The New Method / Substep / Concept Idea Thread

[Tao Yu]

Here's a roux F2B variation I thought up of:

I actually used it some time ago to "cheat" roux, as I was bad at blockbuilding at the time.

Method
I am assuming that you use a white cross.

1. Solve two opposite white cross edges. In the other two positions, place any two edges that are not from the middle layer (so you can place any yellow or white edges).
2. Insert all 4 white F2L pairs. This is the purpose of placing the non-middle layer edges in the cross; it allows you to do F2L as normal without looking for edges in the bottom layer.
3. The F2B should now be solved, finish the solve using normal roux.

This is definitely sub 10 able. The question is: is there any point in using this over roux?

One thing that I like about this is that the first step doesn't require a lot of moves, and allows for a lot of freedom. It is often very easy to see your first pair.

Here's an avg12 I got using this (complete solves). This is faster than my roux avg12 PB (10.69), but I'm not sure if it would be faster in the long run. I wouldn't say this is an accurate representation of what I average though.

Average of 12: 10.46
1. 9.81 R2 D2 R' U2 B2 U2 L' F2 U2 F2 R2 U' F D' U' L B2 F D B'
2. (12.64) F2 R2 B2 U' B2 U' L2 U2 R2 D R2 F U R2 B L U2 B' F2 U2 R
3. 10.69 L' B2 U2 B2 R U2 L' U2 L' D' F' U R B R' D2 B U R'
4. 10.77 L D2 B2 R' U2 F2 R' B2 D2 U2 R2 F U' L' B R2 D' F' U R' U'
5. (8.89) F2 D F2 R2 D R2 U L2 U2 L2 D' L R' D R' B' D2 R D2 U2
6. 10.61 U' L2 F2 D' L' F R B D2 L D R2 D2 L2 D' R2 F2 D' B2 U'
7. 10.82 U R2 D2 F2 R2 B2 D' B2 L2 D2 F2 L U2 B' R2 U R2 D R' D
8. 9.91 U2 L' U2 L U2 R' U2 B2 F2 R D2 F' R D' R' B' U' L D' F'
9. 10.38 B' D2 R2 F2 L2 F' R2 F' U2 B2 R B R2 U B2 R F' L2 B' D2
10. 10.39 F2 U' R2 B2 D2 B2 D' U2 R2 F2 L F' R' F' U' L B2 L R' U'
11. 10.30 L2 D U2 B2 U' R2 U2 F2 R2 U F2 L' D F L2 U' B R' U L' U2
12. 10.95 L2 R2 B2 U2 R2 B' U2 F2 R2 U2 F' D L2 R' U' L2 U F' U F D'

I'd be interested to see how fast CFOP users would be able to solve F2B using this method. Could it possibly be faster than traditional roux F2B? Like with Feliks' F2L?

 

[Renslay]

Here's a roux F2B variation I thought up of:

I actually used it some time ago to "cheat" roux, as I was bad at blockbuilding at the time.

Method
I am assuming that you use a white cross.

1. Solve two opposite white cross edges. In the other two positions, place any two edges that are not from the middle layer (so you can place any yellow or white edges).
2. Insert all 4 white F2L pairs. This is the purpose of placing the non-middle layer edges in the cross; it allows you to do F2L as normal without looking for edges in the bottom layer.
3. The F2B should now be solved, finish the solve using normal roux.

Same disadvantage as CFOP: cube rotations (or d moves, F moves, whatever) during F2L/F2B.

The F2B seems much, much less efficient than the regular way. Higher movecount and less fingertrick-friendly.

 

[Tao Yu]

Same disadvantage as CFOP: cube rotations (or d moves, F moves, whatever) during F2L/F2B.

The F2B seems much, much less efficient than the regular way. Higher movecount and less fingertrick-friendly.

Those disadvantages have never slowed down CFOP though.

But now that I think of it, you might as well do normal CFOP. Having CMLL + LSE after F2L probably isn't much better than OLL/PLL.

 

[TDM]

Same disadvantage as CFOP: cube rotations (or d moves, F moves, whatever) during F2L/F2B.

You could use the M slice to change edge's orientations...

Those disadvantages have never slowed down CFOP though.

They have done. CFOP isn't slower than other methods, but it definitely isn't faster than ZZ/Roux either.

 

[TheOneOnTheLeft]

You could use the M slice to change edge's orientations...

