• Welcome to the Speedsolving.com, home of the web's largest puzzle community!
    You are currently viewing our forum as a guest which gives you limited access to join discussions and access our other features.

    Registration is fast, simple and absolutely free so please, join our community of 40,000+ people from around the world today!

    If you are already a member, simply login to hide this message and begin participating in the community!

JSDMv1 - Jay's Scramble Dependent Method (first version)

ottozing

Platinum Member
Joined
Aug 13, 2011
Messages
3,246
Location
Canberra, Australia
WCA
2012MCNE01
YouTube
Visit Channel
~

(I really wanted to call this method JDSM but I can't make the acronym make sense lmao. I also think this method needs a name with 2-3 syllables maximum but I don't want to just call this "McNeill" so consider JDSM a placeholder for now. If I don't see any possible names I like after a week I'll just suck it up and call it McNeill)

~

Motivation:

The past few weeks, I've been exploring the world of the Mehta method. All in all, I'm impressed with Yash Mehta's efforts (along with key community members in the Mehta discord), and I've been thinking a lot about ways to improve the method

During that time, I stumbled upon an idea that I think has some level of legitimacy as a new method. However, the context is key to its success, and I don't think anyone should be a pure "JSDM" user (especially considering what it literally stands for lol)

This method is intended to be used by CFOP solvers with very good inspection abilities, and the first few steps are very similar to Mehta

This method could potentially become a viable secondary method for both CFOP and Mehta, but I'll explain why I don't think Mehta mains will get the most mileage out of this in the long term

The aspect of method neutrality that gets overlooked:

Method neutrality has become a somewhat popular idea in recent history. I've even toyed with the idea myself, and I think there is a legitimate concept here. However, switching between CFOP and Roux, or CFOP and ZZ, or CFOP and Petrus etc has some major drawbacks which are worth acknowledging

(NOTE: All of what I'm about to say is from a CN or Color Neutral solving perspective. If you only solve on 1-2 colors, some of these drawbacks won't necessarily apply)

When it comes to scrambles that are "good" for Roux or Petrus, they aren't actually bad scrambles for CFOP as long as you have good intuitive block building skills, and they might even be better with CFOP if you look at the further context of the scramble. For example, a scramble with a solved 2x2x2 might have a 2 move Xcross, or a scramble with a solved 1x2x3 might have a 4 move rUgen XXcross etc

When it comes to scrambles that are "good" for ZZ, the issue is more to do with the fact that planning EO in inspection is very different, and much more difficult, than directly solving pieces. You don't even need to know Roux or Petrus to know when you have an easy block, but with ZZ you absolutely need to spend a lot of time training your ability to see EO in inspection (and also solve it efficiently)

The final main issue I see with method neutrality is that any good speedcubing method will have its final few steps be largely algorithmic. The 100's-1,000's of tricks Roux solvers know for CMLL+LSE, or CFOP solvers know for LSLL, or ZZ solvers know for OLS etc are something that take years to hone. I see no good reason to spend years honing skills like these for methods that aren't your main, especially when you're always going to be better with one method

If you're "method neutral" between CFOP and Roux, but CFOP is your main and you use it 75% of the time, over the course of an infinite number of solves, you're absolutely going to become more and more imbalanced with your CFOP skills relative to your Roux skills

In my opinion, the best approach to method neutrality is to use scramble dependent methods that can take you to the "algorithmic state" of your main method.

If you're a ZZ main for example, Petrus is a good secondary method to pick up since easy blocks will be immediate to spot in inspection. This way, you waste minimal time choosing what method to do in inspection while reaching your "algorithmic state" after completing the EO step of Petrus. Since ZZ is best done with EOCross, you would ideally want to use Petrus if you know that you can inspect all the way to 2x2x3+EO+the last cross edge

Even if you can't quite plan that far however, you could potentially mitigate the annoying steps of Petrus by theory crafting compromises if you can only plan 2x2x3. One possible idea would be algs for EO+make sure the DF edge doesn't end up being an F2L edge (eliminates the blind spot and solves the final cross edge 1/5th of the time)

