Will the Fridrich method ever be beaten?

[somerandomkidmike]

What about a five step method, top layer corners, bottom layer corners (2x2 speedcubing method also can), first layer edges, middle layer edges, and last layer edges?
Totally brute force, no intuition.

Edit: best method ever; solve white face, then red face, then blue, then green, then orange.
Once you solve 5 sides of the cube, you solve the sixth side.

I think that is what beginners think of how to solve a cube.

you are better off with waterman, which I believe only one person is really trying to actively use it here

I'm actively using waterman, at the same time as optimizing it

 

[somerandomkidmike]

If my recognition wasn't so bad for CMLL I'd switch to Roux. It has a good move count and look ahead is very easy because after you build the first 1x2x3 you don't have to do ANY cube rotations. I use at least 3 cube rotations during my Petrus solves which is probably adding a second or two on my times.

recognition for CMLL can be very good

 

[qqwref]

But as far as being fun, I have to give it a big fat zero. I just can't stand using that method.

I wonder if any professional track and field athletes sometimes run a mile in competition using a 'silly walk' instead of running. Sure, it might be more fun, but sometimes to be the fastest you have to just do the event as fast as possible instead of worrying whether you're having fun or not

 

[fanwuq]

But as far as being fun, I have to give it a big fat zero. I just can't stand using that method.

I wonder if any professional track and field athletes sometimes run a mile in competition using a 'silly walk' instead of running. Sure, it might be more fun, but sometimes to be the fastest you have to just do the event as fast as possible instead of worrying whether you're having fun or not

Sorry, QQ, but that's a horrible analogy. Silly walk just isn't energy efficient LOL. Petrus is very move efficient.

I can see Petrus with ZBLL being much faster than Fridrich. However, I suck at it. The reason: I can find awesome 2x2x3 block for FMC, I can't build blocks at all during speedsolves. That is the main disadvantage of Petrus and Roux. If you can build 2x2x3 block and orient all edges as fast as a top Fridrich user doing cross+2 pairs, then you can break the WR using Petrus.
If you can do finish both blocks and CMLL before a fridrich user finishes F2L, then Roux would be faster. The main problem is that time restraints allow for worse move count and the steps aren't as fingertrick friendly as Fridrich.

If I can see the far future, maybe 20+ years, I'd say it would be ZZ EOline with thousands of "algs" for multislotting F2L and ZBLL. That's fingertrick and move efficient.

 

[yurivish]

But as far as being fun, I have to give it a big fat zero. I just can't stand using that method.

I wonder if any professional track and field athletes sometimes run a mile in competition using a 'silly walk' instead of running. Sure, it might be more fun, but sometimes to be the fastest you have to just do the event as fast as possible instead of worrying whether you're having fun or not

Sorry, QQ, but that's a horrible analogy. Silly walk just isn't energy efficient LOL. Petrus is very move efficient.

I can see Petrus with ZBLL being much faster than Fridrich. However, I suck at it. The reason: I can find awesome 2x2x3 block for FMC, I can't build blocks at all during speedsolves. That is the main disadvantage of Petrus and Roux. If you can build 2x2x3 block and orient all edges as fast as a top Fridrich user doing cross+2 pairs, then you can break the WR using Petrus.
If you can do finish both blocks and CMLL before a fridrich user finishes F2L, then Roux would be faster. The main problem is that time restraints allow for worse move count and the steps aren't as fingertrick friendly as Fridrich.

If I can see the far future, maybe 20+ years, I'd say it would be ZZ EOline with thousands of "algs" for multislotting F2L and ZBLL. That's fingertrick and move efficient.

It's move-efficient in the sense that you have lots of algorithms and cases to learn, but I don't think the basic method itself is that move-efficient.

ZZF2L, COLL, EPLL = not that big, if any, of a reduction over Fridrich.

Petrus and Roux are inherently move-efficient in that they build up the cube in such a way that it's possible to retain lots of freedom of movement even when some of the steps are finished.

