[Video] Roux as a Beginner Method

[Chree]

There's a movie on DeeDubb's chanel:

https://www.youtube.com/watch?v=zaKmryHSKcc&list=PLajHGvYF36nQ6VlcMyobP3mnK1GNv-gex&index=2

Nice.. that was really helpful.

 

[TDM]

I am not using he's algs I'm using the OLL and PLL algs I use for CFOP I understand that the are not optimal. At this point why not just start learning full CMLL? Is there a reason other than timer?

If you're already using 2-look CMLL then yes, start learning CMLL.

 

[GuRoux]

I am not using he's algs I'm using the OLL and PLL algs I use for CFOP I understand that the are not optimal. At this point why not just start learning full CMLL? Is there a reason other than timer?

if you are serious about learning roux, then start learning cmll. usually when people start, their lse efficiency is pretty bad, so that might be another place to look at. try to usually have 18 moves or less for lse, and you should be good.

 

[King Mike]

Hey DeeDubb!

Great tutorial! Just one question....
Can't you just sledgehammer pairs that have white facing up, instead of avoiding them?

 

[cuber8208]

Can't you just sledgehammer pairs that have white facing up, instead of avoiding them?

Because then you'd be breaking your left block (even though it would be temporary) with the F moves. r U r' or r U R' would be better as well as more efficient and allows you to maintain the death-grip on the left block

 

[philipneri]

Keep me updated on your progress and best of luck.

DeeDubb, I am not so sure how serious you were about keeping you updated on my progress. Nonetheless, here is my 2 month report.

I learned Roux on April 1 using your videos. At that time I was a 2 minute solver. Now on June 1, I am 45 second solver. My step-by-step time breakdown is 10/15/8/12. During inspection, I can plan all of the 1x2x2 square and predict the location of the remain edge and corner.

My personal best solve time is 29 seconds (achieved three days ago). My personal best Ao5 is 37 seconds (achieved about a week ago). I "know" 21 out of the 42 CMLL algorithms. I use know in quotation marks because I still have to use a cheat sheet sometimes.

By July 1, I hope to have "know" all 42 CMLL algorithms and be a 40 second solver. By the end of the summer, I hope to really know all 42 algorithms without a cheat sheet and be a 35 second solver.

Thanks for inspiring me to be a Roux solver. I have enjoyed it so much more than all the other methods (CFOP, Petrus, ZZ, LBL, EF, CF, 8355). Actually, I like Petrus a lot too; I am just not particularly fast at it.

 

[adimare]

Hello!

My brother-in-law wants to learn how to solve the cube and I was thinking about teaching him your version of Roux instead of an LBL approach. I have a couple of concerns tho, the main one being the lack of a fixed point of reference during the first stages of the solve. Let's say he starts with yellow on top and blue on left as his orientation, while solving the first 2x2x1 he buffers the blue-red edge piece on DF using M and pairs it up with its corresponding corner using M2 (so the yellow center will end up in the B face, red on top, etc), won't he have trouble figuring out that the next edge piece he needs to solve is blue-orange? This could be solved by adding the extra step of adjusting the M layer after every pair insertion and making sure yellow is on top, but maybe I'm overthinking it and that's an unnecessary step. Have you or anyone else taught this method to any beginners and ran into this issue? Has anyone that learned this without knowing any other method to solve the cube had trouble with this?

 

[dboeren]

Double post

 

[dboeren]

I'm learning Roux now using this tutorial (along with Waffo's page) and I don't think it's a concern. You just learn that red is front and orange is back. It doesn't really matter what the center pieces in the M slice are at that point in the solve, they should be considered free to rotate as you like.

Currently I'm working on 2-look CMLL from Waffo's page.

 

[DeeDubb]

DeeDubb, I am not so sure how serious you were about keeping you updated on my progress. Nonetheless, here is my 2 month report.

I learned Roux on April 1 using your videos. At that time I was a 2 minute solver. Now on June 1, I am 45 second solver. My step-by-step time breakdown is 10/15/8/12. During inspection, I can plan all of the 1x2x2 square and predict the location of the remain edge and corner.

My personal best solve time is 29 seconds (achieved three days ago). My personal best Ao5 is 37 seconds (achieved about a week ago). I "know" 21 out of the 42 CMLL algorithms. I use know in quotation marks because I still have to use a cheat sheet sometimes.

By July 1, I hope to have "know" all 42 CMLL algorithms and be a 40 second solver. By the end of the summer, I hope to really know all 42 algorithms without a cheat sheet and be a 35 second solver.

Thanks for inspiring me to be a Roux solver. I have enjoyed it so much more than all the other methods (CFOP, Petrus, ZZ, LBL, EF, CF, 8355). Actually, I like Petrus a lot too; I am just not particularly fast at it.

Sorry I haven't been following this as close as I should. That's great to hear! I'm glad you are keeping up with Roux. I have been focusing on other things now, so I haven't been doing Roux solves like I should. Just keep adding the CMLL algs. There's no hurry on those. Block building and L6E make the biggest difference. Congrats on the sub 30 solve!

Hello!

My brother-in-law wants to learn how to solve the cube and I was thinking about teaching him your version of Roux instead of an LBL approach. I have a couple of concerns tho, the main one being the lack of a fixed point of reference during the first stages of the solve. Let's say he starts with yellow on top and blue on left as his orientation, while solving the first 2x2x1 he buffers the blue-red edge piece on DF using M and pairs it up with its corresponding corner using M2 (so the yellow center will end up in the B face, red on top, etc), won't he have trouble figuring out that the next edge piece he needs to solve is blue-orange? This could be solved by adding the extra step of adjusting the M layer after every pair insertion and making sure yellow is on top, but maybe I'm overthinking it and that's an unnecessary step. Have you or anyone else taught this method to any beginners and ran into this issue? Has anyone that learned this without knowing any other method to solve the cube had trouble with this?

