Older cubers discussions

[openseas]

Comms are the cure.

haha, it's true!

As @mark49152 stated, it doesn't seem to be fair to describe comms as algs since they can be derived easily intuitively. I think one of us said that before - learning comms are more like learning intuitive F2L.

Since switching to Orozco (corners, ULB buffer, URB helper), I'm averaging 2 min which is 10~20s slower than my OP era. The main reasons are 1) comms are still not as fast as trained OP 2) need to think (kinda visualize the execution) 3) also even/odd consideration. However, those slow solves which I felt like 3~4 min extremely slow turned out to be 2min solves. The beauty of move count + optimized comms (I'm mostly using Ishaan's) are quite effective. What I'm saying is that I'm turning really slow but end up with not so bad times. Not spamming tps by OP is quite nice feeling + adding accuracy.

I'm summarizing Advanced M2 method (collection of many experts' works including @mark49152 , @Mike Hughey and many others) - will publish next Month once completed.

 

[newtonbase]

I'm summarizing Advanced M2 method (collection of many experts' works including @mark49152 , @Mike Hughey and many others) - will

I look forward to seeing that. I'm doing a lot of edge comms in untimed solves but I'm still learning ways to optimise M2 (and r2) that are nearly as quick to execute and far easier to do without thinking.

 

[h2f]

What I'm saying is that I'm turning really slow but end up with not so bad times. Not spamming tps by OP is quite nice feeling + adding accuracy.

I think it took me almost 2 years to be quite fluent in execution comms but I've wasted a lot of time not to practing comms. Doing solves is not enough. I've changed many of them in December using mostly Ishaan's list though for some cases Im with my ways of doing. And sometimes I think that F/F' setups are just better (and simpler) than 3-4 moves to setup the case. I think now I'm ready to sub-1 constantly.

 

[mark49152]

I use the same awful setup for J and my W is nearly as bad but it's probably too late to change. Comms are the cure.

They are... But you still need that last OP corner in 50% of solves (unless you're doing parity differently).

I used the F2 D setup for a long time but optimised my OP when I optimised my comms recently. For W, I used to use D2 F' but have switched to R2 F. No idea how I managed to pick such bad setups in the first place

 

[newtonbase]

I used the F2 D setup for a long time but optimised my OP when I optimised my comms recently. For W, I used to use D2 F' but have switched to R2 F. No idea how I managed to pick such bad setups in the first place

Yup. That's the one I use. The better one is very obvious. I don't know where it came from either.

 

[newtonbase]

Just checked my notes and the bad setups for J and W are on Noah's tutorial vids

 

[mark49152]

Just checked my notes and the bad setups for J and W are on Noah's tutorial vids

That'll be where I got them, but no excuses anyway . I also changed V from D' F' to R' D' and started using J perm for D instead of F R'.

Any suggestions for F? I hate the F2 setup but can't find anything better. Not really happy with I and X either.

 

[h2f]

About OP targets and parity thing. Have you thought about doing parity else way? I know you do edges first and it means switching UB and UL during it. With last target OP parity you do in a range 17-19 moves (Jperm or Yperm with 2 moves setup). I think it's closer to second amount. If no switch UB and UL you need to add 7 moves. It means 19-26 moves.

You can do parity else way. Switching UBR and UBL while corners means 4-5 moves - all comms in a cycle UBL-...-UBR are nice and easy. During edges you need to add M2 to correct M slice. (+1 move). In the end you do modified Jperms
or Jperm - all 11-13 moves. This means 17-19 moves on average which is close to first way of dealing with parity and shorter to the second. Quite often last target is UB or BU because of cycle breaks so you do only alg for a last target.

These are just few thoughts why I do it this way.

 

[mark49152]

@h2f - Do you mean a modified J perm to switch UBL-UBR and UB-DF? Do you have this alg?

 

[h2f]

@h2f - Do you mean a modified J perm to switch UBL-UBR and UB-DF? Do you have this alg?

Yes, Mark. I found them on Meneghetti's list:
- with DF-UB - U l R U2 r' U' r U2 l' U R' U2
- with DF-BU - U l U' R' U l' U2 r U' r' U2 R U'.