 

[blade740]

ZZ can be very fast, if someone fast would commit to working on it.

The move-efficiency of ZZ is different from that of petrus and roux. ZZ involves restricting your moves to the easiest moves to perform at top speed. Petrus and roux involve restricting your moves to ones that don't touch what you've already solved.

 

[marco768]

Fingertrick friendliness of Fridrich VS Fewer moves for Petrus and Roux
A classic debate topic

Personally I think Fridrich would remain as the most popular method, while remaining as one of the fastest. One reason I think Fridrich is fast is because there are dedicated solvers who provide all sorts of algs for the method, such as fingertrick friendly alternate algs. People also propose and use optimised steps(X-cross, double-slotting).

But I think ZZ has great potential, restricting moves to L,U,R can increase TPS for both normal and OH solving(Compete with Fridrich's "fingertrick-ness"), along with a "possible" 1LLL. I remember a dedicated ZZ user in the forum provided a link to a pretty deep ZZ explanation and guide, one step involves changing the CP(I forgot whether orientation matters) to [UR] subset while retaining a solved 2x2x3 block, which if someone find a method of it successfully, can lead to a 1LLL.

 

[JLarsen]

Fingertrick friendliness of Fridrich VS Fewer moves for Petrus and Roux
A classic debate topic

Personally I think Fridrich would remain as the most popular method, while remaining as one of the fastest. One reason I think Fridrich is fast is because there are dedicated solvers who provide all sorts of algs for the method, such as fingertrick friendly alternate algs. People also propose and use optimised steps(X-cross, double-slotting).

But I think ZZ has great potential, restricting moves to L,U,R can increase TPS for both normal and OH solving(Compete with Fridrich's "fingertrick-ness"), along with a "possible" 1LLL. I remember a dedicated ZZ user in the forum provided a link to a pretty deep ZZ explanation and guide, one step involves changing the CP(I forgot whether orientation matters) to [UR] subset while retaining a solved 2x2x3 block, which if someone find a method of it successfully, can lead to a 1LLL.

I don't see why everyone thinks Petrus is THAT un finger friendly. Sure you can't just hold the cube in one orientation and spam algs all over the place, but you can make a black with no rotation if your good enough, just in the sense that you can slot the entire f2l without rotating if your good enough. Don't forget that a third of F2l is 2 gen.

 

[ErikJ]

so how fast are you with roux?

20s. I'm not used to the type of block building.

 

[Athefre]

I think that a pure color neutral Roux with NMCMLL would minimize move count. and if someone used something like that at full speed, it'd probably "beat" fridrich. But you need to dedicate time to improve look ahead( A LOT).

this is just a thought. Possible? yes... Practical? not so much...
seeing as roux is my main, i try to think of ways to improve it

I see zero improvement being color neutral with "Roux" (at this time). Honestly.

The normal steps:

1. Solve a 1x2x3
2. Solve opposite 1x2x3 (with same color on D)
3. Corners - Requires two thoughts; 1. Orientation, then 2. Permutation color pattern
4. LSE

Let's set it's advantages/disadvantages to 0 (normal).

Color neutral (I'll call it "CN solving" here):

1. Solve a 1x2x3 - Less moves than normal solving --- Advantage.

2. Solve opposite 1x2x3 (using 4 possible colors on D) - The same amount of moves as a normal solve, you don't have time to be looking for the best first pair.

3. Corners - Requires three thoughts; 1. L/R color orientation, 2. Find the two "same" colors on the white corners, 3. Find the two "same" colors on the yellow corners (it really is split like this, try it). NMCMLL = 1.5 times longer than CMLL recognition, even if you try to find the "same" colors while figuring out the L/R color orientation. ---Disadvantage

4. LSE - Also takes longer, because of 2 reasons; 1. It's move average is 1 more than normal solving (because of the last R or L move) and 2. Edge orientation recognition is a little longer because you have to locate the L/R edges to know their orientation. ---Disadvantage

For CN solving that makes 1 advantage, 2 disadvantages, and one undecided (the second block). Even if you worked on the undecided one to make it count as 1 advantage, you are only at 0 (the same as normal solving). You would have to solve that second block much faster (even with its 1-3 fewer moves) than normal solving in order to have the overall CN solve be faster than a normal solve. But with the second block in CN solving being the same as a normal solve I don't see how that's possible. You would have to have great look-ahead abilities while solving the first block to be able to see very short solutions to the second block.

