The New Method / Substep / Concept Idea Thread

[PizzaCuber]

Eo is solved in the first step, so you could use U perms.

Hm, im Stupid. Why did I say ell.

 

[Chris_Cube]

So i wanted to use some spicy edge permutation algs to speed up the last steps. Maybe in the style of L5EP. But with Edges of FL and LL. I will try some other things. Maybe this method can be real good or am I wrong?

 

[shadowslice e]

So i wanted to use some spicy edge permutation algs to speed up the last steps. Maybe in the style of L5EP. But with Edges of FL and LL. I will try some other things. Maybe this method can be real good or am I wrong?

Unless you're a fan of 1980s hardware, I don't think this will be particularly good as because you've solved corners, all of your algs will have to destroy and restore the corners with every step. That means that you'll either end up with questionable ergonomics or a really high movecount. The problem is compounded as you solved the D-layer edges first which means that you'll need to destroy and restore even more.

But like I say to most new methods, I'd like to be proven wrong. I'm confident that I'm not, but try to do it if you believe in it enough.

 

[Chris_Cube]

Hey shadowslice,
This is a good point I thought about this while trying this method. Corners won't be destroyed while solving if you use M and U moves or reverse the moves on the R or L layers while doing the edges. Also you won't solve LBL, I found in some testsolves that some D and U Lines can be solved together
I just need good permutation algs

 

[PapaSmurf]

(2x2)

If you have 3/4 of a face completed, how many algs to solve the rest? (basically 1LLSLL except with a face and not a layer)

614 unique cases plus 1 solved for a layer, then multiply that by 6, so 3689 plus 1 solved.

 

[shadowslice e]

Hey shadowslice,
This is a good point I thought about this while trying this method. Corners won't be destroyed while solving if you use M and U moves or reverse the moves on the R or L layers while doing the edges. Also you won't solve LBL, I found in some testsolves that some D and U Lines can be solved together
I just need good permutation algs

I wasn't saying that you'd destroy the corners and have to resolve them with an alg. I'm well aware that you can solve edges using setup/slice/undosetup. My point is that having to do the setup and undo the setup every time solve a couple of edges will mean that you either have to do more moves/piece than most methods or you'll have to resort to a lot of weird ergonomics involving S/E/M/B movegroups or something (a la bld). While these weird movegroups were fine in the earlier days of cubing when hardware sucked, now that hardware is good enough that the main obstacled to fast solving are lookahead and tps, having a weird movegroup is a far bigger detriment than it used to be.

There is a reason more or less pure cf methods died out.

 

[brododragon]

How does WV/SV > PLL compare to ZBLL in methods where EO is solved.

 

[Triangles_are_cubers]

How does WV/SV > PLL compare to ZBLL in methods where EO is solved.

better recognition and less algs. ZBLL is still kinda worth to learn if you are up to the task though.

 

[Triangles_are_cubers]

guys i have an idea which is based on L2L (Last 2 Layers). i called this variant L2L-P, the P at the end stands for Parrot, cause some people on a different social media called me that.

Steps:
1. Make a layer(probably could block build a part of it to make it more efficient)
2. L2L
2a. Solve edges using comms (might have special cases because of the corners)
2b. L4C


This method isn’t for normal speedsolving, because lookahead would be terrible, but could be a decent method for FMC. Has this method been thought of and/or proposed before though?

 

[Athefre]

After the layer, if you do the middle layer edges, CLL, then last four edges, you've got K4 on 3x3.

 

[Triangles_are_cubers]

After the layer, if you do the middle layer edges, CLL, then last four edges, you've got K4 on 3x3.

Thats nice, but I’m going more towards the L2L4 approach than that.

 

[PapaSmurf]

How does WV/SV > PLL compare to ZBLL in methods where EO is solved.

Insert+ZBLL is better. More efficient and fewer looks (set up pair, WV/SV, PLL vs solve pair, ZBLL). Also, a higher proportion of WV/SV cases are bad than normal last slot.

 

[brododragon]

Steps:
1. Make a layer(probably could block build a part of it to make it more efficient)
2. L2L
2a. Solve edges using comms (might have special cases because of the corners)
2b. L4C

What does L2L solve? Also, blockbuilding doesn't really work in only 2 dimensions.

Insert+ZBLL is better. More efficient and fewer looks (set up pair, WV/SV, PLL vs solve pair, ZBLL). Also, a higher proportion of WV/SV cases are bad than normal last slot.

Is WV/SV > PLL worse than just OCLL > PLL?

 

[Triangles_are_cubers]

What does L2L solve? Also, blockbuilding doesn't really work in only 2 dimensions.

Is WV/SV > PLL worse than just OCLL > PLL?

L2L solves last 2 layers of the cube. Blockbuilding could work in some cases, but not all. I’ll probably think of that.

 

[brododragon]

L2L solves last 2 layers of the cube. Blockbuilding could work in some cases, but not all. I’ll probably think of that.

If L2L solves the rest of the cube, what's the point of solving edges and L4C? Also, I meant blockbuilding isn't that efficient in only 2 dimensions.

 

[Triangles_are_cubers]

If L2L solves the rest of the cube, what's the point of solving edges and L4C? Also, I meant blockbuilding isn't that efficient in only 2 dimensions.

I’m trying to make the method still a little bit efficient movecount wise and not making it have ~200 algs (comparison to other L2L variants). Ok, I’ll keep that into account for this method.

 

[brododragon]

I’m trying to make the method still a little bit efficient movecount wise and not making it have ~200 algs (comparison to other L2L variants). Ok, I’ll keep that into account for this method.

Is this method supposed to be efficient or good for speedsolving?

 

[PapaSmurf]

What does L2L solve? Also, blockbuilding doesn't really work in only 2 dimensions.

L2L solves the last 2 layers (so LBL but you solve the last 2 layers at the same time in some sort of way). There have been some really cool ideas, but one that I think could work is: layer, 3 edges in 2 algs, the rest of the corners, L5E. Even better, layer minus corner, keyhole 2 edges, pseudoslot the third plus solve the layer, solve corners, L5E (these both finish the same way as the Zipper-b method). There's certainly a lot of dabbling and experimentation that could be useful.

I’m trying to make the method still a little bit efficient movecount wise and not making it have ~200 algs (comparison to other L2L variants). Ok, I’ll keep that into account for this method.

Make 500 algs the limit, as 500 algs is totally human learnable and you don't want to miss anything really good for the sake of alg count.

Is WV/SV > PLL worse than just OCLL > PLL?

Depends on the case, but full WV/SV->PLL is slightly worse imo.

 

[brododragon]

Depends on the case, but full WV/SV->PLL is slightly worse imo.

What about Speed-Heise > algorithmic L3C? I've heard L3C is one of the best ZBLL substeps, and Speed-Heise isn't that bad.

(Sorry about all the is x or y better questions)

 

[Chris_Cube]

Speed Heise and L3C is Snyder-Method in some sense. But snyder never wrote down his algs