• Welcome to SpeedSolving.com — the world’s largest puzzle community!

    You’re currently browsing as a guest, which means you have limited access to discussions, resources, member profiles, and community features.

    Join 50,000+ cubers and puzzle enthusiasts from around the world to ask questions, share solves, improve faster, and be part of the community.

    Registration is free, fast, and easy — create your account today.

    Already a member? Log in here to hide this message and start participating.
I just got the dumbest scramble ever. I'm pretty new to 2x2 and I only use Ortega but this scramble had the yellow face already set up and the solution was (F sexy sexy F') and then that was it.

SCRAMBLE: R U2 R2 U' R2 U R2 U2 R' U2

Just a note, these scrambles may not be a rare occurance for everyone else, but for me who is new to 2x2 I thought this was pretty dumb.

:)
 

Attachments

  • IMG_2621.PNG
    IMG_2621.PNG
    88.9 KB · Views: 12
@Metallic Silver it looks good (for OH especially), but you would probably want to have it listed as 3/4 or half belt but usually half, and note that the cross doesn't need to be fully made in Inspection, it can be off by a move which makes it more efficient.

RA-OB (Ribbon-Alpha variant of LLOB) has gotten me many sub-40 example solves, I believe that the method can be really great. It looks like the average speedsolve will have the efficiency of standard Roux, but the average of the solves I've done thus far (about 50) is 43 using HARCS for blocks, but speed optimized TOLS/TTLL. To me at least, the ergonomics of the method, the speed of blockbuilding in Roux, and the TPS spam nature of the last 2 (even 2.5 with MU DF DB) steps can lead to a method that can rival (or even beat) CFOP quite easily.

Here's a 34 move solution from the example solve thread, F2L-C wasn't that great but had excellent TOLS/TTLL cases.

B D2 L2 F2 U2 R2 F U2 B' L D2 U L D U B L' U' R2

z2 y' M' D M B D'//FB (5/5)
U r U' R2 r' U' r U' R M U2 M' U2 M//F2L-C (14/19)
R' D' R U R' D R//TOLS (7/26)
U' R2 U2 R2 U' R2 U' R2//TTLL (8/34)


When I get my camera in I'll post a tutorial on YT and I may start on an official document, let me know if you all think this is good enough to officially release to the public! Any feedback will be appreciated, and if you'd like to help PM me!
 
Look at the HD Method! The method is almost exactly CT on 2x2 but with several advancements! I made a thread on it on the forum so it should be easy to find. Good luck!
 
I think we may have been taking the wrong approach to new method development. Right now we imagine what how we want the pieces to be solved, then we use an algorithm generator to find a way to create that transformation. The problem is the algorithms are usually non-ergonomic.

I am taking up a different approach with very promising results. The reverse approach is to try dozens or hundreds of the shortest, finger friendly move sequences, and then observe their effect on the cube. With this this ultra-fast 'set' of sequences, now design a method to use them.
 
I have noticed for some time that I solve F2L differently than pretty much everyone, so I want to share my F2L method and get some inputs about this way of doing F2L. I also would want to know if this has ever been done before.

I called my method Sideways F2L. Sideways F2L starts after finishing a cross like a regular CFOP F2L, but instead of holding the cube with yellow on top, I would hold my cube sideways so the (white) cross is on the left and last layer (yellow) is on the right (reverse this if you're a leftie). I'll then be pairing and inserting using mostly just l/l', U/U', and R/R'.

Pairing is done intuitively, and is pretty simple as there are significantly less cases than with standard F2L as we can eliminate mirrors and backsides with l and R rotations.

There are two basic insertion algorithms, which I call Thumb Insert (U' R U) and the Pinkie Insert (U R' U') with l or R rotation to choose the side to insert to. Pretty much any pairs can be inserted efficiently to any positions with just those two insertion algorithms. There are also Sledgehammer variants for both Thumb and Pinkie pair inserts, which are useful for LL edge control.

