Thoughts about the BH method

[mazei]

Perhaps I haven't used BH enough to actually get it to a more 'braindead' situation like Fridrich. Then again, the only commutator I've been using consistently is A-perm but I don't even look at it in a commutator way. So I definitely need to practice commutators more.

Solves with BH corners(just corners) for me have been ranging from 1:30 to about 4 minutes. This includes memo which I just do visually(tapping).

 

[cmhardw]

I know how to do it (although I'm slow at it), but it seems to me like I should be able to see the algorithm instantly without a viewpoint shift for every case in BH corners anyway, so I don't really see the point in practicing viewpoint shifting.

Yes I suppose that ideally you should be able to identify and see each case without viewpoint shifting. I view it as a compromise between efficiency and speed. Ideally I should be able to solve the Rubik's cube in sub-35 moves every time, however I find it to be faster to solve in 55 moves with less efficiency. I treat viewpoint shifting the same way. Sure I should be able to identify each case from the angle given, but I find it to be faster to just viewpoint shift the difficult cases.

Here is an example of a case that I can recognize from the angle by itself, but even still I have a difficult time picturing the commutator. I find it to not only be faster, but much easier, to always viewpoint shift on this case:

UBL -> BUR -> FUL

When I see this case, especially when going quickly, I am tempted to think it's a Cyclic Shift, but it's actually an A9. If you viewpoint shift every sticker once clockwise you can see that it's:

LUB -> RBU -> LFU
which is easily done as: R B2 R F2 R' B2 R F2 R2

I am learning to recognize more cases from the actual angle without viewpoint shifting, and maybe it is a crutch that I should stop using, but I find it to speed up my recognition. Perhaps this is only the beginner version, and I should work up to identifying all cases from the original angle. I never thought about that.

--edit--

Then again, the only commutator I've been using consistently is A-perm but I don't even look at it in a commutator way.

To get used to seeing the A perm as a commutator try solving cycles like:
UBL -> BUR -> BRD
UBL -> BRD -> BUR
UBL -> FRU -> FDR
UBL -> FDR -> FRU

They are essentially variations on the A perm.

Chris

 

[Mike Hughey]

Here is an example of a case that I can recognize from the angle by itself, but even still I have a difficult time picturing the commutator. I find it to not only be faster, but much easier, to always viewpoint shift on this case:

UBL -> BUR -> FUL

Funny - I think of that as one of the easier ones to see directly. Although I will admit it's pretty to look at it from the viewpoint shift perspective too - I never looked at that one that way before, and it's certainly pretty.

It just goes to show you that different people see things differently. Certainly viewpoint shifting is a very useful tool when you're first learning BH.

 

[byu]

Here is an example of a case that I can recognize from the angle by itself, but even still I have a difficult time picturing the commutator. I find it to not only be faster, but much easier, to always viewpoint shift on this case:

UBL -> BUR -> FUL

Funny - I think of that as one of the easier ones to see directly. Although I will admit it's pretty to look at it from the viewpoint shift perspective too - I never looked at that one that way before, and it's certainly pretty.

It just goes to show you that different people see things differently. Certainly viewpoint shifting is a very useful tool when you're first learning BH.

Looking at UBL -> BUR -> FUL

I am having a more difficult time seeing the commutator here than I normally would probably due to the fact that I usually use URB as my buffer instead of UBL. But just by rotating the cube, it seems to me that the optimal algorithm would be:

R B2 R F2 R' B2 R F2 R2

Which I would actually execute as:

x' R U2 R D2 R' U2 R D2 R2 x

Which is an A9 case, a 9 move optimal solution, unless I am much mistaken.

However, I myself find it easier to perform most A9s as 10 move commutators (8 move, one setup, no cancellation). Though slower, processing for me requires less thinking, and therefore is just slightly faster.

I'm hoping that eventually I'll be able to execute A9s better.

 

[Mike Hughey]

Here is an example of a case that I can recognize from the angle by itself, but even still I have a difficult time picturing the commutator. I find it to not only be faster, but much easier, to always viewpoint shift on this case:

UBL -> BUR -> FUL

Funny - I think of that as one of the easier ones to see directly. Although I will admit it's pretty to look at it from the viewpoint shift perspective too - I never looked at that one that way before, and it's certainly pretty.

