[Help Thread] Skewb Discussion and Help

[Xishem]

I (among other people) would like to see skewb progress more, and...

Since it's not a puzzle that's been highly explored yet, have this thread.

 

[Xishem]

To people who use Sarah's method:

Do you have any tricks for transitioning between step 2 and L4C? Either predicting the L4C case or whatever. My worst transition stage is between those two steps and I always have to do something like... Recognize, do some ugly rotation, and then do the alg.

The best I've been able to do is maybe predict what center will end up on the front side, but that almost never helps me predict which case it will be. Even being able to predict whether you have to do a z or z' after the last sledge/hedge would be pretty useful.

 

[Ranzha]

To people who use my method, do you...owait. No one uses my method.
Carry on.

Woohoo! Skewb thread =) Even though Skewb probably doesn't deserve it.

In actuality, I feel like Skewb hasn't been explored enough as a puzzle. I haven't seen any sort of collaboration (except between Thom and Meep) to further Skewb endeavours, mainly because Skewb's not official.

Some things that'd be cool to explore:
- 2-side L4C+CLL recognition and solving (for Sarah's and my methods)
- An improved Skewb solver (Meep has one in MATLAB iirc)
- Various techniques and substeps (e.g. centre control, corner control, optimal blockbuilding)

I'm looking forward to seeing this thread grow!

 

[tx789]

I want to be sub 10 tips what method. I've though of learning KirMeep but I'm too lazy to learn l5c

 

[Xishem]

I want to be sub 10 tips what method. I've though of learning KirMeep but I'm too lazy to learn l5c

It's pretty easy to get sub-10 quickly with intermediate Sarah's method. You only need to learn one 4-move alg (and its inverse) along with a few cases.

 

[tx789]

It's pretty easy to get sub-10 quickly with intermediate Sarah's method. You only need to learn one 4-move alg (and its inverse) along with a few cases.

I'm sub 20 with just using the method shown in monkeydude1313's(Chris Bird) skewb tutorial

 

[Ranzha]

I'm sub 20 with just using the method shown in monkeydude1313's(Chris Bird) skewb tutorial

Mitch Lane has averaged 11-12 with Chris Bird's method for a long time.
As far as getting sub-10 goes, try experimenting and figure out how you can use that one algorithm to influence and control pieces.

 

[applemobile]

MY SKEWB IS SO NOISEY

 

[tx789]

Mitch Lane has averaged 11-12 with Chris Bird's method for a long time.
As far as getting sub-10 goes, try experimenting and figure out how you can use that one algorithm to influence and control pieces.

yeah I learned l4c centers but can't remember them(I think the z or h-perm I may of forgotten) I want to same new algs for skewb

 

[pepkin88]

When I purchased skewb I wanted to be able to solve it, but not necessarily in shortest time. So I developed a simple method with only 2 algorithms.
Today I checked a method on the Meep's site, especially the peanut case. I think I found a better algorithm (it's 2 gen.):
R' L R L' R L' R' L
Maybe someone will find it useful.

 

[Xishem]

I'm sub 20 with just using the method shown in monkeydude1313's(Chris Bird) skewb tutorial

Scrambles:

Spoiler

R' B' L' U' L U' R B' U B U R L' U' L' U' B'
U' R L B U' B L' R B R' U' L B' L R U B
R B U' L' B R L' U B' L B U' L' U R' B' U
L' U' R B L R B' R U L' U' R' L R' L R' U'
L U' L U L' U' L B' R B R L R U B L R'
B L U B U B' L' R' B U' L' U' B U L B U'
R' B' R B R' L' U B' L U R L' U R' L' R U
L' B L R B R' L' B' U L R L' U' L B L R'
R' U' R U' L U R' U' R' B' U L B U' R U' L'
B' L B U' B' L' B' R' U B U' R' L' B U' L' R
L' U R U L' U' B' U R U L' R U' R U' L' B'
B' L R' B U B' U' R U L' R' L' U B U R L'

Move counts (S = Sarah's method, C = Chris' method):

S 20, C 32
S 25, C 36
S 23, C 31
S 25, C 33
S 14 (2nd step skip), C 38
S 18, C 37
S 18, C 30
S 21, C 29
S 21, C 33
S 13 (2nd step skip), C 13 (3rd step skip); These two solutions were exactly the same.
S 22, C 24
S 24, C 24 (3rd step skip)

Sarah's method mean movecount: 20.3
Chris' method mean movecount: 30.0

This isn't conclusive by any means, but it shows a pretty strong trend that Sarah's method is a lot more efficient than Chris', and that's why it's faster. It uses the same alg, but uses it in a more efficient way.

