An analysis of cubing methods and times...

[Artic]

A few thoughts I had the other day:

Scenario 1: Two users have the same average of 100. Their time breakdowns are also the same. In other words, their times for cross, F2L and LL are also the same. The only difference is that one is opposite color neutral while the other cuber is completely color neutral.

My analysis: The cuber who is only opposite color neutral is better Why? Because despite his limitation to two cross colors, he is still able to post equal times compared with the other cuber. The cuber who is completely color neutral has a much larger range of easier crosses to make. The disadvantaged cuber has more difficult cross cases, yet he is still able to compensate for this and post equal cross times.

Scenario 2: Two users have the same average of 100. One cuber uses Roux and the other uses CFOP.
My analysis: The CFOP cuber is better. Why? Because Roux is known to have a much lower movecount. And yet despite his move count advantage, the Roux cuber can still only post times equal to his CFOP counterpart.

So for example, Bill Wang should be considered a better cuber than Alex Lau, at least from their official times list. The interesting aspect is that both Bill and Alex had extremely high TPS, and yet, even with this, they still have approximately equal official times. Clearly, there is something difficult or inefficient with Roux, something that limits its efficacy. I suspect the use of M-slices is a permanent impediment and obstacle that offsets their lower movecount advantage. This will forever hinder Roux in my opinion, since it's so M-slice dependent.

Any thoughts on this?

 

[tseitsei]

A few thoughts I had the other day:

Scenario 1: Two users have the same average of 100. Their time breakdowns are also the same. In other words, their times for cross, F2L and LL are also the same. The only difference is that one is opposite color neutral while the other cuber is completely color neutral.

My analysis: The cuber who is only opposite color neutral is better Why? Because despite his limitation to two cross colors, he is still able to post equal times compared with the other cuber. The cuber who is completely color neutral has a much larger range of easier crosses to make. The disadvantaged cuber has more difficult cross cases, yet he is still able to compensate for this and post equal cross times.

Scenario 2: Two users have the same average of 100. One cuber uses Roux and the other uses CFOP.
My analysis: The CFOP cuber is better. Why? Because Roux is known to have a much lower movecount. And yet despite his move count advantage, the Roux cuber can still only post times equal to his CFOP counterpart.

So for example, Bill Wang should be considered a better cuber than Alex Lau, at least from their official times list. The interesting aspect is that both Bill and Alex had extremely high TPS, and yet, even with this, they still have approximately equal official times. Clearly, there is something difficult or inefficient with Roux, something that limits its efficacy. I suspect the use of M-slices is a permanent impediment and obstacle that offsets their lower movecount advantage. This will forever hinder Roux in my opinion, since it's so M-slice dependent.

Any thoughts on this?

You define "better cuber" in a very weird way... Better cuber for me means the one who can complete the event in question (here 3x3 speedsolve) faster. I don't know why you want to overcomplicate this and I also don't think you analysis of those 2 scenarios makes any sense at all because obviously if they have same times they are just as good...

They just are better at different things.

My analysis: The cuber who is only opposite color neutral is better Why? Because despite his limitation to two cross colors, he is still able to post equal times compared with the other cuber.

Well this could as well be said like:
My analysis: The cuber who is full color neutral is better Why? Because even if he isn't as good cross-planner than the other guy he can still post equal times because he has the ability to pick the easier cross most of the time. He has a disadvantage because he doesn't plan as optimal crosses as the other guy but he is still able to compensate for this by choosing the easiest color cross every time.

And for scenario 2 this:

My analysis: The CFOP cuber is better. Why? Because Roux is known to have a much lower movecount. And yet despite his move count advantage, the Roux cuber can still only post times equal to his CFOP counterpart.

Could as well be this:
My analysis: The roux cuber is better. Why? Because Roux has harder lookahead than CFOP (you need to always know what piece is in the DB slot since you can't see it) and roux uses M-moves that are not necessarily as fast as RUF-moves mostly used in CFOP. And yet despite these disadvantages, the Roux cuber can still post times equal to his CFOP counterpart.


