Skyler Variation (SV)

[jskyler91]

Just so you know, rotating before having to do "SV" is the worst idea for speed ever. "EJF2L" was a horrible idea was well, and the only use for this "method" would be for the last slot, dropping its actual use cases significantly. The time it would take to solve last f2l -> rotate -> recog SV -> do alg, is WAAAAAY more than just doing the f2l during your solve and OLL after you finish your last pair. Move count isn't everything to a method. People really need to understand that.

Cool, I agree thats why I put multiple options in.

Congratulations for generating all of the these algorithms and images. I think it's definitely worth learning algs for these LS+OLL cases. I'll probably try to generate some algorithms myself for each case, but I think most of them seem decent enough to learn.



Well.. you still need to look at what your "OLL" is. You still generally look at the same number of pieces as standard CFOP.



Just because almost all of the algs are almost RUF gen, doesn't make the algs decent themselves. I very much doubt that these algorithms are faster than OLL alone overall. When I generate algs I tend to try and find algorithms which are comfortable to perform, short in quarter turns, can be executed from start to finish with very few regrips. Don't restrict yourself to <RUF> too much. There sometimes shorter and more decent algorithms containing other turns.

I did that , although some cases just weren't that great. I might regenerate the EUD cases because those were my first and I was sort of lazy for them. If you decide to generate algs start there. Let me know if you find better ones for them

My thoughts on LS+LL methods;

Learning OLS subsets is akin to learning COLL or the CO OLLCP subset for LL. Similar to learning random ZBLL algs.

While magical random cases will get you a good solve when they come up, the extra recognition and recall overhead from a full implementation of a system like this can cause problems. Saving moves means nothing when you're taking time working out how.

My advice is to treat systems like this how Cubenovice is intending to do so.

I still think it is worth is especially since there are a lot of mirrors and inverses

Just correcting a few mistakes in your post;



You forgot to include the setups to "Skyler Variation" which can be up to four moves, and the fact that average movecount for algs will be longer as people replace your <RUF> optimal ones that are likely bad for speedcubing with better ones, making the moves saved closer to 0 than 5.5 on average.



This is very wrong. F2L cases do not have an equal chance of happening, and you've completely ignored cases where pieces are in open slots.



So both of these numbers are wrong


This negates most of the positives listed in your post.

Interesting, how are they not evenly distributed, can you please cite this with some proof or a source? Also cases with open slots are not part of the general 42 which is what I was basing it on. The average setup is 3 moves which still makes this 2.5 moves quicker and lessens the amount of looks and it is still a far more usable system then any other thus far.

[Kirjava]

Just correcting a few mistakes in your post;

Now when I say Versatile I am referring to the fact that Skyler variation is one of the few (only that I know of ) last slot variations that allow for really easy setups which actually save you time and moves as opposed to wasting them. Seeing as how the average move count for Skyler variation is about 12.5 moves (just a rough estimate, could be slightly higher if you chose different algs than I did for each case) and the average move count for OLL is 9.7 moves, or to make it easy 10 moves, then the only way that you would be wasting moves/ time in setting up for Skyler variation would be if your f2l case were 2.5 moves or less which considering that f2l has an average move count of 7.8 moves (8 for simplicities sake) and the shortest f2l case is 3 moves (unless you start form an R or something)means that you will almost always be saving moves (about 5.5 in general). This makes Skyler variation about 5.5 moves shorter than regular CFOP.

You forgot to include the setups to "Skyler Variation" which can be up to four moves, and the fact that average movecount for algs will be longer as people replace your <RUF> optimal ones that are likely bad for speedcubing with better ones, making the moves saved closer to 0 than 5.5 on average.

When I say usable, I refer to the fact that assuming all F2L cases have the same chances of occurring you have a 50% chance of being able to use Skyler variation for each of the 4 slots this means that if you solve all 4 slots you have about a 94.25% chance (I got his by adding .50 (50 percent for the first slot) plus .25 [25 percent chance you will get it on your second slot if not on the first] plus .125 (12.5% chance if you don't get it on your first or 2nd it will happen on your 3rd) plus .0675 (chance is will occur on the last slot assuming you don't get it ont he first 4) which equals .9425 or a 94.25% chance of getting Skyler variation, please let me know if this math is wrong because I fail at math so it wouldn't surprise me) of getting these cases.

This is very wrong. F2L cases do not have an equal chance of happening, and you've completely ignored cases where pieces are in open slots.

