Anon's Cubing Progression // Beginner Roux gains

[Anon]

I'm an oldtimer who found a 3x3 at a garage sale. The cube is apparently a Moyu AoLong V2 ca. 2015. Eight years is a long time, but after a through cleaning and the cheapest lube I could buy on Amazon, it seems to be turning well. I'd like to get a decent competency with it. I looked through the mainstream methods and settled on Roux, since fewer algorithms appeals to me. Eventually I think I would like to learn Heise and some of the theory that goes into Blind Solving and Fewest Moves, but that's far in the future.

I've spent the last week learning the seven corner algorithms from 2-look OLL. That combined with a J-perm should get me though CMLL in two looks. (If I see a corner swap, I'll just use two J-perms.) I found this tutorial to be quite helpful with remembering the algorithms. I've also learned some basic finger tricks and the standard move notation.

I did 15 solves to set some rough baseline statistics.
total time: 1 min 30 sec
FB: 30 sec
SB: 30 sec
CMLL: 15 sec
LSE: 15 sec

There's lots of room for improvement at least! So much room that it's hard to know where to begin. In FB and SB I feel lost and slow, though I can't articulate exactly why. I keep splitting my timer early during CMLL because I forget that corner permutation comes after corner orientation. And in LSE, I wander aimlessly through Edge Orientation until I stumble across an arrow case.

I plan to vaguely follow this guide on my path to competency.

----

I think I'll start by focusing on EO of the LSE step. I hope that the constrained moveset will help me build intuition and cube knowledge without memorizing a bunch of algorithms for each case. I just need to (1) understand how M' U2 M and M' U M and their variants effect the six edges and (2) how to use those effects to get to the arrow case. Rather than worrying about times, I feel I must go super slow and think things through. Fluency will come (hopefully).

 

[Anon]

Progression log (3x3 Roux, mean of n, standard deviation in every other row)

Date	n	time	FB	SB	CMLL	LSE
01-10-2023	15	1:26.45	31.41	26.59	13.38	15.06
σ		15.31	9.80	7.15	3.36	4.97

(I use means instead of averages to make later statistical analysis easier. Admittedly, I could use some kind of resampling technique to keep averages, but then I'd have to remember all the individual solve times. This way, I can just blindly apply Welch's t-test with only the means and variances. I'll use Bonferroni or, if I'm feeling fancy, Holm to control family-wise error rate. And I'll attach the data in a csv... just in case.)

Spoiler: Personal Goals

• (date) Nothing!
• (upcoming) sub-1min single
• (upcoming) sub-1min ao5
• (2023-10-26) Jorian's Quest: Y-perm
• (2023-10-28) Jorian's Quest: Ub-perm
• (upcoming) Jorian's Quest: T-perm
• (upcoming) Jorian's Quest: Jb-perm
• (upcoming) Jorian's Quest: F-perm
• (upcoming) Jorian's Quest: N-perm
• (2023-11-07) alg: 6-flip LSE
• (upcoming) y2 color neutral FB

 

[Anon]

Today I was mostly busy with work. I did have some time to play, and I realized that both the M' U2 M family and the M' U M family affect pairs of edges in concert. So, I only need to think about pairs of bd, aw, and cu stickers (using Speffz notation) instead of all six edges individually. M' U2 M transports edges between the U and D layers while retaining orientation. Whereas the M' U M family swaps the orientation of 4 edges by storing one column in the bd positions.

 

[Arcanist]

I'm an oldtimer who found a 3x3 at a garage sale. The cube is apparently a Moyu AoLong V2 ca. 2015. Eight years is a long time, but after a through cleaning and the cheapest lube I could buy on Amazon, it seems to be turning well. I'd like to get a decent competency with it. I looked through the mainstream methods and settled on Roux, since fewer algorithms appeals to me. Eventually I think I would like to learn Heise and some of the theory that goes into Blind Solving and Fewest Moves, but that's far in the future.

I've spent the last week learning the seven corner algorithms from 2-look OLL. That combined with a J-perm should get me though CMLL in two looks. (If I see a corner swap, I'll just use two J-perms.) I found this tutorial to be quite helpful with remembering the algorithms. I've also learned some basic finger tricks and the standard move notation.

I did 15 solves to set some rough baseline statistics.
total time: 1 min 30 sec
FB: 30 sec
SB: 30 sec
CMLL: 15 sec
LSE: 15 sec

There's lots of room for improvement at least! So much room that it's hard to know where to begin. In FB and SB I feel lost and slow, though I can't articulate exactly why. I keep splitting my timer early during CMLL because I forget that corner permutation comes after corner orientation. And in LSE, I wander aimlessly through Edge Orientation until I stumble across an arrow case.