You'd have to replace the D edge that this brings into the U layer with another non-E-slice edge, so it's not as useful an option as with Roux.

 

[goodatthis]

No, I know it's a version of CFOP, but it takes away all the good parts of CFOP and incorporates just a few of the less promising aspects of roux. It's clearly less efficient than Roux because you solve the pairs on their own and then insert extra edges. It requires rotations, and you don't have a consistent structure because the empty cross edges will be in different places. This means you have to think during the solve which is never good. It doesn't keep the M slice entirely free, and so makes solving the LL edges a lot slower than it would be if you just did Roux.

Now on lookahead, you missed the point. CFOP lookahead is simple because cross is simple, you get plenty of time TO lookahead, and once cross is done then it reduces pairs into known cases. A partial cross isn't simple because you first have to choose which edges to use in your cross and how many (an extra step), you don't have a full cross foundation to work on, and the basis of CFOP just falls apart. Add to that the fact that you don't get all the edges out of the bottom layer in the first step (the main strength of CFOP) and this just doesn't work. Then if you have a partial cross you need to use MUM' to get the edge there out, adding three moves and the time taken to find the edge, when you could have just inserted it with a single flick when you're doing an efficient full cross.

The yellow edge is an entirely different concept to a partial cross method, it is like non-matching Roux blocks would be to CFOP. You'd always have a complete cross, so the problems I've pointed out don't come into effect. If you don't complete a full cross, there's no point in doing CFOP.

Okay first, can you try to do something other than ruthlessly shoot it down and repeat things I've already said?

1. Yes, I said it's less efficient than Roux. You don't need to repeat that.
2. You have only six places where your cross edges can be. (If you do a 2 edge cross)
3. The LL edges are solved using OLL/PLL, not any other way. The LL edges have nothing to do with this method.
4. Did you really just say that half a cross is harder than a full cross? If you see 2 edges, it is ridiculously simple to just make a cross out of them. 4 moves max. There is literally no intuition involved in placing 2 edges on the D layer. Besides, I've already said that 2 edges adjacent to each other is the best. And it does not make it any harder if you do not have cross edges in place to know where your pairs should go, you should just know the color scheme. Plus you have the centers.

Honestly this isn't even constructive criticism, this is downright disregarding any possible idea. Here's what you should do: Try it. Try using it. I've done like 10 solves using this method and I've gotten easily 35 seconds. I normally average 25. Off of no practice.

Just because you don't like my method in general, for whatever reason (which you're entitled to) don't start coming up with fallacious arguments to support that claim. (Like how a partial cross is more complicated than a full one) As I said, there's kinks to be worked out, and if this method is just not for you, just stop telling me how it won't work and how I'm wrong. Okay? Thank you, I've kind of lost my patience at this point.

 

[10461394944000]

Method
I am assuming that you use a white cross.

1. Solve two opposite white cross edges. In the other two positions, place any two edges that are not from the middle layer (so you can place any yellow or white edges).
2. Insert all 4 white F2L pairs. This is the purpose of placing the non-middle layer edges in the cross; it allows you to do F2L as normal without looking for edges in the bottom layer.
3. The F2B should now be solved, finish the solve using normal roux.

this method has been invented almost as much as the belt method and it is stupid.

 

[TDM]

You'd have to replace the D edge that this brings into the U layer with another non-E-slice edge, so it's not as useful an option as with Roux.

I meant with Tao's suggested method, not with CFOP.

 

[Hypocrism]

Okay first, can you try to do something other than ruthlessly shoot it down and repeat things I've already said?

1. Yes, I said it's less efficient than Roux. You don't need to repeat that.
2. You have only six places where your cross edges can be. (If you do a 2 edge cross)
3. The LL edges are solved using OLL/PLL, not any other way. The LL edges have nothing to do with this method.
4. Did you really just say that half a cross is harder than a full cross? If you see 2 edges, it is ridiculously simple to just make a cross out of them. 4 moves max. There is literally no intuition involved in placing 2 edges on the D layer. Besides, I've already said that 2 edges adjacent to each other is the best. And it does not make it any harder if you do not have cross edges in place to know where your pairs should go, you should just know the color scheme. Plus you have the centers.

Honestly this isn't even constructive criticism, this is downright disregarding any possible idea. Here's what you should do: Try it. Try using it. I've done like 10 solves using this method and I've gotten easily 35 seconds. I normally average 25. Off of no practice.