Some advantages I noticed of Sideways F2L is that it is easy to learn intuitively while allowing high speed pairings and insertions. Since Sideways F2L uses l/l' instead of Y/Y', you never need to do two hands Y/Y' regrips, so you don't need a separate rotationless algorithms to avoid clunky Y/Y' regrips. Also, most of the work is done with your dominant hand so that's a plus if you have a strong preference on your handedness. Doing lookahead is also not too difficult, but it requires a somewhat different lookahead techniques than standard F2L.

I usually also do Sideways Cross, which transitions nicely to Sideways F2L, though I've never been that good at neither regular nor sideways cross. When finishing F2L, I usually insert in the last pair using fingertrick variants that leave the last layer on top, and then just do a standard last layer. Recently, I've also been incorporating edge control for inserting the last pair, which works very well because most Sideways Sledgehammer insertions have fingertrick variants which will also rotate the cube to leave the last layer on top.

I always average about 5-10 seconds faster using Sideways F2L than with regular F2L, even though I practice both roughly equally (my current average is 30s). So it works well for me so far, though since most of the very fast people are using standard F2L, I've also been wondering if I would benefit from focusing on just one way to do F2L rather than trying to practice both.
 
Last edited:
I am taking up a different approach with very promising results. The reverse approach is to try dozens or hundreds of the shortest, finger friendly move sequences, and then observe their effect on the cube. With this this ultra-fast 'set' of sequences, now design a method to use them.

That's what I've been doing as well in my solves. My most frequently used movesets are mostly composed of algorithms that have great ergonomic, easy to execute quickly (avoids F, D, B, and M unless another move had left the finger where it's easy to do those), allow good lookaheads, and somewhat TPS spammy. There are also a number of OLL/PLL cases that I prefer solving using two quick algorithms instead of one clunky ones.
 
Last edited:
I also would want to know if this has ever been done before.

Yes :D

I think this way of solving F2L is very interesting. I never liked y rotations and this replaces them with Lw moves, which is nice. Though, I've also heard lookahead can be harder and there is of course the mandatory z' after F2L is finished.
 
I never liked y rotations and this replaces them with Lw moves, which is nice.

Not just that; you also have to replace lefty F2L algs with either ⟨R,D⟩ algs (okay for three-move inserts, terrible otherwise), an x2 rotation (lolrotations), or do the z' early and continue with F2L on bottom (essentially negates the benefits of COL).

I always average about 5-10 seconds faster using Sideways F2L than with regular F2L, even though I practice both roughly equally (my current average is 30s). So it works well for me so far, though since most of the very fast people are using standard F2L, I've also been wondering if I would benefit from focusing on just one way to do F2L rather than trying to practice both.

As a cross-on-left person myself, I don't think cross-on-left is good in the long run. Looking ahead is complicated by that you can barely see the right face (unless you hold the cube at an awkward angle), which is an important source of information about F2L pieces, and to a lesser extent, that your right hand will often be partially obstructing your view of the front face.

If your goal is to get fast as quickly as possible, focus on the standard cross-on-bottom, imo.
 
I think that Roux is an excellent method, I just like to have fun with this kind of stuff :) Roux with Tyrannical Caterpillar could be pretty insane and even Roux with just CMLL/standard LSE is really good. But I do believe that LLOB is still a method with serious potential. But if I'm wrong, at least I had some fun with it and inspired some people to help progress cubing methods.

I think LLOB w/DFDBEO + ZBLL has excellent potential primarily by allowing CMLL recognition during DFDBEO, eliminating the huge problem in Roux of the pause during CMLL recognition. Tyrannical caterpillar may actually make the pause worse, as now you have to recognize CMLL and also track/identify the FR edge and decide which M/U variant to use. LLOB w/ZBLL tries to eliminate the CMLL recognition pause much like WaterRoux does, but in WaterRoux w/TLEG-1 it is still extremely difficult to see the corner solve in the inspection so LLOB takes a more realistic approach to eliminate this pause. I have made great progress modifying LMCF (which also tries to eliminate the CLL recognition pause), I have created a new variant where you use modified ELL's to solve the first block (after solving the corners) in one algorithm, which is turning out to be extremely fast since all the edges on the cube are free, so the algorithms are extremely short and super ergonomic, and this also fixes the ergonomics problem of LMCF since once the first block and corners are solved the rest of the solve is RrUM.
 
