• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 40,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

Thoughts about the BH method

L

lucarubik

Guest
just watched a 7 min video called how to floating buffers (advanced) and orientation and even number of pieces cycles are still a problem, as it has always been to me to even consider this. now i wouldnt normally say that a method being hard makes a method bad, but i don't think you can get used to the ambiguities of using two buffers, never as good as using one, not in a lifte time. having to focus on weather if the cycles have a closed orientation or not, etc, mainly etc, so im gonna go ahead and say im not a fan, since there are no actual game changers. of course if i see a U perm, or an H or a Z perm ill do them, i think everyone does that, and when i do it its barely better. while it might be "good" to really work on it its definitely not "worth it", same for 5 cycles, i'd call a step forward a way to learn all 5 cycles easily or a way to use floating buffers consistantly, even when one of the two problems appear. this reminds me of swapping pieces during memo when parity; a dirty, overated solution. I give value to having to worry about less things. i could almost meassure it in time.
if someone has actually worked on it and calculated when its worht it to go for it and by how much, and by how much is better to be willing to go for it and these things im not so sure about i'd love to read about it ofc
 

Mike Hughey

Administrator
Staff member
Joined
Jun 7, 2007
Messages
11,305
Location
Indianapolis
WCA
2007HUGH01
SS Competition Results
YouTube
Visit Channel
I've been under the impression that most of the top BLD solvers in the world now do in fact use floating buffers. I assumed that it was just true that enough practice made it possible to be totally reliable and good at that. But if there are top BLD solvers who don't use floating buffers, I'd love to know about it! Maybe it would be a useful thing to survey them.

I know I'm very slow, even with big cubes, so my opinions aren't worth much. But I'll offer my experience anyway: I used floating buffers for center pieces on big cubes BLD for many years. I liked the way that sometimes using floating buffers would give me the chance to cut a few commutators off my solve on a lucky scramble, compared to those who didn't use floating buffers, and I'm convinced I won a few competitions that way. :) But a couple of years ago, when I came back to cubing after having taken a break for a couple of years, I decided it might be a good idea to try not using the floating buffers to see how it went. And I did find that I seemed to be faster and more reliable without them. Even though it was often true I would get to skip a commutator or two on a big cube (sometimes three or four or even more on 5BLD, and even more on really big cubes BLD), the memorization was a little trickier, and solving was also a little trickier. By always sticking with fixed buffers, everything became more braindead, and for me at my mediocre level, it really improved my speed and also my accuracy.

But I still think it's feasible that floating buffers might be a real benefit for people who are world-class at BLD. I am curious if it's true.
 

mark49152

Premium Member
Joined
Oct 29, 2012
Messages
4,719
Location
UK
WCA
2015RIVE05
YouTube
Visit Channel
I think simple floating is quite easy. For example, when there is a single corner 3-cycle involving one U face sticker and two D layer stickers. It saves a whole comm just to set up the U face sticker to the buffer position, and there's no side effect to the setup to think about.

Similarly, a 2-cycle can sometimes be done as a 2c2e PLL, saving a comm.

Of course, the challenge is to think fast enough to execute these shortcut solutions without wasting more time than you save :).
 

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,713
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
Thanks for mentioning the point of floating buffers , and comparing it to a fully floating buffer solve to a 5 style solve.

Even with floating , 5 cycles will maintain ~20-35% more move efficiency over 3-style advanced.

I will be doing an exhaustive statistical comparison of full floating vs 5-cycle , on the move count , and update in the doc.
 
  • Like
Reactions: h2f

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,713
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
I read your document quite quickly as I was short of time, but I didn't see any ideas on breaking down 5-style and making it more manageable. That's essential otherwise it just remains an unrealistic pipedream. The doc just seems to describe how hard it is and illustrate the theoretical benefits should anyone ever actually learn it, but that's not particularly helpful to learning it. Did I miss something?

If you give it a quick thought , it does seem like an unrealistic pipedream , and that is the reason why I procrastinated for nearly 3 years before starting out on learning 5-cycles.

