The New Method / Substep / Concept Idea Thread

[Smiles]

Well even this would be using algs that do too much 'work'. WV specifically orients corners while preserving LL edge orientation, which is somewhat of a restriction, since at this point EO doesn't matter. More efficient algorithms could be generated that would orient corners (or even permute them too) while disregarding EO and inserting the last F2L pair.

Isn't that exactly that OLS does? it has every possible EO so the algs are already generated for you. you just choose the nicest one.\

edit: in case my previous post wasn't clear, i meant use OLS algs for every corner orientation

 

[obelisk477]

Well this could be expanded even further into disregarding DF DL DR and DB. So probably you would find a most efficient one out of all the OLS algs, but maybe not since these preserve the first layer edges

 

[Smiles]

Well this could be expanded even further into disregarding DF DL DR and DB. So probably you would find a most efficient one out of all the OLS algs, but maybe not since these preserve the first layer edges

to my understanding 3 FLE were already completed. maybe i looked at the method wrong. otherwise inserting those edges would be a real pain because of forced rotation, unless they're F/B or L/R.

edit: the advanced part says to blockbuild to get some of the FLE already in there, which makes sense rather than just making columns which is slower when you have to insert the dedges later.

but i guess you could have an alg that only disregards one dedge, but then the orientation of your last pair would be a factor because it might cause you to rotate, and you'd have to learn twice as many algs compared to learning the best OLS alg.

it's kinda like CMLL vs CLL, and if you only know 1 alg per case then CLL is barely worse than CMLL. practically, i doubt disregarding one cross dedge will make much of a difference.

edit 2: if and if you do columns first then it might as well be PCMS and finish with CMSLL rather than COLS, and then the 2 slices after that.

 

[molarmanful]

to my understanding 3 FLE were already completed. maybe i looked at the method wrong.

3 FLE's CAN be completed. It depends on how you blockbuild: if you do a 2x2x2, you finish 2 edges and 1 F2L pair, if you do a 1x2x3, you finish 1 edge and 2 F2L pairs, and if you do a 2x2x3, you finish 3 edges and 2 F2L pairs. Combinations of different blocks might help, too, as long as your combination leaves one edge to be solved.

 

[2180161]

What if you permuted your corners? So like with the LLC, you would instead use COLL+1, which has 96 algs (I think) but would have to insert your F2l edge with it, but hey. Oh, by the way I got a sub-30 time with this. I like it.

 

[qqwref]

Skewb method ("pyraskewb"?):
- V: Solve one corner, the 3 centers next to it, and the 3 corners across from the first corner.
- L7P: Solve last 3 centers + last 4 corners in one look. Suppose you hold these so the 3 centers are on U, F, and R. There are 324 total cases. However: when the edges are solved you can rotate the cube freely; all ccw center cases are mirrors of a cw center case; and you can rotate UFR to "AUF" the UFR corner.

 

[lerenard]

This method has probably never been thought of before simply because of how counterintuitive it is. It would only ever be useful in something like FMC and only if it were very easy:
Solve a 2x2x2 block
Solve another 2x2x2 block completely opposite the other one (ex. First solve BOY and then GRW)
Solve the remaining 6 edges as you would once you two pairs solved in heise.
Solve the remaining corners with commutators.

As I said, this would only be practical if the movecount for each step was very low, which usually is not the case, and if the opposite 2x2 is so easy, it may just be better to use a premove like F2.

 

[mDiPalma]

if the opposite 2x2 is so easy, it may just be better to use a premove like F2.

unfortunately

i've always been confused by methods that solve 2+ non-connected sections of the cube at once (Roux, corners first, a couple weird methods here and there). it just seems like they're intentionally trying to restrict the free moves that you can use.

 

[GuRoux]

unfortunately

i've always been confused by methods that solve 2+ non-connected sections of the cube at once (Roux, corners first, a couple weird methods here and there). it just seems like they're intentionally trying to restrict the free moves that you can use.

in roux, 3/4 of the steps are free moves. cfop, petrus, and zz, the last layer makes only 1/2 of the steps free moves. so though the scope of the free moves get narrowed quicker, you have a greater percentage of the solve that is free moves.

 

[Hssandwich]

Skewb method ("pyraskewb"?):
- V: Solve one corner, the 3 centers next to it, and the 3 corners across from the first corner.
- L7P: Solve last 3 centers + last 4 corners in one look. Suppose you hold these so the 3 centers are on U, F, and R. There are 324 total cases. However: when the edges are solved you can rotate the cube freely; all ccw center cases are mirrors of a cw center case; and you can rotate UFR to "AUF" the UFR corner.