Just because you don't like my method in general, for whatever reason (which you're entitled to) don't start coming up with fallacious arguments to support that claim. (Like how a partial cross is more complicated than a full one) As I said, there's kinks to be worked out, and if this method is just not for you, just stop telling me how it won't work and how I'm wrong. Okay? Thank you, I've kind of lost my patience at this point.

I'm not meaning to attack you, just point out potential weaknesses in an interesting idea.

1) You did say that this was as efficient and faster than CFOP pairs. I don't think will end up as efficient, or as fast, as a normal CFOP pairing, because of the problem of having the hidden edges in the D layer. This makes it difficult to look ahead, requires more partial rotations, and makes getting to those edges difficult, hence lengthening the pair solution.

2) I think choosing between 6 potential 2-pieces crosses will take longer than having your cross end-point predetermined as in CFOP! Then the time taken for choosing between them limits the depth you can look into the solve. Plus there are the extra 3-piece crosses. And if you don't take the time to choose a good first cross, you lose some of the positive things about this idea such as the likelihood of being able to take advantage of block-building opportunities for 2x2x1 blocks.

3) I understand that, but surely it would be better if you did solve them in the way of Roux-that's the problem, it would be more efficient to solve the LL edges as you also place the last cross edges. And that's best done with the corners already solved. Putting the cross pieces in at the stage you've suggested actually seems to make the LL worse than leaving them out because it restricts your moveset more. The best option for a partial cross seems to me to be using opposite cross pieces, enabling the use of M-moves for F2L and potentially LL, and then you really are just doing a Roux variant.

4) I mean that the difficulty of a partial cross comes a little bit from choosing which one to do, and mostly from the problem of look ahead into pairs. It's much easier to look into a solve when you're seeing what the state will be after cross than after cross+1 pair, mainly because solving a pair breaks up what you already have in a way that your cross doesn't. It affects a lot of pieces and makes "look-into" difficult. (I don't know if there's a term for looking into your solve before starting, so I'm going to call it "looking-into".)
In a similar way, if you start looking-into your partial-cross F2L, you'll have a lot of trouble looking past the first pair, because it's more difficult to look-into the solve when you're doing pair-solving moves than it is to look-into a cross solution. Add to this the problem of F2L edges in the first layer and it makes lookahead even more difficult than that.
Also I didn't say anything about colour scheme or having problems knowing where the pairs go. At least I don't think I did. If I did then your response is valid.

I really don't want to discourage you, so how about thinking about this: the idea of not completing your cross fully has potential if the aim is to complete a later CFOP step more efficiently than it would be done if you inserted the edge straight away. Maybe that's a more promising way to take the partial cross idea?

Also well done with getting good times. The skills from CFOP are likely transferrable to this method so it makes a change from the regular grind, some people benefit from switching method briefly to start improving again. I know my speed improved when I spend a while doing ZZ instead. If you get really fast, I'll eat my words about the method!

 

[ryanj92]

2nd cross edge: In this case, the edge is flipped in its position. If we do a y and then a U, notice that both the edges in UF and DF are flipped, so you could use some sort of pure flipping alg to flip them both, and cause an all edges oriented OLL. But for our purposes, we will ignore that. This does open up a neat pathway towards new "last slot" algs. (ZBLCE?) So let's undo that U and we'll do a U'.

The 'last slot' algorithms for this already exist!
http://www.speedsolving.com/wiki/index.php/L5EOP

ultimately, i think the only interesting case for this method is two adjacent cross edges solved (maybe three? i think saving moves would be more hassle than it's worth in this situtation)
if you have two opposite edges solved, it basically becomes what Tao Yu described a few posts ago which has been suggested and tried many times to not much success
one cross edge can be either ignored (especially if ignoring it gives you easier pairs) or made into a two edge case easily

Honestly this isn't even constructive criticism, this is downright disregarding any possible idea. Here's what you should do: Try it. Try using it. I've done like 10 solves using this method and I've gotten easily 35 seconds. I normally average 25. Off of no practice.

but so many very similar things get suggested here all the time, and they haven't worked
i'd definitely try and a few hundred solves and then tell us your opinion - 10 solves doesn't tell us anything useful of course your times won't be -that- far from normal, because you're doing the same LL!

 

[notfeliks]

I've used this idea in a few solves, getting times below average. I think it's pretty decent. Just a slight variation on CFOP that I like to call pseudo-cross.