JSDMv1:

Now let's talk about the method idea I have

With CFOP, once you reach LSLL or solve a cross+3 pairs, as an advanced solver you can basically turn your brain off and just do all the algorithmic mojo you've built up from years of experience with CFOP

Even though the F2L step as a whole is statistically very likely to be algorithmic, it's not 100% since you can occasionally get pairs where both pieces are unsolved in different slots and have no better pair to solve (very rare and also somewhat preventable, but it happens to the best of us)

With the method I have in mind, I wanted it to be something that can reach not only LSLL, but LSLL in a pseudo state (D layer misaligned)

I also wanted it to be something that avoids ugly cases like the F2L thing I mentioned, and also something that avoids the headaches that can come with PSF2L solving (which I'll quickly go over now)

PSF2L or Pseudo F2L is a concept largely popularized by Tymon Kolasiński. For CFOP solvers, if you could plan Cross+3 pseudo pairs in inspection every time, CFOP mains wouldn't really need a secondary method

However, doing this with 15 seconds of inspection, even at home with no pressure, hasn't been 100% achieved yet. In a competition setting, I don't quite see this happening

Another issue with PSF2L is that trying to solve all 4 of your F2L pairs in a Pseudo state forces you to look at the D layer constantly. Even though 16 options for an F2L pair is better than 4 in theory, in practice it's just too much time to spend in a Pseudo state

Here are the steps for JSDMv1 that I currently have in mind. I believe that this is the most practical version of the method out of the dozens of variations I've considered:

Step 1 - Mehta FB (1x2x3 on the D layer)
Step 2 - HB+last cross edge or HBCE (similar to Mehta's 3QB but you solve two E layer edges and a cross edge instead of 3 E layer edges)

[These two steps combined are essentially PSF2L-2 or Pseudo F2L minus 2]

Step 3 - Do a Pseudo F2L pair

[This step gets you all the way to Pseudo LSLL]

Step 4 - LSLL

Some additional notes & thoughts on each step:

(All move count stats will be under the assumption that you're Color Neutral)

FB is a bit more efficient than doing a cross. See here for more info because everything that Yash says applies to this method equally. 5 moves on average, 7 at worst, and good move groups

I think on some scrambles with an easy 1x2x3, doing a Roux style FB into an rUgen XXcross will be better (though I imagine there are situations where this takes way too many moves to justify it being its own scramble dependent method)

HBCE move count statistics don't currently exist and I'm too incompetent to calculate it with HARCS, but the step can be entirely solved with R and u moves. SRgen and fUgen triggers can also be used in similar situations. Since the same number of pieces are being solved as Mehta's 2nd step, I imagine the move count is either the same or 1 worse on average. For now I'm going to set this as 5 moves on average and 7 at worst, but once stats get ran and verified I'll update this

Note that if this method were being used for every scramble, I don't think seeing an optimal FB+HBCE every time would be realistic. However, under the assumption that the cuber only uses this method for scrambles where the first two steps are very easy to see, I don't think it's as unreasonable to assume the cuber spots the optimal solution or within a move of it (could lead to better fingertricks being a move or two less efficient)

Lastly, I think that for HBCE you don't want to solve diagonal E layer edges very often, if ever (given that it's a scramble dependent method). From a look ahead standpoint, having diagonally solved F2L pairs as your final two is generally very annoying and not nice to deal with. Having diagonally unsolved edges alongside adjacently solved corners feels even worse to me on an intuitive level, doubly so in a Pseudo state

If you accidentally solve 1 of the 4 remaining F2L pieces after the first 2 steps, or you can solve an additional F2L piece very easily, you have a very high chance to get to Pseudo F2L-1 in 3-4 moves (especially if the piece you solve is a corner)

For FB and the last cross edge specifically, since we're going for a CFOP style finish, there will be some scrambles where it's significantly easier to build that first block with one of the three edges missing, and instead replaced with an LL edge (this way we still fill in the blind spot). The same could also be applied to the HBCE step. This edge will only ever take 4 moves to solve if it's left to the very end of F2L using something MU, SU, or SRgen, and can be done just as efficiently earlier in the solve with the added possibility of an rU cross edge insert

