ZZ speedcubing method

[Swordsman Kirby]

Why not expand on variant d? It has a 2 gen LL, that's a lot easier than the ZZ LL.

I think the problem is that nobody has been able to find the missing link step – after the left block, before the right side – it's some sequence of moves that changes the cube from RUL-solvable to RU-solvable.

If I understand everything right, the phasing bit towards the end is sort of a replacement. It allows for the LL to be solved in one algorithm, although I'm not sure if it's 2-gen or not.

It CAN be 2-gen, but about half the time that's a waste of moves.

 

[yurivish]

I have a quick question. Is COLL part of ZZLL? I think it should be at least a subset of the cases, the ones with the cross pieces properly permuted.

I only ask because if it is, then I might start learning COLL now rather than waiting for the entire list of ZZLL algorithms.

 

[*LukeMayn*]

I have a quick question. Is COLL part of ZZLL? I think it should be at least a subset of the cases, the ones with the cross pieces properly permuted.

I only ask because if it is, then I might start learning COLL now rather than waiting for the entire list of ZZLL algorithms.

I believe it is a simpiler version of ZZ. and why wait for ZZLL algs when you can FIND THE MISSING LINK AND USE 2-GEN

 

[fanwuq]

I believe some things have been found for permuting corners for petrus before LL. It was on a thread a few months ago, (maybe march?) I didn't pay much attention to it, but I remember that it did link to a detailed site. I think the main concern was that recognition would be horrible.
I'd optimize my F2L and EO line before attempting such crazy LL ideas. Mean while, OLL + PLL works just fine.

 

[Lucas Garron]

I think the problem is that nobody has been able to find the missing link step – after the left block, before the right side – it's some sequence of moves that changes the cube from RUL-solvable to RU-solvable.

Arnaud and I have done it.
(Well, at least I know I have an okay-ish [but, most importantly, working] general solution.)

 

[yurivish]

Care to share? If it's workable for speedcubing, then that's pretty awesome.

it could be renamed the ZZLGAVG method!

Edit - Did you find it when specifically looking for this step of the ZZ method, or was it discovered some other way?

 

[fanwuq]

Another example of EOline
http://www.speedsolving.com/forum/showthread.php?t=3503

Still can't find the CP thread.

 

[*LukeMayn*]

The only real let down with this method is with big cubes like bigger than 3x3 cos you have to spend time after redux (if you use it) to look for the incorrectly oriented edges. I'm gonna learn COLL now for the basic zz method!
EDIT* Oh C'MON LUCAS share the love, you don't even use zz but you know the missing link -_- please share, I'M BEGGING OF YOU!!!

 

[Lucas Garron]

ZZLGAVG? Hah.
http://www.speedsolving.com/forum/showthread.php?p=59828#post59828

It's really not worth it, especially not for speedsolving, which is the main reason we care about 2-gen, anyhow. Unless you have a fantastic recog scheme, even the short fix algs are not that fantastic. But go develop this further, it's always worth it...

 

[*LukeMayn*]

I think something that could improve this method more though is a corners ZB method. So you do EOLine, block build one side, then do the other but this time only do the cross edge and ONE slot next to it. Now when you do the last slot, orient all of the corners then procede to PLL.
Otherwise you could just go with ZBLL. Like do the F2L (which is easier than fridrich cos of no cube movements and 3-gen style) then finish with ZBLL (which would benefit to Jason considering that he knows a lot of it.
Also Lucas Thanks for the link but can you do an ani-cube so I can see what you mean because when I try those algs on a solved cube it pretty much screws it up. Thanks.
*shakes hands after tying hardout *
EDIT***zz (with OLL/PLL) avg. of 12 50.12
-_- so close to sub 50, also just to make th post even longer here are some time schemes for differant variations on zz

ZZ with OLL/PLL

(all times in seconds)
***pro's and con's for LL***
EOLine = 2.5
Left block = 2.5
right block = 2
OLL = 1.5
PLL = 2
+around 1.5 secs of recog =
12

pro's = Very little amount of algs only 28 (20 excluding mirrors and inverse), very possible to get sub 15 avg, Very fast algs.
con's = 2LLL= more recog

ZZ with COLL/EPLL
EOLine = 2.5
left block = 2.5
right block = 2
COLL = 2
EPLL = 1.25
+ around 1.5 secs recog =
11.75

pro's = EPLL VERY fast, can get great averages. Less recog than OLL/PLL for last step.
con's = 2LLL = more recog

ZZ with ZBLL
EOLine = 2.5
left block = 2.5
right block = 2
ZBLL = 2.25
+ around 1 sec recog =
10.25

Pro's = 1LLL = less recog time, can get 10sec avgs with enough practise.
con's = A LOT of algs and easy to forget them if you don't practise, some ZBLL slow, CRAZY hard to do a ZBLL time attack.

ZZ with ZZLL
EOLing = 2.5
left block = 2.5
right block + phasing = 2.25
ZZLL = 2.25
+ around a little less than 1 sec recog =
10.50 or a little less

pro's = Phasing can be done easily on the fly ,far less algs tham ZBLL, can get 10 sec avgs with enough practise.
con's = Phasing takes a little more thinking, only 160 algs (which is actually like 80 after counting mirrors and inverses)

This is measured as if they were the Harris chan of F2L and had good LL with very little stops, recog could be quicker than this though and EOLine could be better or worse down by about 1.25 secs on a lucky solve(0 or 2 edge unoriented) or up by about 0.75 on an unluck solve (6 or 8 edges unoriented)
IMO I will go with ZZ with ZZLL because of only 160 algs (which is actually like 80 after counting mirrors and inverses) and because it has a lot of potential.

