Is ZZ an objectively worse method than Roux and CFOP?

[CarterK]

Sorry, I mean that it would be more pauses, meaning decreased tps, not that it would be slower, and plus the fact that there are 500 algs. I asked Kian what his move counts were and he said 47 STM for 2H and 55 ETM for OH(U2 is 2 moves).

 

[PapaSmurf]

ZBLL gives one less pause. If you do that, you should be doing no pause F2L (if you plan well in inspection and look ahead), and only pause for zbll and auf, which isn’t that hard.

Roux you don’t pause until cmll, then you pause again for eolr. Yes, ideally these aren’t bad pauses, but if you’ve mastered zbll, the total pause time should be exactly the same as pause time for cmll and eolr adder togeter.

 

[CarterK]

ZBLL gives one less pause. If you do that, you should be doing no pause F2L (if you plan well in inspection and look ahead), and only pause for zbll and auf, which isn’t that hard.

Roux you don’t pause until cmll, then you pause again for eolr. Yes, ideally these aren’t bad pauses, but if you’ve mastered zbll, the total pause time should be exactly the same as pause time for cmll and eolr adder togeter.

If you are doing it right, there should be no pause between cmll and eolr, and maybe a small pause for cmll but in most cases you can predict it as your are inserting your last pair, so the pause for ZBLL will be a lot longer than the Roux pauses.

 

[PapaSmurf]

If you know full 2sided zbll recog, you can predict coll in last pair, then just look at 2 edges, then the alg. Which isn’t that bad. And if you’re insane, you could learn to see the zbll case in last slot, but that’s probably impossible/impractical. But I would say that ZZ and roux are equal, not ZZ is better.

 

[CarterK]

If you know full 2sided zbll recog

The problem here is that you won't know this unless you are insane, since you need to learn all 493 algs and the recognition for them. Roux you don't need nearly as many algs, not even as many as OLL.

 

[_zoux]

Wel advaned is roux actually aleast 150+ algs, and "you don't need 500 algs for roux" because it's unextendable

 

[PapaSmurf]

Tbh, 493 algs aren’t that bad. 21 are pll, the rest have good recog. You just have to learn it.

 

[CarterK]

Wel advaned is roux actually aleast 150+ algs, and "you don't need 500 algs for roux" because it's unextendable

Well that's just leaning different CMLLs to predict EO. It's definitely extendable, just not using algs, EOLR is an example of that.


I don't know about you, but I know most people don't want to learn 500 algs, especially if 3x3 isn't their only event. A lot of them definitely don't have good recognition, I'm not sure what you are talking about.

 

[PapaSmurf]

Have you tried to learn zbll? They have good recog, you just have to learn. (Look at my sig, I have experience with this stuff). Also, ask someone like tao yu.

 

[CarterK]

Have you tried to learn zbll? They have good recog, you just have to learn. (Look at my sig, I have experience with this stuff). Also, ask someone like tao yu.

Yes, I know about half of it. A lot of the recognition I wouldn't call good, especially compared to stuff like OLL and COLL. And as I said, most people don't want to learn 500 algs.

 

[PapaSmurf]

But most people don’t bother to be world class. You can get sub 10 with zz and ocll/pll which is 28 algs, all with good recog. Similarly, csp has bad recog but people learn it to be faster. I know it’s not a direct comparison, but the principle carries. The recog can be learnt and made adequate enough to benefit. For example, you could probably sub 2.5 ll every time, quite often getting sub 2 ll if you got your execution and recog down, which is equivalent to lse. I would also say that F2L and blocks are comparable and cmll and EoLine are too. A sub 1.5 EoLine is entirely possible, easily possible. Similarly to cmll+recog by like maybe .1 to .3 seconds. Also, blocks are equivalent.

 

[CarterK]

But most people don’t bother to be world class. You can get sub 10 with zz and ocll/pll which is 28 algs, all with good recog. Similarly, csp has bad recog but people learn it to be faster. I know it’s not a direct comparison, but the principle carries. The recog can be learnt and made adequate enough to benefit. For example, you could probably sub 2.5 ll every time, quite often getting sub 2 ll if you got your execution and recog down, which is equivalent to lse. I would also say that F2L and blocks are comparable and cmll and EoLine are too. A sub 1.5 EoLine is entirely possible, easily possible. Similarly to cmll+recog by like maybe .1 to .3 seconds. Also, blocks are equivalent.

OK gonna go through each thing you said:

Yes, and without ZBLL, ZZ is on average 55 moves.