I plan to vaguely follow this guide on my path to competency.

----

I think I'll start by focusing on EO of the LSE step. I hope that the constrained moveset will help me build intuition and cube knowledge without memorizing a bunch of algorithms for each case. I just need to (1) understand how M' U2 M and M' U M and their variants effect the six edges and (2) how to use those effects to get to the arrow case. Rather than worrying about times, I feel I must go super slow and think things through. Fluency will come (hopefully).

Goodluck! Roux is a very cool method and I think may overtake CFOP one day. So I should probably learn it...

 

[Kami]

Today I was mostly busy with work. I did have some time to play, and I realized that both the M' U2 M family and the M' U M family affect pairs of edges in concert. So, I only need to think about pairs of bd, aw, and cu stickers (using Speffz notation) instead of all six edges individually. M' U2 M transports edges between the U and D layers while retaining orientation. Whereas the M' U M family swaps the orientation of 4 edges by storing one column in the bd positions.

you should look into getting a new cube. if you need as cheap as possible while seill being good, rs3m 2020.
but i personally reccomend the base model dayan guhong pro for a few dollars more on thecubicle.com

 

[Anon]

Good morning all. I hope the weekend is treating you well. I've had precious little free time this week, but I have learned an important Truth! I much prefer reading about advanced Rubik's techniques to actually practicing. -_-

(By the way, forgive me if I'm just badly regurgitating concepts you all already understand quite well. I'm writing this mainly to get my own thoughts in order, and perhaps as a diary/reference.)

I read about Pinkie Pie, which combines some pre-work during SB with the OLL-CP algorithm set. This lets the variant mush CMLL and EOLR into a single step for the low, low price of 331 algorithms. Verdict: Algorithms scare me~

Speaking of which, I also learned about EOLR which combines Edge Orientation and the ULUR substeps (4a and 4b of the beginner Roux method). EOLR involves choosing the 'right' elements of the M' U2 M and M' U M families get a 'good' arrow. Success simplifies the ULUR substep greatly. Verdict: A natural second step given my current EO focus.

Mis-oriented centers (MC) is a twist on EO where the F/B center is held on top instead of the U/D center. Edges are "good" if their F/B stickers are facing up or down. Excepting of course, the UL and UR edges, who have no F or B stickers. They are "good" when the U sticker faces up or down, same as before. MC edge orientation is not strictly compatible with standard edge orientation, but once the UL and UR edges are inserted, it is only an M' away from standard. Verdict: Very cool, but you only use this when you can tell that MC will give you a better case than standard. And I lack the experience to tell.

UFUB is a replacement for the ULUR, where you solve the UF and UB edges instead of UL and UR. Then just throw those edges to the side and solve L4E as normal. Theoretically, it's possible to combine MC and UFUB, but you'd need new rules for identifying "good" edges. ('good' is side color on top except the UFUB edges who want top color on top, or something like that) Verdict: Like MC, an alternative path on the persuit of easy cases. Beyond me.

----

So let me start talking about things I actually plan to practice in the near future. I'm going to swap back to Speffz notation because writing UDFBRL notation in the previous paragraph made me sad.

EOLR (because that is totally appropriate for my skill level) relies on 'good' arrows. If edges bcdu are misoriented (an arrow pointing towards the solver), then 'good' arrows have the bd edges in positions ab, ad, wb, wd, and cu. Then you can choose to solve the arrow so as to stick the bd edges in the D layer, ripe for insertion. (The cu cases requires a quick jaunt into MC.) In particular, if one bd edge is misoriented then it must be on the side of the arrow. And if both are bad, they must be the tip and the target. Both bd edges can only be good by making the solver sad, because he must preform a discrete ULUR substep.

For that matter, EO requires use of the M' U M family to drop into 4 bad edges, then M' U2 Ms to move into the arrow. Since the M' U M family swaps the orientation of 4 edges, the progression goes

6 + (0 - 4) = 2 bad edges

2 + (3 - 1) = 4 bad edges

Theoretically, 6 bad edges can be immediately taken to 4 with 6 + (1 - 3), but there's not room for it on the cube. (Where is the +1 bad edge comming from? All the edges are already bad.) In practice, 6-flips have bespoke algorithms or clever applications of MC.

anyway, tomorrow, I want to practice getting from 2 bad edges to the arrow without M' U2 M repositioning.

 

[Anon]

Thanks to Arcanist for the well wishes!