It just goes to show you that different people see things differently. Certainly viewpoint shifting is a very useful tool when you're first learning BH.

Looking at UBL -> BUR -> FUL...

Which I would actually execute as:

x' R U2 R D2 R' U2 R D2 R2 x

Which is an A9 case, a 9 move optimal solution, unless I am much mistaken.

That's how I do it, yes. I really like that one now that this discussion happened - it's one of the quickest for me to see now because we talked about it.

However, I myself find it easier to perform most A9s as 10 move commutators (8 move, one setup, no cancellation). Though slower, processing for me requires less thinking, and therefore is just slightly faster.

I'm hoping that eventually I'll be able to execute A9s better.

I have decided to never abandon an optimal commutator if it is because it requires less thinking; I figure I can always overcome that by drilling it some more so that's no longer the case. For cases where I can execute faster, though, I've abandoned some cases. For instance, I go ahead and execute the following case:
UBL -> DFR -> UFR
as
y' (R2 D R2 D' R2 U2) * 2 y
simply because I'm actually reasonably fast at that algorithm. I can see the 11-mover, but I just can't execute it as fast.

Chris, I tried to learn to do the 11-mover for it - I used it for about 4 months - but I finally gave up on it because it was so noticeably slower than this, and it just wasn't improving enough with practice. Maybe I can try again someday when I'm a faster speedcuber.

 

[cmhardw]

For cases where I can execute faster, though, I've abandoned some cases. For instance, I go ahead and execute the following case:
UBL -> DFR -> UFR
as
y' (R2 D R2 D' R2 U2) * 2 y
simply because I'm actually reasonably fast at that algorithm. I can see the 11-mover, but I just can't execute it as fast.

Chris, I tried to learn to do the 11-mover for it - I used it for about 4 months - but I finally gave up on it because it was so noticeably slower than this, and it just wasn't improving enough with practice. Maybe I can try again someday when I'm a faster speedcuber.

Were you using L' U' R2 U' L U R2 U' L' U2 L?

You might also like:
R' F' R2 F' L F R2 F' L' F2 R which can be executed as: x R' U' R2 U' L U R2 U' L' U2 R x' (which incidentally I might switch to, because it is very fast!)

There's also the "Caltech style" of:
(R' F R F')*3 U2 (R' F R F')*3 U2

I do sometimes use sub-optimal algs if they are faster as well. For example the case:
UBL -> BDL -> URB

You can do this optimally with a Drop and Catch using U' B D B' U B D' B'

However I find it much easier/faster to use a B9 alg:
U R' D2 R U R' D2 R U2

To each his/her own I say. The BH guidelines is just a philosophy to guide you toward using optimal commutators if they work best. I guess the most fundamental idea behind BH is really "Use braindead, pre-decided algs for all possible 3 cycles of two stickers, starting from a fixed buffer." This way it leaves a lot of wiggle room for personalizing the method.

Chris

 

[Mike Hughey]

For cases where I can execute faster, though, I've abandoned some cases. For instance, I go ahead and execute the following case:
UBL -> DFR -> UFR
as
y' (R2 D R2 D' R2 U2) * 2 y
simply because I'm actually reasonably fast at that algorithm. I can see the 11-mover, but I just can't execute it as fast.

Chris, I tried to learn to do the 11-mover for it - I used it for about 4 months - but I finally gave up on it because it was so noticeably slower than this, and it just wasn't improving enough with practice. Maybe I can try again someday when I'm a faster speedcuber.

Were you using L' U' R2 U' L U R2 U' L' U2 L?

Actually, no, I was using y' L U R2 U L' U' R2 U L U2 L' y.
Yours is nice in that it doesn't have the cube rotation (which really doesn't slow it down much, but maybe a little). Maybe I'll try that if I ever try changing again.

You might also like:
R' F' R2 F' L F R2 F' L' F2 R

There's also the "Caltech style" of:
(R' F R F')*3 U2 (R' F R F')*3 U2

I must admit I don't like either of those. But they are interesting alternatives.

I do sometimes use sub-optimal algs if they are faster as well. For example the case:
UBL -> BDL -> URB

You can do this optimally with a Drop and Catch using U' B D B' U B D' B'

I do it this way. I'm finding it can get faster with practice...