I thought of this a while ago but I never tried to use it or try it out. Great minds think alike

 

[qqwref]

This method has probably never been thought of before simply because of how counterintuitive it is. It would only ever be useful in something like FMC and only if it were very easy:
Solve a 2x2x2 block
Solve another 2x2x2 block completely opposite the other one (ex. First solve BOY and then GRW)
Solve the remaining 6 edges as you would once you two pairs solved in heise.
Solve the remaining corners with commutators.

I don't know about the rest of it, but I've certainly played around with doing the first two steps (2x2x2 block and opposite 2x2x2 block). Usually then I'd try to blockbuild most of the rest, and solve the end with either commutators or Tripod algs.

 

[molarmanful]

Ok, I have created an accompanying last slot alg set for my Pang method called PLS. The algs and info are here: http://www.molarmanful.com/pls.

PLS is different in that it disregards LLEO and only cares about LLCO, so it is compatible with Pang method (and maybe other methods). Enjoy!

 

[supercavitation]

Ok, I have created an accompanying last slot alg set for my Pang method called PLS. The algs and info are here: http://www.molarmanful.com/pls.

PLS is different in that it disregards LLEO and only cares about LLCO, so it is compatible with Pang method (and maybe other methods). Enjoy!

Your "Pang Method" is still almost exactly the same as Michael Gottlieb's Column's First Method, the only difference being that orienting LL corners can be done in a different spot if you don't want to learn COLS algs. The COLS algs are cool, and might be cool with Michael's Method, but aren't really useful for any other method unless you're willing to learn a massive number of algs for the 1LLL finish (If I remember correctly, because I've generated bad algs for all of the cases, it's 156), and without that, it's just less efficient CLL-ELL.

 

[Smiles]

Ok, I have created an accompanying last slot alg set for my Pang method called PLS. The algs and info are here: http://www.molarmanful.com/pls.

PLS is different in that it disregards LLEO and only cares about LLCO, so it is compatible with Pang method (and maybe other methods). Enjoy!

I've looked at your method and made the suggestion of VLS algs, and looking at it now I'm curious what you think:
what are the advantages of the pang method over CFOP? or Roux? (because it's more similar to these 2)

one thing i notice is this concept similar to MGLS, but MGLS never really took off.

 

[lerenard]

Ok, I have created an accompanying last slot alg set for my Pang method called PLS. The algs and info are here: http://www.molarmanful.com/pls.

PLS is different in that it disregards LLEO and only cares about LLCO, so it is compatible with Pang method (and maybe other methods). Enjoy!

I looked on your website, and for the advanced version you basically just said you can solve the first F2L-1x2x2 however you want, then insert last slot and simultaneously solve CO, which can be done with WV, although you've made an optimized set of algs, which is cool. Then you insert the last edge and do EO, and then PLL. This just sounds like a variant of FreeFOP people may not have thought of. It's not a bad idea, but it just seems like a reordering of steps we already know of. I believe the issue here is that EO step is too slow in some cases, and you don't really gain any benefit from waiting to solve the DF edge. It's just easier to so it during the cross or block building. Whenever I try to do something like this, I always accidentally solve my cross edge anyway.

 

[molarmanful]

I've looked at your method and made the suggestion of VLS algs, and looking at it now I'm curious what you think:
what are the advantages of the pang method over CFOP? or Roux? (because it's more similar to these 2)

one thing i notice is this concept similar to MGLS, but MGLS never really took off.

Advantages over CFOP and Roux:
----------
1) MORE WAYS TO START. If you use CFOP or FreeFOP, you can make a Cross - 1E and solve F2L pairs or something like that in any order. If you use Roux, you can build 1 or 2 1x2x3 blocks and finish up the F2L pairs if you haven't already. If you use Petrus, you can build a 2x2x2 and/or a 2x2x3 block, then finish up F2L pairs.

2) AUTOMATIC OLL SKIP (for those who use PLS). Okay, who doesn't love skips? Besides, an OLL skip increases the chance of a PLL skip. MOAR SKIPZ!

3) PREDICTABLE. It is super easy to predict your last layer by the time you start EFL+LLEO. Because EFL+LLEO uses only slice moves, corner orientation is not affected, meaning you can predict either your OLL/COLL case if you don't use PLS, or whether you have headlights PLL, no headlights PLL, or an EPLL case if you use PLS.

 

[supercavitation]

Advantages over CFOP and Roux:
----------
1) MORE WAYS TO START. If you use CFOP or FreeFOP, you can make a Cross - 1E and solve F2L pairs or something like that in any order. If you use Roux, you can build 1 or 2 1x2x3 blocks and finish up the F2L pairs if you haven't already. If you use Petrus, you can build a 2x2x2 and/or a 2x2x3 block, then finish up F2L pairs.