The overall move-count for CN solving would only be 1-3 moves less than normal solving Explanation: minus between 2 and 3 moves for the first block in CN solving (I was averaging 7 moves for the first block), but add .75* moves for the last turn during the LSE.

And I know all of this is strange coming from the person that made the NMCMLL charts on Gilles' page. I used to think that non-matching blocks was the future of the method, but after putting so much time into figuring out exactly how to think during every step of a CN solve, I started to see all of it's little disadvantages, it disappointed me, I really enjoyed solving the weird blocks.

*There are four types of blocks you can have on the right side

The main problem is that time restraints allow for worse move count and the steps aren't as fingertrick friendly as Fridrich.

Not as fingertrick friendly? M/U/R feels faster than F/U/R to me.

If people like Erik or Yu would have chosen Roux or Petrus the Fridrich method would already have been beaten. As far as I can tell no "Blockbuilder" is currently putting as much time into his/her cube as those guys.

Kind Regards
Martin

I know I don't put as much time into my solves as the average Fridrich user, and it doesn't seem like Gilles and Thom does anymore either. Johannes also doesn't solve much anymore.

 

[waffle=ijm]

what Athefre said above

yeah i agree that color neutral Roux with NMCMLL will completely ruin your recognition...it was just a thought seeing that i saw NMCMLL guide and that there are color neutral solvers that can go at full speed. Just combining the two seemed like a good idea but would take lots ad lots of time to practice recognition.

 

[brunson]

what Athefre said above

yeah i agree that color neutral Roux with NMCMLL will completely ruin your recognition...it was just a thought seeing that i saw NMCMLL guide and that there are color neutral solvers that can go at full speed. Just combining the two seemed like a good idea but would take lots ad lots of time to practice recognition.

If color affects your recognition, then you're not truly color neutral.

 

[edavies]

im sad that nobody has mentioned ZZ yet.
once i have EOLine set up (i.e im not very good at it so ill start timing a solve with it done) i can average about 12-14s. add 3s for EOLine execution and i can average of 12 a whole 2-5s better. its ridiculous. for example, if i get the left 2x2x3 block done, the rest is two-gen up until PLL.
i think that if anyone/any method is going to beat fridrich, it would be ZZ.
it beats it on move count, can beat it on tps, beats it on lookahead, beats it on move ergonomy (no cube rotations), and is generally just really nice with a simple Z rotation you can have two gen f2l too.
its only real weakness is EOLine lookahead and execution - it takes a long time to get used to it, and the <=8 moves to do it are not always very nice.
hah, it even beats the cross on average number of moves to completion

Hear hear, ZZ has to be a method with potential.

I don't think ANY method can beat Fridrich at the moment simply because Fridrich has the best resources, reputation and the most time from the community.

Practice obvoiusly counts for more than micro-improvments in method. Just look at how many of the good people are truly colour neutral, or even opposite colour neutral. The chance of a good x-cross really improves, but times, alas do not.

Once beyond that understanding, assessing potential is the crucial thing. The improvements to Fridrich (its variants) have pretty much all focused on LL and involve HUGE numbers of algs. ZZ offers a more ergonomic (therefore potentially faster) F2L and a better last layer (EO already done). The 'cost' of ZZ is the complexity of EOLine compared with the cross. IMHO if you studied the mental involvement of a top Fridrich solver during inspection, they will be working as hard as you need for EOLine anyway; they just haven't practiced EOLine as it isn't mainstream. Look at top cubers 2-gen times for proof that ergonomics counts.