 

[newtonbase]

That'll be where I got them, but no excuses anyway . I also changed V from D' F' to R' D' and started using J perm for D instead of F R'.

Any suggestions for F? I hate the F2 setup but can't find anything better. Not really happy with I and X either.

Looks like I need to change my V too. No idea on F I'm afraid.

Yes, Mark. I found them on Meneghetti's list:
- with DF-UB - U l R U2 r' U' r U2 l' U R' U2
- with DF-BU - U l U' R' U l' U2 r U' r' U2 R U'.

Wow. He has a parity alg for each letter. I already do the edge swap so it's not really worth me changing but its a nice method.

 

[h2f]

Wow. He has a parity alg for each letter. I already do the edge swap so it's not really worth me changing but its a nice method.

The earlier version of the list had only these two algs. The list of parity alg looks complicated in a first glance but I can notice that few of algs are just setups to Jperm or Tperm

 

[JanW]

I'm working on turning my 3-cycles more into algs as well. So far I've worked everything out during the solves based on the same principles, but now I want to optimize the tricky cases.

Here's an interesting alternative I just found for UF-BD-UB, which always has been a tricky case:

U r' U R' U' R' U' R' U R U r R U'

It's a U setup to a U perm with a wide r. To shoot the cycle in the opposite direction I use same alg with U' setup.

I also have a few cases that I notice I can perform the fastest if I do some tricky cube rotations. From what I've seen, blinders don't usually fancy rotations. In particular UF-DB-RB flows very nicely as

x' y' R U R' U' R' U' R' U R U

This is also a U perm with cancellations taken into account. The rotation isn't any slower than any other regrip, I think. Should I still avoid rotations like this?

 

[h2f]

Should I still avoid rotations like this?

Definietly you should avoid y/y' rotations. I think you should start using slices - they make edges much easier. Doing own alg really helps in learning but it's good to check top blinders lists to see if there is a better option.

Here's an interesting alternative I just found for UF-BD-UB, which always has been a tricky case:
x' y' R U R' U' R' U' R' U R U

For example Daniel Lin has this: [U' M' U': [M', U2]] which is U' M' U' M' U2 M U' M U which is shorter and with no roatation.

 

[mark49152]

Yes, Mark. I found them on Meneghetti's list:
- with DF-UB - U l R U2 r' U' r U2 l' U R' U2
- with DF-BU - U l U' R' U l' U2 r U' r' U2 R U'.

I only spent 5 minutes but I can't figure out how to finger trick these without uncomfortable regrips.

The choices for solving parity and last corner target are...

1. Parity alg (7) + Y-perm (15) + OP setups (4 max) = 26 STM
2. UL-UB swap (0) + Y-perm (15) + OP setups (4 max) = 19 STM
3. M2 (1) + extra corner comm (10 typical) + alg from above (12) = 23 STM

Given that Y-perm will have better TPS than the above two algs I still think #1 beats #3. However, #2 beats both.

I asked on here a couple of years ago about number of pairs and someone posted a lengthy proof that with UB-UL swap, there will be one more edge target in 50% of cases and one less in 50% compared to no-swap-and-parity-alg. Given that knowing both methods can be advantageous, e.g. to revert to regular parity to avoid a flip, I expect it would shorten solves in >50% of cases, which is good for an event ranked by single.

My plan remains the same - switch to UB-UL swap. I have been practising tracing that way and am about 1 second slower on average than before. I'm not going to switch to solving that way until after High Wycombe.

Wow. He has a parity alg for each letter. I already do the edge swap so it's not really worth me changing but its a nice method.

I think Gianfranco mentioned at Worlds that he has a parity alg for every corner/edge sticker pair. Like, ~400 algs

The rotation isn't any slower than any other regrip, I think. Should I still avoid rotations like this?

The general advice to avoid rotations is good, but it's a generalisation. Personally I prefer to rotate when it gives an easy case, but limit myself to z and x because they are quite fast and fairly comfortable. I won't even do x' or z' because I don't want to end up with adjacent pairs that require x2 or z2 between them.