And thanks to jawklqnn for the kind recommendation. I see the MoYu RS3 M 2020 is impressively priced while being highly regarded in e.g. this video. However, (1) I feel that I am not yet at the level where the differences between speedcubes will make a significant impact on my performance. And (2) I am somewhat attracted to the mystique of getting good on the same cube that inspired me to look into speedcubing in the first place, the one that I found and repaired from nothing.

That said, with the power of adulthood, I did just randomly throw some money at a RS3 M 2020 on Amazon. It's not adultery. I'm just getting a point of reference, to be sure I tensioned the AoLong's springs properly. ^^

 

[Anon]

With somewhat diligent practice, my ao12 for LSE has gone from ~15 sec to 16.47 seconds! That's a decrease of -1.47 seconds!

I should trust the process, but slower times feel bad. I also think I've bitten off more than I can chew.

So, I'm going to stop thinking about EOLR, and I'm going to stop using 'push through' algorithms like M U M or M' U2 M'. For now, I'll stick to the ones that leave at least 2 edges fixed for better visualization.

I also should cut out some time to treat each of the two-bad edge cases in detail. I feel that I'm not seeing enough of them just randomly in solves.

 

[TheSpeedSkewber]

With somewhat diligent practice, my ao12 for LSE has gone from ~15 sec to 16.47 seconds! That's a decrease of -1.47 seconds!

I should trust the process, but slower times feel bad. I also think I've bitten off more than I can chew.

So, I'm going to stop thinking about EOLR, and I'm going to stop using 'push through' algorithms like M U M or M' U2 M'. For now, I'll stick to the ones that leave at least 2 edges fixed for better visualization.

I also should cut out some time to treat each of the two-bad edge cases in detail. I feel that I'm not seeing enough of them just randomly in solves.

Its very cool learning about Roux with you, and you explain it very well! Good look on your further pursuits, and well done on your progression.

 

[Luke Garrett fan]

Get a new cube like the ys3m bc or tv3 flagship

 

[Anon]

Its very cool learning about Roux with you, and you explain it very well! Good look on your further pursuits, and well done on your progression.

Cheers I'm glad you are enjoying the thread. I do think I've done well on my progression. I'm very happy to have improved by negative 1.47 seconds; that is, I got slower. XD

Thanks to Luke Garret fan for more speedcube recommendations. Those will be far in the future however, for reasons I outline elsewhere in the thread.

 

[Anon]

I've changed the way I'm visualizing the last six edges. Previously, I was imagining points of lights above every bad edge and tracking their movement. However, M U M' changes the bad edges, for example. So, I'd have to turn off some lights and turn on others for each M-slice move. It was too much.

Instead, I'm imagining 3 pipes which pair the aw, bd, and cu edges (since these edges pair are preserved by the LSE algs). The pipes are kinked so that they protrude from the side sticker. For example, when the cube is solved, there is a u-shaped pipe protruding from the i and k stickers and a straight pipe running through the e and m stickers. As I solve EO, I can visualize the pipes moving as rigid bodies. Then I can just look at the kinks to find where the side stickers are. For example, to get into an arrow, I just need to get 3 pipe ends pointing up and one pointing down.

To aid in memory, I visualize the F-face pipe as being made of a fuzzy pink pipe cleaner, the U-face as a glass tube, and the B-face as concrete rebar.

I'm not sure if any of this is coherent to read, but I'm going to try it out and report back.

 

[Anon]

It's been a while. IRL has distracted me from cubing. I'm making slow progress on EO and EOLR. As a break, I decided to learn a Y-perm. A Y-perm will prevent me from needing to use two back-to-back J-perms for diagonal corner swaps cases in my CMLL.

I started with Brian Sun's video and Jperm's video to pick an algorithm. I chose

(F R U' R' U') (R U R' F') (R U R' U') (R' F R F')

i.e.

(F R U' R' U') (R U R' F')(sexy)(sledge)

And I believe this is the standard algorithm. I used both of Mike Shi's videos to help remember the algorithm. As Mike notes, it's interesting that the two halves of this Y-perm algorithm are OLL algs, with (F R U' R' U') (R U R' F') solving OLL 37 and (sexy)(sledge) solving OLL 33.

Of course, you can pair any OLL algorithm with a second OLL algorithm that fixes whatever OLL case the first alg maps the solved state into, and that will give you a PLL algorithm. (Standard break and restore.) But what is interesting is that it's the most popular algorithm.

For example, swapping the two OLL halves gives

(sexy)(sledge)(F R U' R' U') (R U R' F')

which is a T-perm algorithm. But certainly not the best one. (Wait, I just realized, this _is_ the best T-perm after cancellation. WTF?)