The final idea I have is a very weird one & requires ridiculous foresight & spatial awareness. To follow along, take a scrambled 3x3 and solve a cross, an E layer belt, and two adjacent D layer corners on the left

Now on this cube, do r' E2 r

A possible FB alternative build is something like this with the opposite center

More interestingly, if you can place the cross edge in FL in this specific orientation, and you can also place 2/4 E layer edges like they are on the right (UR and DR solvable with R, and FR/BR touching the F/B centers), then r' E2 r gives you PSF2L-2

If you're doing this for PSF2L-2 and you want to avoid diag edges, then you have to be a little careful with which edges you place on the right

You can also do this to get PSF2L-1 by having a D layer corner solved and placing 3/4 E layer edges like they are on the right, since r' E2 r preserves the corners on a solved cube

It's also possible to ignore setting up the cross edge and just make sure there's an LL edge in FL to make this technique easier

r E2 r' works the same way but the edge pattern is mirrored

r E/E' r' and r' E/E' r can be theory crafted the same way but holy hell the recognition gets awful

Example Solve:

F2 U2 R2 D' F2 L2 U L2 B2 R2 B2 D2 L' F' R B2 F2 U2 L' D2 U'

x
r' f' R U' L // FB+1
u' R' F R2 F' // HBCE
u' U R U R' U2 R' U' R // PSF2L-1
U' F' U' F // PSF2L
D2 U' R U R' U R U' R' U R U' R' U R' U' R2 U' R' U R' U R U // ZBLL
 
Joined
Apr 1, 2020
Messages
54
(I really wanted to call this method JDSM but I can't make the acronym make sense lmao. I also think this method needs a name with 2-3 syllables maximum but I don't want to just call this "McNeill" so consider JDSM a placeholder for now. If I don't see any possible names I like after a week I'll just suck it up and call it McNeill)
A possible name would be ACFOP
Alternative CFOP
 

ottozing

Platinum Member
Joined
Aug 13, 2011
Messages
3,246
Location
Canberra, Australia
WCA
2012MCNE01
YouTube
Visit Channel
After some thinking, I've decided on BFOP as the name (similar to @Rouxvolutionist's suggestion)

I like this because it communicates the most common steps of the method (Block, F2L, OLL & PLL) and simultaneously communicates the fact that it's a CFOP alternative through a single letter change (it also has nice meme potential because 🅱️)

If a mod wants to update the thread title to something like "BFOP method proposal" I would be greatly appreciative <3

Also, I've created a Discord server for anyone who wants to help develop the method or just hang out with other people interested in the method - https://discord.gg/B5MP263qyN
 

xsac

Member
Joined
Dec 9, 2020
Messages
16
my solve with bfop

R2 B' D' F B L D2 F' D F2 L2 F2 D2 F2 L F2 B2 U2 R L B2

z2
F' U2 L F2 u F R' F' // FB + HBCE
U2 R U2 R2 U F R F' // 3rd pair + eo
U' R' U R U' R' U R // LS
U R D' R' U R2 D' r' D2 r' U' r2 D' R2 U2 D' // ZBLL
 

effperm

Member
Joined
Apr 17, 2020
Messages
28
Location
philippines
YouTube
Visit Channel
ooh this method might be okay
it gets rid of the learn 1400 something algs in mehta
fb still seems a bit awkward and it still comes off as a worse way to do petrus
but with fb+1 and stuff techniques maybe
 
Last edited:
Joined
Apr 1, 2020
Messages
54
You only have
ooh this method might be okay
it gets rid of the learn 1400 something algs in mehta
fb still seems a bit awkward and it still comes off as a worse way to do petrus
but with fb+1 and stuff techniques maybe
You only have to learn 130 algs or 900. Not 1400
 
Thread starter Similar threads Forum Replies Date
F Puzzle Video Gallery 4
Top