For example, z rotations give nice <R,U,D> comms for pairs that combine I, U, S and pairs that combine M, N, O, P. For the latter it would be more intuitive to do z' and interchange on top, but instead I do z for both and interchange M, N, O, P on the D layer to avoid ever needing z2.

Having said that, I'm not keen on my I, U, S solutions because they are A perms. The regrip to do Lw' is more annoying than the z rotation, so they are on my list to get changed.

The only ones I do x for now are I, J, K, L pairs. The x rotation is more awkward than z but the algs are fast enough that it's worth it. Like for KL, I do [x: [R' D2 R, U']].

This all suits me for now but it's not as good as what the fast guys do. If I ever get close to sub-50 I will look to eliminate x rotations and Lw/Lw' regrips, although I might keep z rotations for M, N, O, P a while longer.

 

[h2f]

I only spent 5 minutes but I can't figure out how to finger trick these without uncomfortable regrips.

The choices for solving parity and last corner target are...

1. Parity alg (7) + Y-perm (15) + OP setups (4 max) = 26 STM
2. UL-UB swap (0) + Y-perm (15) + OP setups (4 max) = 19 STM
3. M2 (1) + extra corner comm (10 typical) + alg from above (12) = 23 STM

The second one is just Jperm with wide moves. I think Feliks uses this one (without wide moves) in normal solves. I agree the first is not fingertrick friendly but I used to it.

In the second option UL-UB swap means 1 extra target during edges, I think (around 4 moves?). It's similar to extra comm in 3rd option when you solve last target and add another target. It's really hard to compare all of them but I think option 3 has a potence because few edges - the most I break during cycles like UR, UB, UL - can be solved to pure algs like Jperm. For example if I finish with UR I do last cycle UB-UR and got pure Jperm which is fast.

 

[JanW]

For example Daniel Lin has this: [U' M' U': [M', U2]] which is U' M' U' M' U2 M U' M U which is shorter and with no roatation.

You quoted the wrong alg there. For that case I use the first alg I listed, no rotation.

My problem with slices is mainly that I'm really bad at them. When I've been training with my bldtrainer, the cycles I solve with 4 move commutators like M' U2 M U2 tend to be among the slowest. This is in part because I'm still not seeing the direction of that comm in an instant, need an extra second to figure it out, but also because my M moves are so slow. That example alg from Daniel Lin takes me half an eternity to execute. Maybe I should spend 6 months doing roux, then it would become more comfortable...

Definietly you should avoid y/y' rotations.

What's so special about y rotation? In the example I gave, which started with x' y' R, the right thumb stays on the same face of the cube to immediately strat the R move. Flows very nicely imo. The rotation back is a bit awkward though.

 

[h2f]

ou quoted the wrong alg there. For that case I use the first alg I listed, no rotation.

My problem with slices is mainly that I'm really bad at them.

Sorry, I've missed it. I know the problem with M slices but learning them - and others like E or S - leads to many possiblities. The direction is a matter of practice. For example I know what happens on cube with M, D2 alg but it was hard to me to memorize whats going on with M', D2. Finally I used to it.

What's so special about y rotation?

For me it's around 1 second lost, maybe more. But of course it's personal and there are situations when it's needed.

 

[mark49152]

The second one is just Jperm with wide moves. I think Feliks uses this one (without wide moves) in normal solves. I agree the first is not fingertrick friendly but I used to it.

Ah yes I didn't notice that. I don't use that J perm. I've never been able to get fast at switching between L and R faces. When there's a good <L,U,R> solution I usually z and do it <R,U,D> instead, like the comms I mentioned above. I doubt that would work well with wide turns though.

In the second option UL-UB swap means 1 extra target during edges, I think (around 4 moves?). It's similar to extra comm in 3rd option when you solve last target and add another target.

As I wrote above, it's one extra target for 50% of solves and one less target for 50%, so net change of zero.

 

[h2f]

As I wrote above, it's one extra target for 50% of solves and one less target for 50%, so net change of zero.

Ah ok. I didnt understand it at first time.