Recognition for CSP doesn't matter as long as it's in inspection - it doesn't take up any solve time. I'm not saying ZBLL is bad, in fact I think it's better than ocll/pll, but it's very time consuming to get good with when you could be spending that time on something else that's much more worth it.

CMLL + Recog I'm not sure what the times usually are, but I think it's around what you said, probably a bit lower.

Blocks are most definitely NOT equivalent. FB is usually done in 8 moves or so and can be done regripless if you know what you are doing(and can be planned in inspection!) and second block has a lot more freedom with Rw/M moves.

 

[PapaSmurf]

Second block has a lot of regrips. And a lot of moves relatively. I get your point of FB being good, but second block isn’t brill, and there are also more moves in cmll. So I’d say that they’re mostly equal, but the more pieces you solve in algs makes up for that.

 

[Petro Leum]

ZBLL is an investment for the long haul.

Sure, in the time you take to learn and drill 500 algs, you could e.g. work on lookahead, which is a lot more worthwile. However, as you become faster, working on lookahead/F2L solutions brings less and less to the table, and the small difference ZBLL makes becomes much more important and noticeable. And when you get to that point, you will wish you had learned ZBLL ages ago; as with every alg set, and especially because it is so large, the earlier you learn it, the better you get used to it.

ZBLL, while a lot of effort to learn and maintain, is very possible for anyone with the dedication, and I'm sure it will become one of the basics to learn for competitive speedcubers in the near future.

 

[PapaSmurf]

I think we should do something like this: FB=1 sec, SB=1.5 sec, CMLL=1.2 sec, LSE =2 sec. 5.7 theoretical average.
EoLine=1 sec, F2L=2.7 sec, ZBLL=2 sec. 5.7 theoretical average.
Please note that these numbers are pushing limits for both of the methods, so an extra .2 or 3 seconds will be expected for a non near perfect solver who is still world class.
CFOP would be something like 0.5, 3.7, 0.8, 1.2, probably .4 seconds faster with an x cross.

If you think that I’ve got something wrong, say.

 

[CarterK]

I think we should do something like this: FB=1 sec, SB=1.5 sec, CMLL=1.2 sec, LSE =2 sec. 5.7 theoretical average.
EoLine=1 sec, F2L=2.7 sec, ZBLL=2 sec. 5.7 theoretical average.
Please note that these numbers are pushing limits for both of the methods, so an extra .2 or 3 seconds will be expected for a non near perfect solver who is still world class.
CFOP would be something like 0.5, 3.7, 0.8, 1.2, probably .4 seconds faster with an x cross.

If you think that I’ve got something wrong, say.

Where are you even getting those numbers from? Those are totally arbitrary. LSE should be a lot lower for sure, FB and CMLL should be about the same, and SB and LSE should be about the same. I'm not really sure about CFOP and ZZ though.

 

[PapaSmurf]

Tbf, the roux ones are kinda arbitrary, although I’m pretty sure cmll is about 1.2 including recog. If you have more accurate ones, go ahead. But I’d say that the zz ones are pretty good.
Better ones could be: 1.2, 1.8, 1.2, 1.5?

 

[CarterK]

Tbf, the roux ones are kinda arbitrary, although I’m pretty sure cmll is about 1.2 including recog. If you have more accurate ones, go ahead. But I’d say that the zz ones are pretty good.
Better ones could be: 1.2, 1.8, 1.2, 1.5?

Sure, go ahead and make the numbers whatever you want. You could just take numbers that add up to 5.7, or any other number. We can definitely do better than 5.7 with current methods.

 

[adimare]

I think we should do something like this: FB=1 sec, SB=1.5 sec, CMLL=1.2 sec, LSE =2 sec. 5.7 theoretical average.
EoLine=1 sec, F2L=2.7 sec, ZBLL=2 sec. 5.7 theoretical average.
Please note that these numbers are pushing limits for both of the methods, so an extra .2 or 3 seconds will be expected for a non near perfect solver who is still world class.
CFOP would be something like 0.5, 3.7, 0.8, 1.2, probably .4 seconds faster with an x cross.

Heise is the better method then:

4 squares: 1 sec
Orient edges: 1.2 sec
AB3C: 1.4 sec
Last 3 corners: 2 sec
5.6 sec theoretical average.

Brb switching to Heise. Am I doing it right?

 

[PapaSmurf]

I don’t think people like my logic... ok, you guys give me what you think for roux.