2) AUTOMATIC OLL SKIP (for those who use PLS). Okay, who doesn't love skips? Besides, an OLL skip increases the chance of a PLL skip. MOAR SKIPZ!

3) PREDICTABLE. It is super easy to predict your last layer by the time you start EFL+LLEO. Because EFL+LLEO uses only slice moves, corner orientation is not affected, meaning you can predict either your OLL/COLL case if you don't use PLS, or whether you have headlights PLL, no headlights PLL, or an EPLL case if you use PLS.

First, more ways to start isn't much of an advantage when you only have 15 seconds of inspection time, which all 3 of the methods you listed take advantage of quite well (ok, CFOP doesn't if you're not CN and don't want to go for first pair during inspection, but switching methods isn't the solution to that).

Second, CFOP and Petrus also have algsets that force automatic OLL skips (Roux doesn't really do OLL), this is not unique to this method. Also, an OLL skip does not change the chance of a PLL skip, it only changes the chance of an LL skip.

Third, lookahead later in the solve comes at the cost of difficult lookahead early in the solve, since you have lots of pieces hidden where you'd normally have cross edges, were you doing CFOP. This is one of the weaknesses normally ascribed to Roux, but with far more hidden pieces than Roux.

Overall count: 1 slight advantage, 1 not an advantage, 1 advantage at a huge cost.

 

[GuRoux]

Advantages over CFOP and Roux:
----------
1) MORE WAYS TO START. If you use CFOP or FreeFOP, you can make a Cross - 1E and solve F2L pairs or something like that in any order. If you use Roux, you can build 1 or 2 1x2x3 blocks and finish up the F2L pairs if you haven't already. If you use Petrus, you can build a 2x2x2 and/or a 2x2x3 block, then finish up F2L pairs.

2) AUTOMATIC OLL SKIP (for those who use PLS). Okay, who doesn't love skips? Besides, an OLL skip increases the chance of a PLL skip. MOAR SKIPZ!

3) PREDICTABLE. It is super easy to predict your last layer by the time you start EFL+LLEO. Because EFL+LLEO uses only slice moves, corner orientation is not affected, meaning you can predict either your OLL/COLL case if you don't use PLS, or whether you have headlights PLL, no headlights PLL, or an EPLL case if you use PLS.

doesn't seem like a bad method. probably around as good as any colunm's first method.

 

[molarmanful]

First, more ways to start isn't much of an advantage when you only have 15 seconds of inspection time, which all 3 of the methods you listed take advantage of quite well (ok, CFOP doesn't if you're not CN and don't want to go for first pair during inspection, but switching methods isn't the solution to that).

Second, CFOP and Petrus also have algsets that force automatic OLL skips (Roux doesn't really do OLL), this is not unique to this method. Also, an OLL skip does not change the chance of a PLL skip, it only changes the chance of an LL skip.

Third, lookahead later in the solve comes at the cost of difficult lookahead early in the solve, since you have lots of pieces hidden where you'd normally have cross edges, were you doing CFOP. This is one of the weaknesses normally ascribed to Roux, but with far more hidden pieces than Roux.

Overall count: 1 slight advantage, 1 not an advantage, 1 advantage at a huge cost.

You seem to be the biggest critic of my method; you've been blasting it from the day I proposed it to this site. I'd like to ask how you would improve this method to your liking. Or do you think that this method is hopeless and that I should just abandon my work? I've heard enough about your misgivings; I have similar doubts. Now it's time I hear other things.

 

[mDiPalma]

ADVANCED
Build a 2x2x2, 2x2x3 (recommended), 1x2x3, or a combination of the previously-mentioned blocks and insert remaining F2L pairs to solve all columns and some FLE's. Move count is lower, but gets even lower with F2LL. This is highly recommended and will force an OLL skip.

PROS
-Forces an OLL cross (and maybe an OLL skip).
-EFL + LLEO uses only M and U, and can be very fast. This step can also be done intuitively.
-Predictable: If you don't use F2LL, you can predict your OLL case when you finish columns. If you DO use F2LL, then you can predict whether you will get headlights or not during PLL when you finish columns.

CONS
-Not very efficient for the most part; only the advanced method with LLEO is sort of decent with efficiency. If you don't know some form of F2LL, you are basically screwed to inefficiency.
-Horrible for OH (obviously).

What do you guys think?

i think it's a really good method.

2x2x2
2 F2L pairs.
pair and PLS
permute U-layer corners while inserting a D-layer edge
last 5 edges per columns first.

or

2x2x2
3 F2L pairs
CLL (of some sort)
insert 1 D-layer edge while orienting U-layer edges.
EP5 (really good alg set)