Older cubers discussions

[Jason Green]

Congrats! My first sub-20 was also an 18.xx. Good job!

That's cool. Thanks!

 

[h2f]

I had my first sub 20 tonight, wow! I don't even know if it was a skip or what. I'll be on a cubing high for a bit now!!

Nice feeling, isnt it? Congratulations.

 

[Jason Green]

Nice feeling, isnt it? Congratulations.

Thank you, yeah it was an awesome feeling!

 

[muchacho]

Congrats! Very nice average also.

 

[JanW]

Congratulation indeed, sub 20 is very impressive!

I haven't done any timed solves at all in the last few days. Got too distracted by that commutator stuff, which got me interested in how algs actually work. As a result I've now developed my own system for edge 3-cycles, based on the very simple 2-gen alg R U R U R U' R' U' R', which allows me to solve any edge 3-cycle with 1-2 setup moves. (I know that doesn't align the U-face correctly, but this is intentional, as the last U-face alignment very often will cancel with setup moves and I have a separate rule for how to deal with that.) With variations to that alg I can solve any 3-cycle from UF to BR to any one of UL, UB, UR, FR or DR. It can also of course be done to the left, or from UB to FR instead of UF to BR, and so on. Or any of those cycles in reverse. Basically it is a set of quite straight forward rules that determines how to alter the alg for each case. The algorithm itself stays just as simple all the time. As I can solve such a great amount of different cases, the amount of setup moves can be kept very small, keeping the total move count for each 3-cycle at 10-14. Does any of this sound familiar to anyone? I mean, am I reinventing the wheel here, or should I write a more in depth tutorial on this to share with the cubing world?

 

[Logiqx]

I haven't done any timed solves at all in the last few days. Got too distracted by that commutator stuff, which got me interested in how algs actually work. As a result I've now developed my own system for edge 3-cycles, based on the very simple 2-gen alg R U R U R U' R' U' R', which allows me to solve any edge 3-cycle with 1-2 setup moves.

Yes, I've also used that trick myself when free-styling with just a few random edges to solve (sighted, not blind).

You might want to look at the MU forms of the U-perm as well since they are also good. I prefer MU algs for edge cycles, except on big cubes and solving one-handed (RU algs).

 

[h2f]

Yes, I've also used that trick myself when free-styling with just a few random edges to solve (sighted, not blind).

You might want to look at the MU forms of the U-perm as well since they are also good. I prefer MU algs for edge cycles, except on big cubes and solving one-handed (RU algs).

There are also few M U tricks from Turbo. I mean M U M' U2 M U M' or M U' M' U2 M U' M. Adding few setups you may solve some 3cycles. In my opinion Turbo helps a lot to transist to full 3cycle edges. I also do Uperms with M slice but only in turbo.

 

[JanW]

Yes, I've also used that trick myself when free-styling with just a few random edges to solve (sighted, not blind).

Yeah, I don't know, maybe this is already widely known, but I haven't managed to find any post or video that describes this technique. It's so much more efficient than for example Turbo edges, that I'm a bit surprised it hasn't been defined as a technique. It's more efficient, because it allows a lot more different setups, meaning less setup moves, while it also basically requires only one algorithm. On top of that the R U' R U R U R U' R' U' R2 alg, that I assume many people use for turbo, uses this technique with setups and undo of setups. So why do setup to one alg that does setup to another alg, then undo both setups, when you can just head straight to a shorter alg?

Just to take a couple of examples from Zane's Turbo tutorial. UF->BU->LB, he needs 3 setup moves (L B' R') and an 11 move algorithm (U' R U R' U' M' U R U' Rw' U). I'd do 1 setup move (b') and a 10 mover (U2 R U R U R2 U' R' U' R'). Next example UF->RB->RF he does R b setup, then 11 mover. I'd do x R U' R' U' R' U2 R U R U l. And for the next one, his least favorite turbo case, UF->LD->BL, he suggests S2 L setup and again an 11 mover. If S moves are okay to use, then just do S setup and L' U' L' U' L' U L U L U.

 

[h2f]

It's so much more efficient than for example Turbo edges

I dont think so. I mean - pure comms are more efficient than Turbo, but the techinque you've described is not pure comms. Turbo is efficient because you dont have to think much about setups - they are maxiumum 3 movers and algs are very easy to execute sub2. And you dont have to figure them out. And adding some easy comms you can make them shorter. Your example UF->BU->LB you can do: [F' L2 F, M] which takes only 8 moves. That's why this technique is not best option - better is to learn pure comms.