See what I did there?

 

[IRNjuggle28]

A few thoughts I had the other day:

Scenario 1: Two users have the same average of 100. Their time breakdowns are also the same. In other words, their times for cross, F2L and LL are also the same. The only difference is that one is opposite color neutral while the other cuber is completely color neutral.

My analysis: The cuber who is only opposite color neutral is better Why? Because despite his limitation to two cross colors, he is still able to post equal times compared with the other cuber. The cuber who is completely color neutral has a much larger range of easier crosses to make. The disadvantaged cuber has more difficult cross cases, yet he is still able to compensate for this and post equal cross times.

I disagree. I think they are the same. I think being better or worse is identical to being faster or slower.

Yes, the opposite neutral solver has fewer crosses to choose from, but he has easier recognition due to always having the same two colors on U and D. But even if I hadn't made that observation, I'd still think all the evidence points to neither of them being "better," because being good or bad is synonymous to being fast or slow, in this context.

If anything, I'd say that the CN guy is better because he has more potential to be better in the future, simply because he has a higher ceiling for how fast he can get. If you assume that both cubers can improve their recognition to the point where they almost never pause, the CN guy will be faster because of the lower movecount. There are no bounds to how much skill can improve, but only having 2 colors to choose from instead of all 6 is a limiting factor that no amount of practice can counterbalance.

I don't think the assessments you're making can't be useful, I just think you need more specific and less subjective descriptors than "better" and "worse" for it to work.

Scenario 2: Two users have the same average of 100. One cuber uses Roux and the other uses CFOP.
My analysis: The CFOP cuber is better. Why? Because Roux is known to have a much lower movecount. And yet despite his move count advantage, the Roux cuber can still only post times equal to his CFOP counterpart.

My analysis: The Roux solver is better. Why? Because Roux is known to have much more difficult recognition. And yet despite the advantage that a method with easier recognition gives, the CFOP cuber can only post times equal to his Roux counterpart.

See?

You seem to cite movecount as a thing that make a method or a cuber better or worse, but not recognition. I think that's a mistake. I can understand why; movecount is much more measurable than recognition is, and therefore easier to make comparisons about. That doesn't make it more deterministic of the skill of a cuber or the efficiency of a method, though.

As I did with the previous scenario, I would argue that the Roux solver has more potential to be faster in the future, as lowering the movecount of a method is less doable than improving your lookahead.

So for example, Bill Wang should be considered a better cuber than Alex Lau, at least from their official times list. The interesting aspect is that both Bill and Alex had extremely high TPS, and yet, even with this, they still have approximately equal official times. Clearly, there is something difficult or inefficient with Roux, something that limits its efficacy. I suspect the use of M-slices is a permanent impediment and obstacle that offsets their lower movecount advantage. This will forever hinder Roux in my opinion, since it's so M-slice dependent.

I'm pretty sure Bill has higher TPS than Alex at the times they're equal speeds.

I had thought the LSE was actually the best step in the method. So many pieces are solved 2gen, and M slices are really nice to fingertrick. Can you explain more why you think of it as an impediment?

EDIT: tseitsei said similar things... partially ninja'd

 

[rjcaste]

A few thoughts I had the other day:

Scenario 1: Two users have the same average of 100. Their time breakdowns are also the same. In other words, their times for cross, F2L and LL are also the same. The only difference is that one is opposite color neutral while the other cuber is completely color neutral.

My analysis: The cuber who is only opposite color neutral is better Why? Because despite his limitation to two cross colors, he is still able to post equal times compared with the other cuber. The cuber who is completely color neutral has a much larger range of easier crosses to make. The disadvantaged cuber has more difficult cross cases, yet he is still able to compensate for this and post equal cross times.

Scenario 2: Two users have the same average of 100. One cuber uses Roux and the other uses CFOP.
My analysis: The CFOP cuber is better. Why? Because Roux is known to have a much lower movecount. And yet despite his move count advantage, the Roux cuber can still only post times equal to his CFOP counterpart.

So for example, Bill Wang should be considered a better cuber than Alex Lau, at least from their official times list. The interesting aspect is that both Bill and Alex had extremely high TPS, and yet, even with this, they still have approximately equal official times. Clearly, there is something difficult or inefficient with Roux, something that limits its efficacy. I suspect the use of M-slices is a permanent impediment and obstacle that offsets their lower movecount advantage. This will forever hinder Roux in my opinion, since it's so M-slice dependent.

Any thoughts on this?

I roughly understand what you're trying to say. Although two cubers may have equal times, one cuber is "better" than the other because he has the harder method/subset to use or has a disadvantage outside of something they control and still gets equal times. Such as if someone were to be using beginner's method and getting equal times with someone using CFOP. Obviously, the cuber using the beginner's method would have superior skill since it is difficult to get fast with the beginner's method yet he is as fast as a cuber using CFOP.

 

[Petro Leum]

look, it's easy.

how "good" a cuber is equals how "fast" they are.

there are 3 factors that determine your speed

1. Turning speed (includes raw potential turning speed, but also ability to execute certain moves and algorithms properly)
2. Lookahead (includes lookahead midsolve, recognition times and planning out during inspection, with the shared goal of avoiding pauses)
3. Efficiency (includes Movecount and ergonomy; 50 moves can be better than 45 moves, when the moveset of the solution is better)

now, you can either just measure speed of a cuber and determine who is "better" by just checking who is faster

OR

you could compare the specific factors of the cubers, for example:

Cuber A and Cuber B both have an equally fast average of 100.

Cuber A spends 1.2 per solve not turning any layers, Cuber B spends 1.8 seconds per solve. Cuber B has more and/or longer pauses and therefore worse lookahead than cuber A

you can also compare the efficiency of the two Cubers or the turning speed, but you can never compare two different factors.

In your first example, it comes to mind that the completely color neutral cuber has better efficiency, because his cross solutions are shorter. However, he must then have inferior turning speed or lookahead, because although the opposite-neutral cuber has longer cross solutions, he can achieve equal times.

BUT in the end, both cubers are still equally fast!

get the point?


EDIT: if anyone has something to add to my 3 factor system or any additional thoughts, please post them!

 

[MWilson]

...

if anyone has something to add to my 3 factor system or any additional thoughts, please post them!

Maybe focus.

Consider two solvers that are equal in turning speed, look ahead, and efficiency at the same competition. They both make the same mistake at the same point in the same scramble, after the same time has passed in the solve. The one that loses focus will likely start turning faster in an attempt to make up for the mistake, and may make another mistake or at least have worse performance one or more of the other three factors. This results in a worse time than the other solver, who is otherwise identical, making focus another factor in speed.

There could be many other examples than just making mistakes, such as being better at focusing while solving in front of a crowd.

Maybe nerves is a better term.

 

[qqwref]

I agree with your general point, Petro, but I think instead of "Lookahead" we should just call it "Pauses". And then we can simply write:
Solve time = Pauses + Movecount / TPS
You mention ergonomy as part of efficiency, but I'd consider it a tradeoff; using more ergonomic moves is a tradeoff giving you high efficiency at the cost of high tps, whereas using less ergonomic moves is a tradeoff in the other direction.

So yeah, the Roux and Color Neutral solvers are probably more efficient, but to have the same time as the Fridrich and Opposite Color Neutral solvers they must either have less TPS or more pauses.

 

[Smiles]

Maybe OP didnt word things the exact way for people to understand it, but I totally understand what OP means...
and it makes perfect sense too (at least scenario 1 makes perfect sense).

Better cuber in this context doesn't mean faster in times, but just who can do better given the same circumstances of method.
It's almost like better potential cuber.
So if A uses CFOP and B uses a beginner method, and both average 10 seconds, then B is "the better cuber" or the "better potential cuber" because B obviously can perform just as well on a cube given a disadvantage. The theory is that if B switches to CFOP, B will be faster and therefore B is a better cuber. This is all theoretical, obviously everyone has preferences but that is not taken into account here.

I think the logic can be seen in your scenario 2 but I think the reality of it breaks down here. Just because Roux is more move efficient in STM, that doesn't mean it's more efficient in HTM or QTM or SQTM.
These are important metrics too because Roux does have a lot of M2 and U2 moves, which can each be counted as 2 moves (slower). CFOP doesn't have so many.
Also, M moves cannot be performed as fast as algorithms used in CFOP that seem like they're made to be done fast.
Example: H perm vs T perm = 7 STM vs 14 STM yet H perm usually isn't twice as fast.

I guess my point is yes Roux users have a move count advantage (in STM) but methods can't be compared on the basis of just that. Doing 60 moves in CFOP obviously is faster than doing 60 moves in Heise; CFOP isn't designed to be move-efficient, it's just fast cause people have fast fingers and PLL doesn't require look ahead like Roux LSE does.

 

[Artic]

And for scenario 2 this:

Could as well be this:
My analysis: The roux cuber is better. Why? Because Roux has harder lookahead than CFOP (you need to always know what piece is in the DB slot since you can't see it) and roux uses M-moves that are not necessarily as fast as RUF-moves mostly used in CFOP. And yet despite these disadvantages, the Roux cuber can still post times equal to his CFOP counterpart.

This is a completely wrong and irrational statement. I was comparing indisputable known facts between methods, lower move count for Roux vs higher for CFOP, while you are trying to argue using a baselessopinion. Difficulty of lookahead is subjective and cannot be argued one way or another. Many people think CFOP has extremely hard and tricky lookahead. In fact, there's a new thread on F2L lookahead every week! Some people might think Roux had harder lookahead, and many others might say CFOP. That is debatable. But what cannot be disputed is move count. So in that regard, you argument holds no water.

I disagree. I think they are the same. I think being better or worse is identical to being faster or slower.

I don't think the assessments you're making can't be useful, I just think you need more specific and less subjective descriptors than "better" and "worse" for it to work.

My analysis: The Roux solver is better. Why? Because Roux is known to have much more difficult recognition. And yet despite the advantage that a method with easier recognition gives, the CFOP cuber can only post times equal to his Roux counterpart.

You seem to cite movecount as a thing that make a method or a cuber better or worse, but not recognition. I think that's a mistake. I can understand why; movecount is much more measurable than recognition is, and therefore easier to make comparisons about. That doesn't make it more deterministic of the skill of a cuber or the efficiency of a method, though.

As I did with the previous scenario, I would argue that the Roux solver has more potential to be faster in the future, as lowering the movecount of a method is less doable than improving your lookahead.

Again, an argument based on subjective factors and not on facts. F2L lookahead is just as difficult if not more so than Roux. Ask the hundreds of new cubers who post F2L help threads.

I roughly understand what you're trying to say. Although two cubers may have equal times, one cuber is "better" than the other because he has the harder method/subset to use or has a disadvantage outside of something they control and still gets equal times. Such as if someone were to be using beginner's method and getting equal times with someone using CFOP. Obviously, the cuber using the beginner's method would have superior skill since it is difficult to get fast with the beginner's method yet he is as fast as a cuber using CFOP.

Thanks. That's an extreme example, but it does get the point across. I don't think anyone would argue that if a cuber averaged 6 seconds using the beginners method, then he should without a doubt be considered the best cuber.

Maybe OP didnt word things the exact way for people to understand it, but I totally understand what OP means...
and it makes perfect sense too

thanks

 

[IRNjuggle28]

Again, an argument based on subjective factors and not on facts. F2L lookahead is just as difficult if not more so than Roux. Ask the hundreds of new cubers who post F2L help threads.

I can point out some objective facts that make F2B recognition more difficult than F2L. When you solve a cross, you have all centers in the correct place, and can use them to judge where corner/edge pairs belong. But during Roux blocks, the M layer edges and centers are not solved, and I have to memorize my color scheme to know where things belong. Furthermore, edges can only be in the U layer or E layer during F2L. During F2B, edges can be in the D layer, where they are more difficult to track than during CFOP. Both of these things make finding pieces as well as knowing what slot those pieces belong in more difficult.

The other (obvious) objective way of measuring is speed. There are many CFOP solvers that average sub 6 for F2L, and as far as I know, only Alex Lau is that fast at F2B, even though F2B solves less pieces than F2L, and has a lower movecount. All of this seems to point towards Roux having more difficult recognition.

 

[adimare]

You're basically saying that you can determine that cuber A is better than cuber B based solely on their TPS. That is a very subjective statement.

 

[guysensei1]

Does it really matter how efficient someone is?

If Person A averages sub-7 but averages 70 moves, is he really not as good as a person who averages sub-7 and gets 30 moves on average?

 

[TDM]

Does it really matter how efficient someone is?

If Person A averages sub-7 but averages 70 moves, is he really not as good as a person who averages sub-7 and gets 30 moves on average?

Tbh, I would be more impressed by someone who averaged 30 moves in a speedsolve. We've seen people do 10+ TPS in solves before, but if someone averaged 30 moves in a speedsolve, I'd like to see them do FMC

 

[ottozing]

while you are trying to argue using a baselessopinion.

I wouldn't exactly call it baseless, although maybe worse look ahead wasn't the right way to put it. The way I see it, there's very good reason to believe Roux has harder recognition than CFOP does, which results in worse look ahead. Aside from everything IRN said about colour scheme stuff and edge placement, second block solutions aren't really done anywhere near as algorithmically as CFOP pair solutions are (More specifically the first 1x2x2 of the second block). This results in harder look ahead because with CFOP, it's just lookahead, see pair, spam short alg. With Roux it seems like you need to look ahead, come up with a solution on the fly, and while doing moves that aren't 100% ingrained into your muscle memory, do that solution while looking ahead to what you have to solve next. That seems a lot harder to do than spamming F2L pairs, and I can definitely see that being a downfall.

Many people think CFOP has extremely hard and tricky lookahead. In fact, there's a new thread on F2L lookahead every week!

Maybe you should refrain from using baseless opinions yourself. I mean geez that's pretty bad reasoning, considering the fact that the majority of new cubes on this website go for CFOP as their main method instead of Roux, so obviously there will be more of them asking about look ahead since they don't grasp the concept.

 

[tseitsei]

This is a completely wrong and irrational statement. I was comparing indisputable known facts between methods, lower move count for Roux vs higher for CFOP, while you are trying to argue using a baselessopinion. Difficulty of lookahead is subjective and cannot be argued one way or another. Many people think CFOP has extremely hard and tricky lookahead. In fact, there's a new thread on F2L lookahead every week! Some people might think Roux had harder lookahead, and many others might say CFOP. That is debatable. But what cannot be disputed is move count. So in that regard, you argument holds no water.

watwatwat? Baseless opinion?! I even wrote this right after that lookahead part of my post:

(you need to always know what piece is in the DB slot since you can't see it)

Also the edges you need to look for can be in other D-layer slots (DF and DR) while in CFOP during F2L they can only be in U or E layers making them a lot easier to find/track. And on top of that you need to only track 2 pieces to make a CFOP pair but you have to track 3 pieces to make a Roux square (1x2x2 square)...

What you are saying is basically that we can only measure movecount of a method but not recognition/lookahead/some other stuff like that.
Well if that's the case then I suppose Heise method is the most awesome speedsolving method ever
But clearly that is not the case because it's too complicated to be used in speedsolve.
Similarly I would say that CFOP is good speedsolving method because it's so damn simple... Roux is also a good speedsolving method. A bit more complicated than CFOP, but makes up for that with it's lower movecount.

Now please understand that there needs to be more variables involved while comparing methods than just simple movecount.

 

[Petro Leum]

I agree with your general point, Petro, but I think instead of "Lookahead" we should just call it "Pauses". And then we can simply write:
Solve time = Pauses + Movecount / TPS
You mention ergonomy as part of efficiency, but I'd consider it a tradeoff; using more ergonomic moves is a tradeoff giving you high efficiency at the cost of high tps, whereas using less ergonomic moves is a tradeoff in the other direction.

So yeah, the Roux and Color Neutral solvers are probably more efficient, but to have the same time as the Fridrich and Opposite Color Neutral solvers they must either have less TPS or more pauses.

Good point, plus that formula really sums up what my 3 factors are about.

About the ergonomy and Movecount tradeoff; that is also directly connected with my point from the factor Turning speed "being able to execute certain moves properly " - if you are able to execute a lower Movecount solution with a perceived inferior
Moveset equally fast, a more ergonomic higher Movecount solution wouldn't be a tradeoff in your favor.

@mwilson:

I like your idea of focus.
It certainly is important for speedcubing, but I'm not sure how to incorporate it....
It certainly doesn't work as a fourth factor, since it influences the other 3 factors.

 

[Escher]

Looking at cubers recorded official solves doesn't tell you anything about 'better' unless you wish the word to be interchangeable with 'faster'.

Roux is better than CFOP because the approach of the method allows the trade-off for lower movecount so much better than other non-CFOP methods with lower movecounts. For an example, the Roux approach towards L6E is so well designed that most optimal solutions 'use' the method, just a move-cancelled version.

CN is better than single/dual because there is no reason why there should be a trade-off, unless there is something about the brains pattern recognition system we don't fully understand yet. And EVEN IF there is something inhibiting towards CN, it would still be better simply because movecount isn't the only metric that is improved by CN, 'solvability' is far far better once you get to higher levels. One gets so many more awesome options presented to them with 4/5 additional starts to choose from.

Single or dual colour CFOP is mega popular because it builds upon the method pretty much everyone learns, and colour neutrality (or lack thereof) appears to be a set in stone thing from very early on.

We have such little data on Roux and CN that it's somewhat unreasonable to make such claims based on data rather than reasoning, and the reasoning suggests that CNCFOP>CFOP and Roux>CFOP. Sorry I can't write much right now, but I'll return to the thread later on.

 

[sneaklyfox]

Solve time = Pauses + Movecount / TPS

But isn't TPS taken over the whole solve? Even if you can do short bursts of 10 TPS the pauses already factor into TPS for a full solve. (A 10-second solve at 60 moves means 6 TPS.) So unless you have a different way of defining TPS,
Solve time = Movecount / TPS

 

[Bindedsa]

But isn't TPS taken over the whole solve? Even if you can do short bursts of 10 TPS the pauses already factor into TPS for a full solve. (A 10-second solve at 60 moves means 6 TPS.) So unless you have a different way of defining TPS,
Solve time = Movecount / TPS

The point is it's how fast you can turn when you are actually turning. It's not how TPS is generally defined, but in the context of comparing the turning speed of solvers, it's the better metric, IMO.

 

[GuRoux]

I can point out some objective facts that make F2B recognition more difficult than F2L. When you solve a cross, you have all centers in the correct place, and can use them to judge where corner/edge pairs belong. But during Roux blocks, the M layer edges and centers are not solved, and I have to memorize my color scheme to know where things belong. Furthermore, edges can only be in the U layer or E layer during F2L. During F2B, edges can be in the D layer, where they are more difficult to track than during CFOP. Both of these things make finding pieces as well as knowing what slot those pieces belong in more difficult.

The other (obvious) objective way of measuring is speed. There are many CFOP solvers that average sub 6 for F2L, and as far as I know, only Alex Lau is that fast at F2B, even though F2B solves less pieces than F2L, and has a lower movecount. All of this seems to point towards Roux having more difficult recognition.

i average 5 for first two blocks. probably by the time a rouxer gets to sub 12 they will have sub 6 f2b.