Another factor nobody's mentioned here is the technology (Ok this is a method thread sorry guys). My uncle heard about my involvement in the UK open and was astounded that the top 6 cubers were (AFAIK anyway) using £4 chinese plastic cubes. In what other 'sport' do you see the elite using crappy kit. Golf clubs, tennis rackets, you name it; pros pay £400 for it. I'm waiting for a ridiculous rich kid to commision MIT's engineering department to produce him a carbon fibre, graphene nano-tech lubed wondercube with titanium screws and a silicone core. He could throw in ball race bearings and a hydraulic suspension system, who knows? The V-cubes are proof that there's a wealth of developments that haven't happened.

 

[Dene]

what Athefre said above

yeah i agree that color neutral Roux with NMCMLL will completely ruin your recognition...it was just a thought seeing that i saw NMCMLL guide and that there are color neutral solvers that can go at full speed. Just combining the two seemed like a good idea but would take lots ad lots of time to practice recognition.

If color affects your recognition, then you're not truly color neutral.

Lol, another person that doesn't understand the concept of CN, huh brunson?

 

[MistArts]

How about this method...

Orienting all the F2L pieces and place midges in the middle layer
Solve F2L
OLL/PLL or CLL/ELL

 

[rckclmb124]

How about this

1.Heise style f2l-one slot
2.finish f2l and edge orient and permutate if posible
3.Cll

I think it could be very fast if you could get the f2l-one slot done fast and efficiently if you follow the steps of the method. The only problem is there are probably a ton of algs needed for finishing the f2l and orient plus permutate the edges. I am very interested in looking into this if it has not been thought of already.

 

[yurivish]

We shouldn't just sit here and talk; we should actually form a group of people dedicated to improving ZZ. There seems to be some consensus that it's a method worth developing and has the capacity to be a worthy alternative to Fridrich in terms of speed.

I use ZZ; switched over to it a few months ago when that thread came out. I currently average around 22-23, with my best average of 12 being 20.80.

- Finding a "missing link", something that can be done on-the-fly after finishing the left block, in order to make the rest of the cube 2-gen, would be a big step forward, though it might not be possible - there has been some effort in this area already.

- Figuring out several "standard" optimal/fingertrick ways to solve certain edge orientation configurations, along with some special cases would also be a boon. Right now I do ALL of them by grouping edges on the F and B faces and then turning them one 90º turn. However, there are cases for which there are efficient M-slice solutions. Some of these can be taken directly from Roux, such as M' U M. One additional restriction for us is that we want the centers to remain constant after orientation.

- "Algorithms" for the F2L – I've developed an internal set of ways that I do certain cases. I'd be more than happy to share. I think that mine are all pretty straightforward, but I think it would be a good idea to get a list of different ways people solve the same cases. I'm sure that several of the cases have better solutions than the one I have been using.

- Special cases. Is there any equivalent to the XCross for ZZ, by which I mean any advanced technique that can be used at the outset to do more than just the EOLine?

 

[blade740]

- Special cases. Is there any equivalent to the XCross for ZZ, by which I mean any advanced technique that can be used at the outset to do more than just the EOLine?

EOcross works.

 

[brunson]

what Athefre said above

yeah i agree that color neutral Roux with NMCMLL will completely ruin your recognition...it was just a thought seeing that i saw NMCMLL guide and that there are color neutral solvers that can go at full speed. Just combining the two seemed like a good idea but would take lots ad lots of time to practice recognition.

If color affects your recognition, then you're not truly color neutral.

Lol, another person that doesn't understand the concept of CN, huh brunson?

Yeah, you pretty much drilled it home in the other thread.

 

[DavidWoner]

- Special cases. Is there any equivalent to the XCross for ZZ, by which I mean any advanced technique that can be used at the outset to do more than just the EOLine?

EOcross works.

or EOLine+1x2x2

I think ZZ users should also work on blockbuilding for the left and right blocks, instead of using 2-gen Fridrich.