This means 9/10 of your solves will be on average 5.5 moves

So both of these numbers are wrong


This negates most of the positives listed in your post.

[Kirjava]

Interesting, how are they not evenly distributed, can you please cite this with some proof or a source?

URUR' and RU'R' are counted as a single case, but are actually two different ones.

Also cases with open slots are not part of the general 42 which is what I was basing it on.

Pieces in open slots produce cases outside of 'the general 42' for non-LS cases, you have to include them to correctly determine the probability.

[aronpm]

Interesting, how are they not evenly distributed, can you please cite this with some proof or a source? Also cases with open slots are not part of the general 42 which is what I was basing it on. The average setup is 3 moves which still makes this 2.5 moves quicker and lessens the amount of looks and it is still a far more usable system then any other thus far.

If you are using mathematics you need to use the actual number of cases, which is NOT 42 for the first slot.

If you have a solved FR pair, that is a certain probability (determined by the number of permutations and orientations). The probability of having the edge solved and the corner in the U layer is 4 times larger (and these are the same F2L case), because there are 4 times more locations for the edge to be, and thus is 4 times more likely

[jskyler91]

URUR' and RU'R' are counted as a single case, but are actually two different ones.



Pieces in open slots produce cases outside of 'the general 42' for non-LS cases, you have to include them to correctly determine the probability.

I don't see how your first part affect that averages. Also, I am not very good with the math stuff, all I know if that I have a high chance of getting these cases.

If you are using mathematics you need to use the actual number of cases, which is NOT 42 for the first slot.

If you have a solved FR pair, that is a certain probability (determined by the number of permutations and orientations). The probability of having the edge solved and the corner in the U layer is 4 times larger (and these are the same F2L case), because there are 4 times more locations for the edge to be, and thus is 4 times more likely

Again, I am really bad with math, if someone wants to generate the exact percentages I would greatly appreciate it.

[masterofthebass]

you really shouldn't be so worried about getting your name on some method no one is ever going to learn. it's not that important and you've already made yourself memorable to the community

[Robert-Y]

You've mentioned less looks in general as a pro, yet in the examples you've posted, you haven't demonstrated this. There are 6 looks for the 2nd example but that's because the first step is x-cross.

[jskyler91]

you really shouldn't be so worried about getting your name on some method no one is ever going to learn. it's not that important and you've already made yourself memorable to the community

Oh, I am not doing this for recognition, I just honestly think it will be a very fast method. Assuming I can memo all of the algs, there is no reason that i see why it wouldn't be at minimum as fast as regular CFOP and i see quite a few reasons why it would be faster. The only reason I am posting this is to share an idea with the community which I think will make them all faster cubers. I said the whole get my name on it thing because I wanted to make sure i didn't spend a bunch of time generating algs and then find that someone else had done it slightly faster than me and they took the credit.

You've mentioned less looks in general as a pro, yet in the examples you've posted, you haven't demonstrated this. There are 6 looks for the 2nd example but that's because the first step is x-cross.

The actual execution of the solves will have less looks in that I will be able to plan out multiple parts while doing the simple RUR' insert. Also, when the pair is in place it is one less look which wil sill be more often then say in Rowe Variation because he will almost always have to recognize the first f2l case, solve it up until it is a pair and then recognize the RV case. This looks thing is a minor addition I just though I would bring up, its not like I think this will really save much time, although it will make looking ahead easier because simple inserts affect less of the cube so you could look ahead better.

Another way of looking at this is that say in RV only 2 f2l cases actually have formed pairs, while in my variation there are 5 cases which come like that so 2.5 times as likely. Also my cases often set up with a simple RuR' whereas his often require 4-5 moves to set up into a pair. Again, I am not bashing RV here, just comparing.

[Kirjava]

I don't see how your first part affect that averages.

It's like how chance of getting a specific PLL isn't 1/22.

Also, I am not very good with the math stuff, all I know if that I have a high chance of getting these cases.

How?!

[masterofthebass]

Oh, I am not doing this for recognition, I just honestly think it will be a very fast method. Assuming I can memo all of the algs, there is no reason that i see why it wouldn't be at minimum as fast as regular CFOP and i see quite a few reasons why it would be faster. The only reason I am posting this is to share an idea with the community which I think will make them all faster cubers. I said the whole get my name on it thing because I wanted to make sure i didn't spend a bunch of time generating algs and then find that someone else had done it slightly faster than me and they took the credit.

its ok. you don't need to lie.