Keep in mind that this was using the pure versions of each method (intermediate in the case of Sarah's), so any tricks for avoiding bad cases weren't used. This could bring the mean movecount of Chris' method down (or vice versa). I just wanted to show that the general solutions were quite a bit longer in Chris' method.

When I purchased skewb I wanted to be able to solve it, but not necessarily in shortest time. So I developed a simple method with only 2 algorithms.
Today I checked a method on the Meep's site, especially the peanut case. I think I found a better algorithm (it's 2 gen.):
R' L R L' R L' R' L
Maybe someone will find it useful.

That alg doesn't preserve the bottom face.

Edit: But now that I think about it, you could use that and just do L5C on the bottom 5 centers, but this would favor worse cases, I think. You're always going to get some adjacent case since the top (previously bottom) center will never be solved. Not sure if this is a good trade off for an alg that probably isn't much faster -- 2 gens aren't as good on skewb as they are on NxN puzzles, imo.

 

[pepkin88]

That alg doesn't preserve the bottom face.

Edit: But now that I think about it, you could use that and just do L5C on the bottom 5 centers, but this would favor worse cases, I think. You're always going to get some adjacent case since the top (previously bottom) center will never be solved. Not sure if this is a good trade off for an alg that probably isn't much faster -- 2 gens aren't as good on skewb as they are on NxN puzzles, imo.

You're right, I forgot about the bottom center.
I think that 2 gens can be even better in skewb than 3x3, because you often don't need to do any regripping (similar case in pyraminx). But it's just a theory, I'm not fast enough to prove it empirically.
Although some 2 gen algs can be not that good at all.

 

[Xishem]

You're right, I forgot about the bottom center.
I think that 2 gens can be even better in skewb than 3x3, because you often don't need to do any regripping (similar case in pyraminx). But it's just a theory, I'm not fast enough to prove it empirically.
Although some 2 gen algs can be not that good at all.

I can't do 2-gens without any regrips. I can do really fast regrips, but not regripless. How?

 

[pepkin88]

That alg for peanut I do without regrips. Before start my thumbs lay on the front centers, index and middle fingers on the back centers, ring and pinky fingers on the bottom face. I don't know if it's an optimal way to do this, for me it is quite comfortable, but I'm not very fast so that may be misleading.

 

[Meep]

When I purchased skewb I wanted to be able to solve it, but not necessarily in shortest time. So I developed a simple method with only 2 algorithms.
Today I checked a method on the Meep's site, especially the peanut case. I think I found a better algorithm (it's 2 gen.):
R' L R L' R L' R' L
Maybe someone will find it useful.

The alg on the website looks ugly, but the execution feels like R B' R' B y R B' R' B:

But really, it's preferable to just avoid the peanut case entirely by forcing a pi or a skip.

 

[tx789]

Solving a layer and a center at the same time is kinda hard to do in very few moves. Then to solve the cube in one look there's a lot less cases than if you were to not to

Has this been thought of before?

 

[Xishem]

Solving a layer and a center at the same time is kinda hard to do in very few moves. Then to solve the cube in one look there's a lot less cases than if you were to not to

Has this been thought of before?

I'm not sure what you're asking? Are you asking if solving a face (center + 4 corners) + another center has been done?

Ranzha's method does this, but in two steps: 2 corners + center, then the other 2 corners + another adjacent center.

 

[Ranzha]

I'm not sure what you're asking? Are you asking if solving a face (center + 4 corners) + another center has been done?

Ranzha's method does this, but in two steps: 2 corners + center, then the other 2 corners + another adjacent center.

Acubist's solution also did this. His method is cached on the Wayback Machine.
His first step and mine are identical. He came up with it first, and I discovered it independently afterwards.

 

[tx789]

Are you asking if solving a face (center + 4 corners) + another center has been done?
another adjacent center.

I guess

 

[Ranzha]

I guess

My method: