Briggs (3x3x3 method)

[Robert-Y]

Where did you get 24 quarter turns? I count 19. Also, it depends on what 3-cycle is used. This one is one of the worst 3-cycles due to the positioning of the corners which were cycled. Most of the time you won't get a case which requires 2 double turn accessible corners so the regular move count in SQTM would be more likely 1+8+7= 16 moves (it is also a regripless step).

Ok 19, which is still bad. You still haven't explained how to do CP properly. And you say it's more likely to be 16 which is still bad. I've asked a bunch of roux solvers to do an average of 12 for the first block in SQTM and the best of us are getting about 8 or 9 moves for this. I could just do a niklas at LL and get around 16 moves too. So why are you proposing to solve CP so early on. What led you to believe that this was a good idea? Also it's a regripless step? So what? People use a number of regrips in CFOP solves and can still achieve decent times. Is this such a big thing to you? It isn't to me.

Also, LOOKAHEAD. You can plan the entirety of the first step in the inspection without too much effort once you have worked out the 3-cycle to correct CP as well as some of the second step if you are good with lookahead (or 3-cycles if you like BLD). And you don't have to try to recognise corner permutation later in the solve which will inevitably lead to a long pause or wasted moves getting RD corners to their positions. So this is much more fluid for the whole solve.

"You can plan...without too much effort" I'd like it if you can back up this claim for me, otherwise I don't believe it. I don't know you think recognizing corner permutation later in the solve will "inevitably lead to a long pause" when most of the top solvers can recognize it fairly fast with enough practice. Have you tried it yourself? Or is this another claim without much evidence to back it up?

Lastly, if you go FB then CP the way that is described by the method you posted, you still get around 8+3+3=14 moves (and these are some of the lower numbers assuming good blockbuilding) and I'll be damned if some of them aren't double turns (so the actual movecount in SQTM would be at least 16) and then there is still the aforementioned problem with lookahead. Thus, I would say that 3 extra moves (or none in STM plus we haven't factored in how my method movecount could be lower if the corners weren't in such horrible positions) for a greatly increased lookahead and less algs is a good tradeoff.

"greatly increased lookahead"...I don't believe you. Again, I would like evidence to back this up.

So, yes Rob, I agree with you that movecount in SQTM is important (which is why this method aims for a movecount of 50- about the same as the "big four" (discounting petrus))but other things are important to consider as well; the lookahead given by this method is much better than many other methods which can help to give more fluid solves and forcing of good cases in the way in which could potentially lower the movecount as well.

Yes, I think SQTM is somewhat important, or at least we should stop using HTM for speedsolves. Yeah I don't believe that the lookahead given by the method is much better than many other methods. Seems like you just made this up to me.

 

[shadowslice e]

Ok 19, which is still bad. You still haven't explained how to do CP properly. And you say it's more likely to be 16 which is still bad. I've asked a bunch of roux solvers to do an average of 12 for the first block in SQTM and the best of us are getting about 8 or 9 moves for this. I could just do a niklas at LL and get around 16 moves too. So why are you proposing to solve CP so early on. What led you to believe that this was a good idea?

Ok, first rob, you forgot to factor in how you would need to place the DL corners after FB so that's a few more moves after the FB which combined with the niklas is around 18 moves anyway and you would have to recognise the CP case before excecuting which is time that you will be able to avoid when doing my first step. In addition, I say 16 moves which is: solve 1x1x2 (1)+ 3-cycle+ setup (8)+ finish FB (7- and this is being conservative considering you only have 2 edges to place)

You can plan...without too much effort" I'd like it if you can back up this claim for me, otherwise I don't believe it. I don't know you think recognizing corner permutation later in the solve will "inevitably lead to a long pause" when most of the top solvers can recognize it fairly fast with enough practice. Have you tried it yourself? Or is this another claim without much evidence to back it up?

My claim for this is that the CP will not affect the edges at all (as you are doing a corner 3-cycle) so they can be planned during inspection whereas CP cannot be planned so easily during the inspection. Certainly, when I solve, I can somewhat predict where the DB edge will be after FB and CP. In addition, you need to practise your recognition a lot more if you want to recognise CP quickly.

Regarding the "long pause", I'll admit i was exaggerating a bit but i was comparing it to the pause that occurs in my method: namely, none (at least during the first step).

You still haven't explained how to do CP properly

Ok, this I will admit I am guilty of. In the most simple way I can describe it, you solve the remaining DR corner while permuting the remaining ones to a 2-gen solve using either a 3-cycle or a 2-swap.

To do this, you must first recognise what set of 2-swaps there are present (images of which are in the website link I provided in the OP; imagine the diagrams as if you were looking at the cube after a y rotation) the way that I do this is by finding the two DR corners and treating them as the first 2-swap. The remaining pairs are of opposite colours apart from on the U facelet (so a red/green would be paired with a orange/blue). To be 2-gen, they must fit into the 2-swap pattern that is dictated be the locations of the DR corners.

Clearly, as one of the corners in the pattern must be that of the remaining DL corner, the pattern is not complete so must be fixed after identifying what corner needs to be in the place it currently occupies.

The corner which is currently in the DL must also be cycled, so the three corners are cycled together with a 3-cycle (you can do this in less moves but I don't because the three cycle mean the edges can stay in place and so be planned more easily).

I'm afraid that that's the best explanation I can give in type. I will try to make it clearer but that's all I have ATM. Hopefully I'll be able to make it clearer if I verbally explain it to someone IRL and convert that to written form.

 

[Robert-Y]

Ok, first rob, you forgot to factor in how you would need to place the DL corners after FB so that's a few more moves after the FB which combined with the niklas is around 18 moves anyway and you would have to recognise the CP case before excecuting which is time that you will be able to avoid when doing my first step. In addition, I say 16 moves which is: solve 1x1x2 (1)+ 3-cycle+ setup (8)+ finish FB (7- and this is being conservative considering you only have 2 edges to place)

Maybe I wasn't clear enough or that I misunderstand what you're trying to say. I'm saying that solving the first 1x2x3 block (and placing it on DL) seems to require about 8 or 9 moves for a "decent" Roux solver. However I think your second point is kinda fair although I still think that we can recognize the CP case fairly fast with enough practice. We can already recognize CxLL cases fairly fast, I think it's just a matter of learning what corner swap each CxLL case corresponds to.

My claim for this is that the CP will not affect the edges at all (as you are doing a corner 3-cycle) so they can be planned during inspection whereas CP cannot be planned so easily during the inspection. Certainly, when I solve, I can somewhat predict where the DB edge will be after FB and CP. In addition, you need to practise your recognition a lot more if you want to recognise CP quickly.

I don't know if this is a fair comparison but with CFOP, many of us can plan the cross and the first pair during inspection which is 6 pieces which is the same number of pieces for FB and DB so I do not see this as an advantage when I look at it this way. (Nor do I see it as a disadvantage)

Ok, this I will admit I am guilty of. In the most simple way I can describe it, you solve the remaining DR corner while permuting the remaining ones to a 2-gen solve using either a 3-cycle or a 2-swap.

To do this, you must first recognise what set of 2-swaps there are present (images of which are in the website link I provided in the OP; imagine the diagrams as if you were looking at the cube after a y rotation) the way that I do this is by finding the two DR corners and treating them as the first 2-swap. The remaining pairs are of opposite colours apart from on the U facelet (so a red/green would be paired with a orange/blue). To be 2-gen, they must fit into the 2-swap pattern that is dictated be the locations of the DR corners.

Clearly, as one of the corners in the pattern must be that of the remaining DL corner, the pattern is not complete so must be fixed after identifying what corner needs to be in the place it currently occupies.

The corner which is currently in the DL must also be cycled, so the three corners are cycled together with a 3-cycle (you can do this in less moves but I don't because the three cycle mean the edges can stay in place and so be planned more easily).

I'm afraid that that's the best explanation I can give in type. I will try to make it clearer but that's all I have ATM. Hopefully I'll be able to make it clearer if I verbally explain it to someone IRL and convert that to written form.

Alright thanks for trying. I think I have a better idea of how this step is accomplished now. Still, if you or anyone has some more examples, that would be great to take a look at.

 

[shadowslice e]

I don't know if this is a fair comparison but with CFOP, many of us can plan the cross and the first pair during inspection which is 6 pieces which is the same number of pieces for FB and DB so I do not see this as an advantage when I look at it this way. (Nor do I see it as a disadvantage)

well, it's an ok comparison but you don't do CP or Eo in CFOP so it's not quite as simple as Cross+1 in that although you are solving the same number of pieces, you are influencing many more than CFOP does (at least normally).

Maybe I wasn't clear enough or that I misunderstand what you're trying to say. I'm saying that solving the first 1x2x3 block (and placing it on DL) seems to require about 8 or 9 moves for a "decent" Roux solver

I get what you are saying here in that FB can be solved pretty efficiently but (in almost every method that I have seen which deals with corner permutation) the DL corners must be placed in order to permute the L6C using a form of CPLL in order to avoid some really horrible recognition for all the corners at the same time. However, this adds moves to the overall movecount and the recognition for the CP is still not great.

Alright thanks for trying. I think I have a better idea of how this step is accomplished now. Still, if you or anyone has some more examples, that would be great to take a look at.

This I will be working on but I'm slow with notation because I'm terrible at writing it. I will try to get at least a couple more first step examples by tomorrow evening (GMT).

 

[whauk]

An idea for CP after FB:
After placing the first block on the left as usual you have a "natural" bottom color. Find the two corners of this color and put them on D in any orientation (should require not much more than 3 moves.
Now check if any corners on U need to be swapped for being solved. If two adjacent ones need to be swapped apply F R F' or F R' F' in order to swap the two corners along UB. If two diagonal corners need to be switched L' U R2 U' L does the job. Notice that the D corners don't need to be in their correct position. If they are swapped you have to produce a diagonal swap on the U-layer in order to reduce the cube to <Rw,R,U>.
Example: (please notice, that I don't do Roux very often, so my block might be really bad.)
F U2 F L2 B D2 L2 D2 R2 F' D2 R' B U L2 F L B' F2 L2 D

FB: D F2 L F2 Rw2 B' U2 B2
place D-corners: R' U' R2. Now CP requires to swap URF and URB, but the D-corners are switched so we swap the other two with U F R F'.
The cube is in <Rw,R,U> now.

IMO a very big plus of this method is that LL-recognition reduces to 2GLL, which is a very manageable one-look-LL-set.

 

[shadowslice e]

An idea for CP after FB:
After placing the first block on the left as usual you have a "natural" bottom color. Find the two corners of this color and put them on D in any orientation (should require not much more than 3 moves.
Now check if any corners on U need to be swapped for being solved. If two adjacent ones need to be swapped apply F R F' or F R' F' in order to swap the two corners along UB. If two diagonal corners need to be switched L' U R2 U' L does the job. Notice that the D corners don't need to be in their correct position. If they are swapped you have to produce a diagonal swap on the U-layer in order to reduce the cube to <Rw,R,U>.
Example: (please notice, that I don't do Roux very often, so my block might be really bad.)
F U2 F L2 B D2 L2 D2 R2 F' D2 R' B U L2 F L B' F2 L2 D

FB: D F2 L F2 Rw2 B' U2 B2
place D-corners: R' U' R2. Now CP requires to swap URF and URB, but the D-corners are switched so we swap the other two with U F R F'.
The cube is in <Rw,R,U> now.

IMO a very big plus of this method is that LL-recognition reduces to 2GLL, which is a very manageable one-look-LL-set.

https://www.speedsolving.com/forum/showthread.php?41986-Noah-s-CP-Block-Method-2-0

Also, you are essentially doubling up moves as well as having to pause during the first step which is what I was trying to avoid by doing CP earlier during inspection.

Also it doesn't provide too much of a movecount decrease which is lost as soon as you have to pause to recognise CP.

 

[shadowslice e]

Sorry it's really, really late but I have 3 more examples of how to do CP. Please not that these may not be the best way to execute the moves in terms of orientation going into the rest of FB.

1) Scramble: D2 R U B L' B2 L' R U' D F' U R' F D' B2 L2 F

z'// inspection +1x2x2
U' F R U R' U' F' U L// CP+ 1x1x3

2) L U R D R U2 B R2 L' F R' U' F R2 L F U D2
x2 z// inspection
M// 1x2x2
y U2 L' U L D' L' U L D'// CP+1x1x3

3) R B R2 L' U' B2 F2 U' D F' D2 L2 F U2 B' L2 R F

z'// inspection
M// 1x1x2
y' R U' R' D R U R'// CP+1x1x3

 

[Phaint]

I'm interested in learning this method and have tried going though all your example solves. I really don't fully understand some of the steps and get lost in the explanations with all the shorthand I'm not used to seeing. Do you have a way to make a video of a solve with you explaining your thought process? If you aren't great with notation I would think that would be easier. Just me though, I do better with audio/video learning than text.

 

[shadowslice e]

I'm interested in learning this method and have tried going though all your example solves. I really don't fully understand some of the steps and get lost in the explanations with all the shorthand I'm not used to seeing. Do you have a way to make a video of a solve with you explaining your thought process? If you aren't great with notation I would think that would be easier. Just me though, I do better with audio/video learning than text.

I will make a video as soon as I can but no guarantees as to when this will be as I'm pretty busy with college.

 

[shadowslice e]

I finally posted a video with the correct link. Also there's a link in the OP.

http://youtu.be/fXnkFs_v6Qo

 

[shadowslice e]

85 for 2GLL and how many for the 3-cycles?
Also, I was practicing what I proposed in the thread (phasing before 2GLL) and was starting to get sub-25 averages with it. I think that sub-20 with this is certainly possible.

I finally got around to answering this question and concluded that you need 108 algs to solve any case optimally but realistically you only need 21 algs including reflections and inverses (or 3 algs if we remove them(!)) with a 1 move setup (2 sets of 3 cycles, and their inverses plus 1 2 swap times 3 orientations) or 9 with a 2 move setup. In addition, most of these would be very easy to learn as they would be pretty intuitive. Think of them as sort of like BH algs with a DFL buffer.

 

[shadowslice e]

Is this a double post? It's been 4 days so I don't think it qualifies anymore.

It might be a good idea to change which side you build the initial FB on based on which side has the LF and LB edges as you would be able to solve them using only L l and U without destroying CP if you keep it pseudo 2 gen before moving on to the next step.

Perhaps an example would be clearer: scramble a cube using only R, r, U and u moves. Now scramble using only L l U and u moves. Now solve FB using L, l and U moves, do Eo and now finish F2L with R and U moves. You now have a 2gll case which shows that as long as the LDF and LDB corners are disturbed using a (pseudo)2-gen scramble-ie LU, and are replaced still using only those moves, CP is preserved.

 

[wir3sandfir3s]

You were right, these methods are complicated... Still gonna attempt a video on them.

 

[Micki]

first 6 solves i did on cam with this method

 

[shadowslice e]

first 6 solves i did on cam with this method

Wow nice! I am impressed.

Would you say this a reasonably representative average with the method for you?

 

[Arc]

first 6 solves i did on cam with this method

This is awesome. Briggs is a ton of fun for me and I think it has a lot of potential, so I'm actually using it as my main method (I'm still really slow) as of a few days ago. I'm using the CPFB variant though, I think it's feasible (but difficult) to plan it in inspection and it cuts down on a few moves and removes <u> from the solve.

One thing I'm still really exploring atm is different ways of solving EO, if you'd like to share anything you found that was interesting I'd appreciate it.

Also it's nice to see someone else solving in blue/white.

 

[Micki]

Wow nice! I am impressed.

Would you say this a reasonably representative average with the method for you?

yes atm, but since i learned CP 2 days ago and my block extensions + eo is kinda ineffecient right now it's definitely not a good average considering what you can get with this method. my only good steps are everything after 2x2x3+cp+eo since i'm sub 10 with CFOP and know 2GLL. I'm also pretty good at the first step but that's easy to get below 7 moves i'm really liking solving with CP first methods, it makes you feel really advanced xD

 

[Micki]

This is awesome. Briggs is a ton of fun for me and I think it has a lot of potential, so I'm actually using it as my main method (I'm still really slow) as of a few days ago. I'm using the CPFB variant though, I think it's feasible (but difficult) to plan it in inspection and it cuts down on a few moves and removes <u> from the solve.

One thing I'm still really exploring atm is different ways of solving EO, if you'd like to share anything you found that was interesting I'd appreciate it.

Also it's nice to see someone else solving in blue/white.

i'm also exloring EO, I haven't found a good way yet to do it without having a bunch of pauses. But the way i'm doing it is pretty straightfoward. I solve my DF/DB edges while orienting as many pieces as i can, and that will always leave me with either 0 or 2 edges where i can just do r (R', R, R2) r' to orient the rest. I dont think this way is a bad way of doing it, the reason for me having a lot of pauses is just because i'm not used to orienting edges that way yet.