Edit for UF - RB - RF i would do B' and [R F R' F', M2] but your alg is efficient. For UF - LD - BL it's enough to make B and [F' L' F, M2]. Etc.

Edit2: I think it's fine you try to figure out your way of learning comms but it's not efficient as it might seem. Sometimes is, sometimes it is not. It is good to know few ways of doing it to find best solution.

 

[mark49152]

Just to take a couple of examples from Zane's Turbo tutorial. UF->BU->LB, he needs 3 setup moves (L B' R') and an 11 move algorithm (U' R U R' U' M' U R U' Rw' U). I'd do 1 setup move (b') and a 10 mover (U2 R U R U R2 U' R' U' R').

As an M2 user, I would do B' (U R' U') M2 (U R U') M2 B to solve BU->LB from my DF buffer.

Next example UF->RB->RF he does R b setup, then 11 mover. I'd do x R U' R' U' R' U2 R U R U l.

B M2 (B' R2 B) M2 (B' R2), again from DF.

And for the next one, his least favorite turbo case, UF->LD->BL, he suggests S2 L setup and again an 11 mover. If S moves are okay to use, then just do S setup and L' U' L' U' L' U L U L U.

Regular M2 for this one, so (B L B') M2 (B L' B') (U' L U) M2 (U' L' U). However, unlike TuRBo, M2 doesn't force me to solve two pieces at a time, so I have the option of combining with their other adjacent pieces instead. This example is my letter pair GR. If it came up in a sequence AI GR BW then I would execute as A (IG) (RB) W instead. IG = M2 (B L B') M (B L' B) M. RB = L (U M' U2 M U) L'.

 

[JanW]

I dont think so. I mean - pure comms are more efficient than Turbo, but the techinque you've described is not pure comms. Turbo is efficient because you dont have to think much about setups - they are maxiumum 3 movers and algs are very easy to execute sub2. And you dont have to figure them out. And adding some easy comms you can make them shorter.

Technically it is a kind of commutator as well. UF-BR-UR could be written as [U R U R, U], UF-BR-DR is [U R U R, U'] and UF-BR-FR is [U R U R, U2]. The only difference is that the interchange move is undone in the other layer, so the execution of the first example would be (U R U R) U (R' U' R' U') R', the second (U R U R) U' (R' U' R' U') R and the last one becomes (U R U R) U2 (R' U' R' U') R2.

I don't quite think of thes like that though, I only look at the locations of one piece and where it should go. The important piece is the one that stays in the interchange layer.

The steps for UF to 2 pieces on outer right layer are:
1. You should have one piece in UF and two pieces on outer right layer (UR, BR, DR or FR)
2. Notice which of the pieces is supposed to stay on the right layer and take note in which direction it is moving and where it should end up. For now I'll call this piece P.
3. Move P to BR
4. Do U R U R ?? R' U' R' U' where ?? will be U' if the P is moving clockwise around R layer, U if P is moving counterclockwise and U2 if P should move 180 degrees.
5. Once done, P will be back at BR, bring it from there to the location it should end up

So for example UF-FR-DR, the R layer is the interchange layer with two pieces and the U layer has one lone piece. The piece currently at FR is the one staying in the interchange layer, and I notice that it should do a counterclockwise move to DR. I bring it to BR with R2, then do (U R U R) U (R' U' R' U'), then bring it from BR to DR with an R move. This way I automatically take care of any cancellations in the R layer moves.

If I have 2 pieces on U layer and one piece in BR, it works the same way. Bring the piece that is supposed to stay on U layer to UF, do alg R U R U ?? U' R' U' R', where ?? is R' if the piece on U layer is moving clockwise, R if moving counterclockwise and R2 if a double move, then finish by bringing that piece from UF to it's final destination.

To expand to further directions, all you need to know is that the piece travels within it's own layer should be opposite to the single piece in the other layer (for example UB to 2 pieces on outer left layer, then the piece staying on left layer should be brought to FL) and the alg always starts by bringing the single piece away from the other layer. So UB to 2 piece on outer left layer, you start by moving UB away from L layer and it becomes [U L U L, ??]. Same rules apply to the interchange move as earlier. UF to 2 pieces on left layer starts by moving UF away from L layer, so it becomes U' L' U' L', and so on. This might appear confusing at first, but after a few dozen solves with this technique, the execution part of the alg has become pretty much automatic. I don't need to think if it should be start U' R' or R U or whatever, just move the single piece away from the other layer and the rest follows automatically.