If a good prediction/recog style can be done based on CLL cases in ZBLL, it would be absolutetly insane. I will certainly take a look at it. RA-OB (LLOB with Ribbon Alpha) is really great too, so if I can't come up with something we certainly have something to call back on :)

That LMCF variant seems interesting, I'm a little confused on it though. Do you have to put the edges in a certain way before solving FB or is it mostely intuitive? Other than that, good luck and I hope to see how it turns out!

WaterRoux for me is a good idea in theory but it seems completely unrealistic for a majority of people. Sure the efficiency is nice, but lookahead is subpar and there are a lot of algs to learn with a very tedious last step. I may finish it someday, but as of now it doesn't strike me as good enough if a method to do so. Perhaps this new variant of LMCF will be the "next phase" of WaterRoux, I think I'll play around with LMCF and see what I come up with!
 
That LMCF variant seems interesting, I'm a little confused on it though. Do you have to put the edges in a certain way before solving FB or is it mostely intuitive? Other than that, good luck and I hope to see how it turns out!

The state of the cube with all corners solved, and no edges solved, is one of the most amazing states since you can move 3-4 edges in ultra short algorithms of 3-5 moves that are ridiculously fast and ergonomic (in fact, you literally look for super ergonomic 3-7 move sequences and just see what they do to the edges...) The big shift from previous LMCF is to try to accomplish a lot more during the transition phase, solving edges on the U layer before switching to R/L solving.

To make use of this requires a very effective method of edge recognition & lookahead. This is what I have been working on. The idea is after solving the corners, you are already looking ahead at the U layer edges. By the time the corner algorithm is finished you have already identified the U layer edge set up which will consist of some combination of edges destined for U, D, and 'junk' edges from the E-layer. The idea is to solve U layer edges on the U-layer, sometimes adding an E-layer edge destined for U, and sometimes pushing a top layer edge down to D and solving it (by doing a D/D' before hand to make sure it goes in the right place). The end effect is you can solve 2-4 edges almost instantly, you then do a z rotation and finish Waterman style RrUM. Furthermore, during the U-layer edge 'algorithm', you are also setting up the first E2L (ERL) pair. I'm still tweaking the system but it has huge promise and reduces the move count even more (due to the incredible efficiency of these 3-7 move sequences). The fastest case is when you finish the corners and find (by chance) that you have 3 U layer edges already in the U layer but unsolved (actually happens quite often). Now an extended ELL set allows you to solve them all at once, and the ELL-style algorithms are FAR shorter than normal ELL's because the whole rest of the cube is free to be broken up.

In addition I have been tweaking the algorithms & finger tricks for the Waterman L6E. Now that I actually know a significant fraction of this large set I am finding it truly useful and incredibly fast; you can never really be sure about how something works until you start using it in real, full speed solves.

Having said that, although this LMCF variant has promise, I honestly think that LLOB w/ZBLL is the best method developed thus far. In the LLOB thread I proposed the name ZBRoux = Zebra for it. However ZBRoux only shines with full ZBLL (I'm trying to convince Anthony Brooks to switch since he already knows ZBLL). Therefore ZBRoux, kind of like WaterRoux, is one of those methods that really only shine with a huge algorithm set.
 
I apologize for not having read the whole thread, so all of this may have been already suggested. I mentioned these ideas in the ZZ vs. CFOP etc. thread, but they probably fit better here. Has anybody tried:
Instead of EOLine going EO[2x2x3]Block? I can't even plan full EOLine in inspection, but if somebody was really good at ZZ it seems like it could combine the best of ZZ and Petrus.
Secondly, and this is truly half-baked: let's assume you do an EOBlock on the bottom left as described above, then build the easiest 1x2x3 block instead of the one that goes on the bottom. Would recognition be too difficult to finish the solve with ZBLL, AUF, ARF?
 
Back
Top