Only when it's convenient to do so at the start of a solve, you solve the cross in such a way that two of the adjacent edges are swapped (they can be solved with M2 U2 M2). Otherwise, if it's not really convenient to do so, then you just solve the cross and proceed with the solve as normal.

Then at any point throughout F2L or OLL, you can change a case you don't like and correct the cross at the same time by doing M2 U2 M2.

If you like all of the cases and end up at PLL, then your solve is most likely going to have been pretty fast anyway, and the quick M2 U2 M2 trigger followed by PLL isn't going to slow you down all that much. Alternatively, you can learn some case control and PLL cases with the two cross edges swapped, but that's probably not really worth it.

Here's an example solve:

(Scramble with blue on top and white in front for a white cross)

B2 U2 B2 D L2 D' L2 B2 R2 U2 F D' B L' F2 D R2 U2 F2 R2

x' y
L' F D // Pseudo-x-cross(!)
U' L' U2 L y' R' U' R // F2L 2
R U' R' U2 y' R' U' R // F2L 3
U' L' U' L U' y' F R' F' R // F2L 4 + edge control
U R U R' U R U2 R' // OLL
U y M2 U2 M2 // Fix cross + case control
R U R' F' R U R' U' R' F R2 U' R' U2 // PLL

PLL was influenced on that solve by fixing the cross in such a way that the orange bar was preserved which forced an easy PLL - in this case a J-perm.

Advantages:

Not something you have to use on every solve
Can make a cross much easier
You can avoid a case you don't particularly like

Disadvantages:

You might have to rotate the cube back so that the swapped edges can be swapped again (ie they are at DL/DR instead of DF/DB) which slows you down a bit
You might have to slow down to notice cases you can change
Initial pair placement might be slightly confusing with the out of place edges
Probably not usable for OH because of the M-slices


Thoughts? Does anyone else already use this?

 

[TheOneOnTheLeft]

Daniel Sheppard used it in the finals of the UK Championships last year: http://cubesolv.es/solve/1796

 

[TDM]

Daniel Sheppard used it in the finals of the UK Championships last year: http://cubesolv.es/solve/1796

I don't think that was on purpose

Also, notfeliks,

this method has been invented almost as much as the belt method and it is stupid.

 

[notfeliks]

I don't think that was on purpose

Also, notfeliks,

Several things:

I never said that I "invented" this. I said it was an idea that I used on occasion, which I put forward, and then proceeded to ask if it was already in existence in the last line of my post.

This is not a "method", nor did I claim it to be so.

Why is this stupid? I put forth ample evidence to invite constructive criticism and you reply by spitting in my face. If you are going to spit in my face, at the very least do it in a way that makes you look like less of a complete joke and more like someone deserving of an ounce of respect.

I apologise if this as already been proposed before, but that does not call for a complete lack of respect.

 

[TDM]

Several things:

I never said that I "invented" this. I said it was an idea that I used on occasion, which I put forward, and then proceeded to ask if it was already in existence in the last line of my post.

This is not a "method", nor did I claim it to be so.

Why is this stupid? I put forth ample evidence to invite constructive criticism and you reply by spitting in my face. If you are going to spit in my face, at the very least do it in a way that makes you look like less of a complete joke and more like someone deserving of an ounce of respect.

I apologise if this as already been proposed before, but that does not call for a complete lack of respect.

Sorry, I'll admit that whilst the idea of that post was relevant, it may have been a bit harsh. It wasn't worded as I would put it, and I apologise.

However, if you do search this thread, it has been mentioned more than once, and generally, the extra substep increases your time by more than you would save by not including it. You'd have to think more during PLL, and as you said, you may have to do unnecessary rotations. It would be better to do a normal cross in many situations, and there you'd have to remember whether you'd done it or not. You'd have to be very careful to not accidentally miss this or do an M2 U2 M2 when you had a normal cross. The disadvantages outweigh the advantages and, as well as for a couple of other reasons I can't remember that have been said before (if you'd searched you'd have found them), it would be better to use normal CFOP most of the time.

 

[Hypocrism]

I've used this idea in a few solves, getting times below average. I think it's pretty decent. Just a slight variation on CFOP that I like to call pseudo-cross.

Only when it's convenient to do so at the start of a solve, you solve the cross in such a way that two of the adjacent edges are swapped (they can be solved with M2 U2 M2). Otherwise, if it's not really convenient to do so, then you just solve the cross and proceed with the solve as normal.