However I find it much easier/faster to use a B9 alg:
U R' D2 R U R' D2 R U2

But this is pretty amazingly better. I think I may have to switch.

To each his/her own I say. The BH guidelines is just a philosophy to guide you toward using optimal commutators if they work best. I guess the most fundamental idea behind BH is really "Use braindead, pre-decided algs for all possible 3 cycles of two stickers, starting from a fixed buffer." This way it leaves a lot of wiggle room for personalizing the method.

I like this definition for BH. I've been using this to slowly customize M2 for edges until it's almost like BH. Basically, I just treat each pair of M2 single-piece algorithms as if it were a BH edges "candidate", and then optimize it if I don't think it's fast enough. Right now I probably only use pure M2 for about 1/3 of the edge pair cases. Probably 30% of my edge cases are now using optimal algorithms for a given pair, and most of the rest are fairly fast.

 

[deadalnix]

I just dicover this alg which is pretty nice for BH :

uR2D'R2'u'R2uR2'DR2u'R2'

It look ugly, but it's very finger friendly

 

[cmhardw]

I just dicover this alg which is pretty nice for BH :

uR2D'R2'u'R2uR2'DR2u'R2'

It look ugly, but it's very finger friendly

I posted this same alg in another thread, but I think I must have misunderstood the case people were looking for.

For the case you posted I think your alg is good, but I really think Daniel Beyer's alg cannot be beaten in terms of speed and ease of execution. It is so incredibly fast, and with some practice easy to visualize during a blindfolded solve. It would take quite an amazing alg for me to switch to something different:

U l2 U' l2 U' R2 U l2 U l2 U' R2

Chris

P.S. Hey Amaury!

 

[deadalnix]

oO

I really don't see how this alg can be fast. I have tried several thing, but none is fast. Can you describe how to do it ?

 

[cmhardw]

oO

I really don't see how this alg can be fast. I have tried several thing, but none is fast. Can you describe how to do it ?

Well, when I say fast I mean faster than all the others algs I have tried. On my best cube I probably average around 3.5 seconds to execute this alg, and I can get singles under 3 seconds.

I execute it like this:
U ^ (l)2 (U')* (l')2 (U')** R2 U ^ (l)2 U (l')2 (U')** R2

^ high grip (left thumb on U face, left fingers on D face)
*pull with left ring finger
** left index finger

The alternating directions of the double l layer, and the high grips before starting them is what helps speed up the alg a lot.

Chris

 

[trying-to-speedcube...]

I still don't understand what that alg is good for... :confused:

 

[deadalnix]

This is very nice !

Adopted

The reverse is nice also.

 

[cmhardw]

I still don't understand what that alg is good for... :confused:

Maarten, it does the corner sticker cycle:
URB -> DRF -> DLB

And it's amazingly fast (comparatively)

Chris

 

[trying-to-speedcube...]

I end up doing a corner 5-cycle and an edge 3-cycle with that alg. Is...

Oh, wait. I was still thinking bigcubes. Sorry It's an awesome alg indeed

 

[MatsBergsten]

I end up doing a corner 5-cycle and an edge 3-cycle with that alg. Is...

Oh, wait. I was still thinking bigcubes. Sorry It's an awesome alg indeed

@Maarten: Ha, I did like you, took a 4x4 and executed the alg to see what it did. Ended up with nothing nice at all

Then I read your post and understood too

@Chris: This is one of the few cases where I have not used the BH-corner-alg but a 13-mover of my own (which I thought was easier to turn and remember). But this is way better, I agree.

 

[cubeninjaIV]

i gave up BLD a looong time ago (got irritating and time consuming.)

 

[Mike Hughey]

i gave up BLD a looong time ago (got irritating and time consuming.)

I gave up 3x3x3 speedcubing a while ago too (got irritating and time consuming).

Then I discovered that getting better at 3x3x3 speedcubing actually helps my BLD results more than just about anything else I can do. So now I am pretty heavily practicing 3x3x3 speedcubing. But it's still irritating and time consuming. I just do it because it helps me get better at BLD, which is fun.

Just goes to show you that different people value things differently, I guess.

 

[Edmund]

I wish I could solve blind.

 

[deadalnix]

I wish I could solve blind.

So work your blind instead of posting unrevelant things into a BH related topic.