 

[Lofty]

Ok guys... chill out and read around before you start spouting all this stuff out.
Phasing is to permute two of the LL edges thus reducing the number of cases of ZBLL down to the smaller ZZLL.
The missing link is not really missing and has been throughly discussed especially in relation to the Petrus method. The conclusion is that orienting corners at that stage of the solve has terrible recognition and doesn't really work well for speedsolving.
As for orienting the corners yes you could just do a very simplified version of MGLS (not very simplified you just get to cut off ELS). You could also do the Winter Variation (aka F2LL) but that requires that you have a connected pair created to orient your corners while inserting the last pair.
I think the method is best as is. It only takes a few moves to "phase" and reduce ZB to ZZ. You could also just ignore phasing and learn VHLL and then ZZLL and then 2GLL and then you have a little chunk of ZBLL already memorized.
Edit: I am seriously considering switching to this method after Nationals this weekend. It seem to have a nice blend of everything. And being similar to Fridrich we can use things we already though be slightly more efficient as the whole cross is not in the way. I think in the future how much that is able to be accomplished in the inspection time will be important, not that it isnt already important now. Both Fridich and Petrus solve only 4 pieces in the first step while Roux solves 5 so roux wins there, this solves only two pieces but oreints all your edges which i think is a very good step to get out of the way right at the start of the cube. Then the simplified blockbuilding LUR F2L with a 1LLL its my dream, I don't know why I stopped considering it before.

 

[*LukeMayn*]

Lofty:
You can't use 2GLL without permuting corners (which I have heard is really annoying in a solve.) Also Thanks for clearing that ZZLL and ZBLL differance for me, I didn't really get it. Thanks again Lofty!

 

[Lofty]

Lofty:
You can't use 2GLL without permuting corners (which I have heard is really annoying in a solve.) Also Thanks for clearing that ZZLL and ZBLL differance for me, I didn't really get it. Thanks again Lofty!

I know you cant use 2GLL without permuted corners... but if you already know COLL you can solve the what 1 in 3 i think times your corners are permuted in 1 look, assuming you are referencing my saying learn different LL alg sets. So I'm not saying you would use it every time just increase your probability of a 1LLL from 1 in 12 to something greater than 1 in 12 lol.

 

[*LukeMayn*]

Lofty:
You can't use 2GLL without permuting corners (which I have heard is really annoying in a solve.) Also Thanks for clearing that ZZLL and ZBLL differance for me, I didn't really get it. Thanks again Lofty!

I know you cant use 2GLL without permuted corners... but if you already know COLL you can solve the what 1 in 3 i think times your corners are permuted in 1 look, assuming you are referencing my saying learn different LL alg sets. So I'm not saying you would use it every time just increase your probability of a 1LLL from 1 in 12 to something greater than 1 in 12 lol.

I'll just learn ZZLL

 

[fanwuq]

Doing EO line in 6 moves is amazingly hard on average. I usually can't see the whole thing during pre-inspection.

LUR F2L for some reason feels really awkward. I end up rotation without reason just out of habit. So cases won't work with my regular F2L algs.
Would Multislotting be a lot easier to recognize with ZZF2L than Fridrich F2L?
I'm considering MGLS and PLL for the finishing.

Expected move counts during a speedsolve:
EOline: 9
L Block: 8
one more 2x2x1 on R: 5
MGLS: 13
PLL: 13
Total: 48
Not bad if I can improve my Block building on L and the EOline.
The recognition would be quite easy for this method.
It's pretty much a 5 look method. (Of course, EOline is 2 looks for me now, first block might also be 2 looks, MGLS can be 2 looks.)

 

[MHordecki]

I've did a statistic concerning phasing. Out of 20 solves, overhead generated by phasing step is 1.95 on average. It was suprising for me that in 18 of 20 solves, overhead was either 4 or 0 moves(0 meaning 'skip'). I like it, cause phasing do not require any algorithms (or maybe I'm not doing it optimally) - it's usually some 2-gen moves before last move of a slot-inserting algorithm.

 

[Lotsofsloths]

Can someone make a youtube video..like a walkthrough solve?

 

[fanwuq]

How do you do phasing exactly?
Does it get rid of nicer LL cases such as the Sune?

 

[MHordecki]

Yup, some of the nicest LL do not apply to ZZLL, but some of them are retained ( such as (F R F') r (U R') U' r' - that's my favourite COLL, by the way ).

As for phasing, my current method (I'm still a beginner, though) involves learning no algs (so, probably, it's not so optimal, but it's still quick and ~2 move avg).

Given the example:

In classic F2L, we would do R U R'. Then, we would end with:

(it's actually COLL case I hate the most ;P )

This time, we want do do phasing. So, we do R U2 R':

Now, phasing is done (with UF edge in UR and UB edge in UL), we need to solve c+e pair - we can do it using double move: U R U2 R':

Done! It's 7 moves instead of 3, so +4 moves this time.

 

[GuyWithFunnyHat]

here's the corner permutation thread, http://www.speedsolving.com/forum/showthread.php?t=3818

as was said before, the main concern was that recognition would be terrible.

another concern is that all edges are already correctly oriented after the 2x2x3 block is made, and the alg to switch two corners is 7 moves, the same as niklas, so it would be much better to do a 2LLL with niklas first, if you really really really want to use 2GLL.