZZ-CT's chances of skips aren't very good. I suppose learning OLS would increase the chance to 1/72 LL skip, but even then the chances aren't that good. I think the best way to get a good single would be learning how to force easier F2L cases and avoid bad cases.

my point was the only benefit I see with ct is skips, I don't care about skips I want a good average, also if I learn it now I will eventually have to relearn all the algs because they're not very optimised yet

 

[CubingGenius]

i've been looking at zzll algs and they don't seem to bad.
Should I use baum-harris recognition?
is it ever worth learning advanced phasing?
could I realistically use zzll for oh?



my point was the only benefit I see with ct is skips, I don't care about skips I want a good average, also if I learn it now I will eventually have to relearn all the algs because they're not very optimised yet

Isn't the change of TSLE skip 1/108 and TTLL 1/360? Not very good chances at all...

Advanced phasing would be a good idea.

Baum-Harris sounds good. All you need to do is compare one edge's sticker to a corner.

Simon Kalhofer uses ZZLL for OH and has a 13.70 OH average.

 

[TheBrutux168]

Critique please? I feel like my F2L is inefficient but I don't really know how it compares to good ZZ solves. And also at times I feel like the efficient way of doing things ends up being unergonomic. Do ZZ solvers always go for efficiency in F2L or is there a balance? Or should I just practice the things I feel are unergonomic?