(R U R' U') (R' F R F')(F R U' R' U') (R U R' F') = (R U R' U') (R' F R2 U' R' U') (R U R' F')

Did I not need to learn a J-Perm at all?

As an ending question, if a PLL algorithm, P, can be factored into two pieces, P = A B, and both A and B are OLL algorithms, will B A necessarily be a PLL algorithm too? (not in general, see below)

 

[Anon]

Oh, ofc, the answer is false because A and B don't change the orientation of specific pieces like the Green-Orange-Yellow corner. They just change the orientation of whatever piece happens to be in, e.g., the UFR corner.

As a counter example,

(Y-perm) (A-perm) =(F R U' R' U' R U R' F') (R U R' U' R' F R F')(x R' U R' D2 R U' R' D2 R2 x')

is a PLL algorithm and factors into

A = (F R U' R' U' R U R' F')

and

B = (R U R' U' R' F R F')(x R' U R' D2 R U' R' D2 R2 x')

but

B A = (R U R' U' R' F R F')(x R' U R' D2 R U' R' D2 R2 x') (F R U' R' U' R U R' F')

does not preserve last layer orientation. So the Y-perm is lucky in this way.

 

[Jorian Meeuse]

(R U R' U') (R' F R F')(F R U' R' U') (R U R' F') = (R U R' U') (R' F R2 U' R' U') (R U R' F')

Did I not need to learn a J-Perm at all?

Notice how a different rearrangement of the parentheses gives a J perm: (R U R' F') (R U R' U') (R' F R2 U' R' U')
J, T and Y perms are simply rearrangements of each other.
Then, an F perm is just an R' U' F' setup into a T perm, with a cancellation at the end.
An N perm is just an R U R' U setup into a J perm

 

[n0nco]

its cool to see older people learning how to solve the cube! i think most people learn at a younger age, i personally know cfop and barely know how to use roux so i don't understand much, good luck!

 

[Kami]

Thanks to Arcanist for the well wishes!

And thanks to jawklqnn for the kind recommendation. I see the MoYu RS3 M 2020 is impressively priced while being highly regarded in e.g. this video. However, (1) I feel that I am not yet at the level where the differences between speedcubes will make a significant impact on my performance. And (2) I am somewhat attracted to the mystique of getting good on the same cube that inspired me to look into speedcubing in the first place, the one that I found and repaired from nothing.

That said, with the power of adulthood, I did just randomly throw some money at a RS3 M 2020 on Amazon. It's not adultery. I'm just getting a point of reference, to be sure I tensioned the AoLong's springs properly. ^^

How’s the new cube? (assuming you already have it)

 

[Anon]

Notice how a different rearrangement of the parentheses gives a J perm: (R U R' F') (R U R' U') (R' F R2 U' R' U')
J, T and Y perms are simply rearrangements of each other.
Then, an F perm is just an R' U' F' setup into a T perm, with a cancellation at the end.
An N perm is just an R U R' U setup into a J perm

Wait, what!? That's nuts!

 

[Anon]

its cool to see older people learning how to solve the cube! i think most people learn at a younger age, i personally know cfop and barely know how to use roux so i don't understand much, good luck!

Much appreciated! I not sure I'll ever be amazing, but I think sub-20 is eventually attainable, even for an old fogey like me. I totally get that though, every method is like it's own world. I was idly watching some EO-line stuff for ZZ and the ideas were completely different from Roux.

How’s the new cube? (assuming you already have it)

I actually have the RS3M 2020 in a box on my desk right now. I gave it a few turns, and it feels really nice. It even has little magnets in it! How nifty is that? A little sandy though, needs lube out of the box.

That said, I decided to leave the RS3M in its box until I hit a sub-45 Mo3, both for the sentimental reasons I outlined above and as a little extra motivation to practice.

 

[Bingus Ringus McDingus]

I'm seeing a lot of improvement here, and I think that you could definitely become really good pretty quickly! I learned pretty slowly because I had a chicken-crap method of learning where I just carried a little pamphlet around with me until I understood it instead of actually drilling algs, and I think with your good practice techniques, sub-40 is easily attainable! Roux is a great method to start off with.

Much appreciated! I not sure I'll ever be amazing, but I think sub-20 is eventually attainable, even for an old fogey like me. I totally get that though, every method is like it's own world. I was idly watching some EO-line stuff for ZZ and the ideas were completely different from Roux.

Don't doubt yourself about your age versus your skill, look at @Mike Hughey who is 61 and still averaging a solid 15. Good luck!