Sure, there are certainly better ways for many cases as well. I've now only tried to find out how far this technique can be stretched, which is why I've done solves where I apply this to every 3-cycle. So far I haven't found a case that would require more than 2 setup moves. For real solves I would for sure use some basic M-slice commutators as well. I mean, a 4-move commutator is a 4-move commutator and there's no competing with that. But for freestyling, these rules can surely help you come up with good algs for many different cases.

 

[h2f]

Thanks for the explanation. Now it looks more clear to me. I wonder where it will bring you.

 

[JanW]

I wonder where it will bring you.

At least it brought me my first blind edge solves. Based on these first few blind attempts, I'd say it is definitely easier than corner commutators. Accuracy was also pretty good. While only 3 of my first 6 attempts were successful, two of the failed attempts were caused by me not knowing my letter scheme. Once I memorized the wrong letter, the other time I memorized Q, but when it came up in the solve I somehow read Q as sticker location X instead... In all those solves there was only one mistake in execution. I solved a cycle in the wrong direction. I have a feeling this could be quite fast once I get the hang of it. But next I'll drill those parity algs so that I can attempt full blind solves.

 

[mark49152]

First five 3BLD solves of the day: 2:12, 2:13, 2:15, 1:45, 2:12. Ao5 = 2:12.93. Followed by a few DNFs . I think more 4BLD practice is helping my 3BLD too!

 

[h2f]

Nice Mark.

During last month I did around 300 solves with 3style/turbo. Yesterday I had first two sub 1:30 in a row. I could do mo3 sub 1:30 but dnfed third. I'm around my M2 level at the time maybe a little slower.

 

[mark49152]

Nice Mark.

During last month I did around 300 solves with 3style/turbo. Yesterday I had first two sub 1:30 in a row. I could do mo3 sub 1:30 but dnfed third. I'm around my M2 level at the time maybe a little slower.

Good progress!

I'm feeling confident enough now with my advanced M2 that I'm motivated to start looking at corner comms again. I'll have another try at including some in my solves. At the moment I don't use them because they make me pause.

 

[MarcelP]

First five 3BLD solves of the day: 2:12, 2:13, 2:15, 1:45, 2:12. Ao5 = 2:12.93. Followed by a few DNFs . I think more 4BLD practice is helping my 3BLD too!

Wow, that is pretty professional I have not started practicing BLD yet but have watched my first M2 video on Youtube.. I still have to watch it a few times again and start doing it myself.. But time is far and between lately...

 

[h2f]

Good progress!

I'm feeling confident enough now with my advanced M2 that I'm motivated to start looking at corner comms again. I'll have another try at including some in my solves. At the moment I don't use them because they make me pause.

Thanks. I must start including more edge comms. Corner comms still makes me psuse too but it's worth. You need tons of solves to make them fast, but when I got easy cases in my memo I can do it very fast. So it's worth switching.

 

[muchacho]

BLD progress:
- corners: None, I'll use Boomerang but I'll try to find better algs.
- edges: Learned lettering scheme and most of the setup moves.
- memo: I've generated a list of words and I've been practicing (just a bit) to generate sentences and remembering them.

3x3 progress: Not much, but Ao100 today, 31.91 (edit: 31.78 now) with the Tanglong I received a few days ago, it's good, I just need to lube it and resticker it.

 

[mark49152]

Thanks. I must start including more edge comms. Corner comms still makes me psuse too but it's worth. You need tons of solves to make them fast, but when I got easy cases in my memo I can do it very fast. So it's worth switching.

In sighted, untimed solves it's easier, but in timed solves I always find myself going for the safer, pause-free option. Actually at UKC I found myself doing basic M2 and avoiding any of the M2 tricks I've been using for weeks at home, for the same reason. I was quite frustrated with myself for that.

I'm focusing on the very easiest corner cases first, like GC, SC, GK, SK and their inverses. Once I have the hang of spotting and including comms, I'll add cases.

@Marcel: Looking forward to you joining in with 3BLD

Nice clip of Ben succeeding in that Toy Testers TV video