Then at any point throughout F2L or OLL, you can change a case you don't like and correct the cross at the same time by doing M2 U2 M2.

If you like all of the cases and end up at PLL, then your solve is most likely going to have been pretty fast anyway, and the quick M2 U2 M2 trigger followed by PLL isn't going to slow you down all that much. Alternatively, you can learn some case control and PLL cases with the two cross edges swapped, but that's probably not really worth it.

Here's an example solve:

(Scramble with blue on top and white in front for a white cross)

B2 U2 B2 D L2 D' L2 B2 R2 U2 F D' B L' F2 D R2 U2 F2 R2

x' y
L' F D // Pseudo-x-cross(!)
U' L' U2 L y' R' U' R // F2L 2
R U' R' U2 y' R' U' R // F2L 3
U' L' U' L U' y' F R' F' R // F2L 4 + edge control
U R U R' U R U2 R' // OLL
U y M2 U2 M2 // Fix cross + case control
R U R' F' R U R' U' R' F R2 U' R' U2 // PLL

PLL was influenced on that solve by fixing the cross in such a way that the orange bar was preserved which forced an easy PLL - in this case a J-perm.

Advantages:

Not something you have to use on every solve
Can make a cross much easier
You can avoid a case you don't particularly like

Disadvantages:

You might have to rotate the cube back so that the swapped edges can be swapped again (ie they are at DL/DR instead of DF/DB) which slows you down a bit
You might have to slow down to notice cases you can change
Initial pair placement might be slightly confusing with the out of place edges
Probably not usable for OH because of the M-slices


Thoughts? Does anyone else already use this?

I can see it being useful to form a pseudo cross if there's a REALLY easy F2L following a short p-cross, but for use consistently I doubt it (then again if I'm reading you right, it doesn't seem like you intended it for that anyway!)

Just be careful it doesn't devolve into one of the CFOP variants of roux, where you're solving your pseudo cross, doing F2L, CLL and then solving LSE

 

[notfeliks]

Sorry, I'll admit that whilst the idea of that post was relevant, it may have been a bit harsh. It wasn't worded as I would put it, and I apologise.

However, if you do search this thread, it has been mentioned more than once, and generally, the extra substep increases your time by more than you would save by not including it. You'd have to think more during PLL, and as you said, you may have to do unnecessary rotations. It would be better to do a normal cross in many situations, and there you'd have to remember whether you'd done it or not. You'd have to be very careful to not accidentally miss this or do an M2 U2 M2 when you had a normal cross. The disadvantages outweigh the advantages and, as well as for a couple of other reasons I can't remember that have been said before (if you'd searched you'd have found them), it would be better to use normal CFOP most of the time.

Thank you. A few things though: You would only have to 'think' about what to do if you were inexperienced using psuedo-cross. No-one that is sub-20 thinks about what F2L, OLL or PLL case they're solving as they're doing it, so with sufficient practise this becomes the same. On the occasional solve I do with this, I don't forget about or screw up the pseudo-cross. And yes, it would be better to do a normal cross in most situations, but this is, after all, a use-when-convenient substep.

 

[TDM]

Thank you. A few things though: You would only have to 'think' about what to do if you were inexperienced using psuedo-cross. No-one that is sub-20 thinks about what F2L, OLL or PLL case they're solving as they're doing it, so with sufficient practise this becomes the same. On the occasional solve I do with this, I don't forget about or screw up the pseudo-cross. And yes, it would be better to do a normal cross in most situations, but this is, after all, a use-when-convenient substep.

I see what you mean; you'd use it differently to how I was thinking about it i.e. you would only use it occasionally. I was thinking more of using it all the time. If you only use it occasionally, it would probably be something you'd remember when solving.
Also wouldn't a normal cross be useful 50% of the time?

 

[TheOneOnTheLeft]

I don't think that was on purpose

I think it was - this post certainly suggests so to me.

Also, Ben's comment was in response to Tao Yu's post, not notfeliks's.

 

[TDM]

I think it was - this post certainly suggests so to me.

Also, Ben's comment was in response to Tao Yu's post, not notfeliks's.

Well if it's 50% lucky then it wasn't entirely on purpose, but it is hard to determine what he meant by that post.

And I know his comment was meant for Tao's post, but it still applied to notfeliks' too: it was something that had been suggested several times before and had a few obvious disadvantages.