I don't think this is an issue unless you're coming from a cross first method. A new solvers won't be focused on the M Slice centers because it hasn't been "programmed" into them yet. So, as long as they memorize a fixed block. Blue on the left, white on the bottom in my videos, then they will also just memorize blue red on front and blue orange on the back. I have taught a few people Roux and haven't had any issues with this.

 

[dboeren]

I'm solving in about 60 seconds now, though I'm still timing my blocks and "the rest" separately so there may be a little bit of extra recognition time in between. Close enough...

At this point I'm doing 2-look CMLL from Waffle's page (minus the diagonal permutation, I just do adjacent twice right now) and using M' U M' to iterate through the bad edge cases until I get to the arrow. So clearly there are improvements that can be made in the second half of the solve - most of which is memorization stuff:
Full CMLL
Learn direct solves for all the bad edge cases
Learn direct solves for each of the final edge cases

What I feel I really want to improve though is my block building which is just shamefully bad My best ao12 right now is a hair under 30 seconds for just doing right/left blocks, but I know my move count is not very good as I'm still using Deedubb's "beginner" method of putting the edge in the front bottom loading spot. If you ever have time to make another video, I nominate a "next steps" to block building showing better techniques for this stage.

One thing in particular that I seem to have trouble with is the 2nd block bottom edge. After I build my left block, my next step is to find and place the green/white edge on the bottom right and if it ends up in a bad position it takes me like 5 moves to get it there which just seems really awkward. I've also tried using my inspection time to solve both the blue/white and green/white edge first before trying to take on the first blue pair - does that sound like a better approach? The plus is easier placement of the green/white edge since the cube is still pretty "open". The drawback is that I have to maintain that edge while doing the blue pairs and occasionally I forget to restore it after making some move and the solve goes off the rails because I probably don't notice for a while

Anyway, just wanted to say "hi" to all the other Roux newbies out there and encourage you to keep practicing.

 

[GuRoux]

if i were you i would first learn all the lse cases, be able to sub 25 moves on average. then work on block building, i think 35 moves is a good goal. and i wouldn't worry about cmll now since it's less interesting and makes very little difference in your speed.

 

[theROUXbiksCube]

I would learn the Y perm though because double adjacent swap is okay but one alg is good. Learn the cycles for each EO case and stuff

 

[PenguinsDontFly]

I would learn the Y perm though because double adjacent swap is okay but one alg is good. Learn the cycles for each EO case and stuff

what about your glorious V perm? or ryans overpowered E perm?

 

[supercavitation]

what about your glorious V perm? or ryans overpowered E perm?

I still use N perms

 

[dboeren]

if i were you i would first learn all the lse cases, be able to sub 25 moves on average. then work on block building, i think 35 moves is a good goal. and i wouldn't worry about cmll now since it's less interesting and makes very little difference in your speed.

Thanks, it's good to know what to prioritize. I'm glad it's not CMLL too, too many algorithms tend to be boring. I will benchmark my LSE on the next few solves to get an idea of how many moves I'm using right now.

I would learn the Y perm though because double adjacent swap is okay but one alg is good. Learn the cycles for each EO case and stuff

Yes, it's definitely on my to-do list. Noob question though, do the Y, J, E, and N perms all accomplish the same thing at this state?

I did a quick search and found this page which seems to document each algorithm (assuming these are the correct ones):
https://www.speedsolving.com/wiki/index.php/PLL

Since edges are not solved yet at this step in Roux, I believe that means that it doesn't matter whether the algorithm you use alters the edges, or that a high level solver could look at the edges, decide if he liked them, and then apply an algorithm that preserves them OR another algorithm that doesn't hoping for a better configuration (or even knowing the final configuration)?

If I can choose any one of these, then I suppose it's just a matter of feel and benchmarking to see which version works best.

Edit: The diagonal permutation on Waffle's page is not one of Y/J/E/N. It is:
R' U L' U2 R U' x' U L' U2 R U' R x'

So what's up with that? It doesn't seem to be used anywhere else...

 

[TDM]

Edit: The diagonal permutation on Waffle's page is not one of Y/J/E/N. It is:
R' U L' U2 R U' x' U L' U2 R U' R x'

So what's up with that? It doesn't seem to be used anywhere else...

This is an N perm executed slightly differently:
(R' U L' U2 R U' L)2 = R' U L' U2 R U' L R' U L' U2 R U' L
he's done some rotations to change a few of the moves a little, but it's still very similar to the N perm.

 

[DeeDubb]

The only compelling reason to use the Y Perm over other perms is because it will also help you if you plan on getting into blind solving (most people start with OP corners, which uses a slightly modified Y Perm) Otherwise, use any diagonal you want.

 

[dboeren]

This is an N perm executed slightly differently:
(R' U L' U2 R U' L)2 = R' U L' U2 R U' L R' U L' U2 R U' L
he's done some rotations to change a few of the moves a little, but it's still very similar to the N perm.

I wonder what the motivation was to add cube rotations?

Anyway, I think I will go back to the original N-perm version for now instead - thanks for pointing this out.

Definitely no plans to go blind at this point, or likely ever. If I eventually manage to get good at 3x3 I will probably try 4x4 next.

 

[TDM]

I wonder what the motivation was to add cube rotations?

Possibly to make it easier to execute - the first one replaces two moves with zero moves, and the second one allows you to do an R move instead of an L move, and most right-handed people find R moves faster.