The doc had more content about just plain comparison , as I just wanted to address all kinds of scepticism that people might have when they hear about 5 cycles for the first time.
I do have a learning mechanism to get the ~130000 edge algs and ~70000 corner algs , into our memory , and most of the algs into muscle memory , but for that learning to be effective enough for any level of blindfold solver, I will have to develop the learning method more.
 
  • Like
Reactions: h2f

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,713
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
I've been under the impression that most of the top BLD solvers in the world now do in fact use floating buffers. I assumed that it was just true that enough practice made it possible to be totally reliable and good at that. But if there are top BLD solvers who don't use floating buffers, I'd love to know about it! Maybe it would be a useful thing to survey them.

I know I'm very slow, even with big cubes, so my opinions aren't worth much. But I'll offer my experience anyway: I used floating buffers for center pieces on big cubes BLD for many years. I liked the way that sometimes using floating buffers would give me the chance to cut a few commutators off my solve on a lucky scramble, compared to those who didn't use floating buffers, and I'm convinced I won a few competitions that way. :) But a couple of years ago, when I came back to cubing after having taken a break for a couple of years, I decided it might be a good idea to try not using the floating buffers to see how it went. And I did find that I seemed to be faster and more reliable without them. Even though it was often true I would get to skip a commutator or two on a big cube (sometimes three or four or even more on 5BLD, and even more on really big cubes BLD), the memorization was a little trickier, and solving was also a little trickier. By always sticking with fixed buffers, everything became more braindead, and for me at my mediocre level, it really improved my speed and also my accuracy.

But I still think it's feasible that floating buffers might be a real benefit for people who are world-class at BLD. I am curious if it's true.

Floating buffers is really beneficial in improving the blindsolving technique , but it is really scramble dependent on how much , floating buffer cycles you are able to extract. Floating buffers is quite useful , in case of hard scrambles ,where all the pieces are stuck in a 2-cycle or a 4-cycle , and need to cycle breaks (an extra comm) to get solved.

Floating buffers is the way to go , but it is not the roof of the complexity we can develop for blindfold solving.
Eventually , we can start improving on 3-cycles and scale them up.
 
  • Like
Reactions: h2f

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,713
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
not sure what you mean by the “to compare with 5style” part but there are a few places you can find walkthroughs/recons with floating. Off the top of my head I can think of some of Jake klassens walkthroughs, Josh Weimer has a vid, and then just going on Facebook in the bld support group lots of people can help you with stuff there :)

I do follow Jake cool walkthrough solves , and they are very helpful.
I am already a proponent of the 3-style , and the modern 3-style development.

It is just that I think the plunge into 5 cycles is worth considering , even if we have many advanced techniques for 3-style like parity avoidance, flipped edge+parity , parity +last comm , floating , CEEC/ECCE neutrality , ZBLL for parity , 2e2e and many more advanced techniques.
 
L

lucarubik

Guest
maybe a reduced 5 style, with a very efficent combination of way less algs would be interesting? maybe?
 
L

lucarubik

Guest
It looks like few blinders can.
ill try it for myself, maybe its better in practice than it sounds in theory
i would have to learn cycles wtih all pieces as buffer but two, right? two of my choice, the two last would be 100% overlap
i guess ill test it for edges, since they are the most likely to be worth it on, by a lot, wich is a shame cause i already know all the 63x8 corner cases
 

Mike Hughey

Administrator
Staff member
Joined
Jun 7, 2007
Messages
11,305
Location
Indianapolis
WCA
2007HUGH01
SS Competition Results
YouTube
Visit Channel
A variant of total floating buffer that would only be minimally slower and would require no extra algorithm memorization would be to simply reorient the cube to put the new buffer where your current one normally is and then execute the appropriate algorithms for it there as you're using the new buffer. It probably depends on how you visualize the pieces as to how difficult it might be to see where the other pieces wind up when you do this. I think it would confuse me, but I get the feeling other people don't find it that difficult.
 
L

lucarubik

Guest
Not necessarily. Simple floating is much easier. I gave an example above.
even if that set ups end up cancelling and its literally te same, i rather think about it as completely different buffer
A variant of total floating buffer that would only be minimally slower and would require no extra algorithm memorization would be to simply reorient the cube to put the new buffer where your current one normally is and then execute the appropriate algorithms for it there as you're using the new buffer. It probably depends on how you visualize the pieces as to how difficult it might be to see where the other pieces wind up when you do this. I think it would confuse me, but I get the feeling other people don't find it that difficult.
yes i didnt even think abouit it, there is a lot of overlap, a lot of symetry, i dont think i need to do any of that, just actual floating buffer
 
L

lucarubik

Guest
I'm not an expert in this but yes - it needs learning algs for few buffers. But few of top blinders changed they buffer from DF to UF and UBL to UFR so it makes it easier.
tell me about it lol
 

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,713
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
At 10 algs per day that'll take you about 60 years, by which time your brain won't even recall your name and your fingers will have dropped off. Good luck!

I do not think extrapolating linearly works while learning algorithms. Otherwise it is just insane amount of linear time.
We always have computer and computing power at our disposal , and our brain increases its complexity with each passing day , maybe exponentially for some people.

60 years sounds too pessimistic.

If I assume a fully driven person , it should take him/her , 4 years to get complete 5 cycle into their mind and fingers.
If the person is just doing blindfold solving as a hobby , maybe around 10 years to completely exhaust the set and drill
If the person is just doing speedcubing as a hobby , then your calculations match up , ie , 10-60 years , and lot of best wishes for that crazy person.
 

mark49152

Premium Member
Joined
Oct 29, 2012
Messages
4,719
Location
UK
WCA
2015RIVE05
YouTube
Visit Channel
60 years sounds too pessimistic.

If I assume a fully driven person , it should take him/her , 4 years to get complete 5 cycle into their mind and fingers.
If the person is just doing blindfold solving as a hobby , maybe around 10 years to completely exhaust the set and drill
If the person is just doing speedcubing as a hobby , then your calculations match up , ie , 10-60 years , and lot of best wishes for that crazy person.
4 years = 1461 days. Let's say 12 hours per day. 200,000 algs comes out at roughly one new alg every 5 minutes. You did say "driven" so I've assumed no breaks and no other life.

Of course learning isn't linear, but an average rate is still a reasonable way to show how crazy it is. Maybe you'll learn algs faster as you go, but you'll also have to spend more time reviewing what you already learned.

I won't waste time on the 10 year scenario as it's not much better. I think maybe you don't grasp the numbers and the scale of the task. You're welcome to prove me wrong though and I look forward to seeing you back here in 4 years! :)
 

abunickabhi

Member
Joined
Jan 9, 2014
Messages
6,713
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
4 years = 1461 days. Let's say 12 hours per day. 200,000 algs comes out at roughly one new alg every 5 minutes. You did say "driven" so I've assumed no breaks and no other life.

Of course learning isn't linear, but an average rate is still a reasonable way to show how crazy it is. Maybe you'll learn algs faster as you go, but you'll also have to spend more time reviewing what you already learned.

I won't waste time on the 10 year scenario as it's not much better. I think maybe you don't grasp the numbers and the scale of the task. You're welcome to prove me wrong though and I look forward to seeing you back here in 4 years! :)

Hmmm , this 10 algs a day batch approach never works,

if suppose there are 20 types of commutators in 5 cycle on contrary to 5 in 3 cycle (pure comm, A9 , cycle shift , columns , per special) , then just understand the alg and its various setup would be crucial . The task would be to accurately retrieve the each alg , and the type of comm+setup we have mapped it into.

I do not expect everyone to start learning this up, this is still in its baby phase, maybe in few years I can make it modular enough for every calibre person ,to grasp appropriate parts of the whole system of 5cycle.
 
Top