• 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 35,000+ people from around the world today!

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

CFOP-Breaker? Mehta method

trangium

Member
Joined
Jul 24, 2019
Messages
50
WCA
2019TRAN10
What if you did L5EP with the DR edge instead of the DF edge? That way you could have <R, U> L5EP algs instead of <M, U>, and 6CP algorithms would improve. I expected this to only be viable for one-handed, but many of the algs are so good that they outperform the <M, U> algs for even two-handed. Both P perms, both O perms, and the Qa perm are especially good compared to their respective <M, U> algs. Also consider that switching from <R2, U, D> turning to <M, U> turning requires a grip shift.

Here's a full list of L5EP algs with the DR edge unsolved.
CasePictureAlgNotes
I
1/15
S' U2 SLeft index leaves the cube during U2. Use a left index push for S.
Ua
1/15
R U R' U R' U' R2 U' R' U R' U RThis alg is relatively long but it flows well. To memorize, watch the pairs move around.
Ub
1/15
R' U R' U' R3 U' R' U R U R2Start with the right thumb on top. The R3 is intentional and is done as a rolling R3. It's there to prevent a regrip.
Pa
1/15
R U R' U' R' U' R' U RVery easy to memorize: Take out a pair, R', insert the pair. Make sure not to regrip after the third R'.
Pb
1/15
R' U' R U R U R U' R'Mirror of Pa.
H
1/60
M2 U M2 U2 M2 U M2
Z
1/30
M2 U2 M U M2 U M2 U M
Ca
1/15
R U R' U R' U' R2 U' R' U2 R U R U' R'The first 9 moves are the same as the Ua perm. A long alg, but it flows well. Do the 3rd to last move as a rolled R.
Cb
1/15
F2 U' r2 F2 R2 U' F2 r2This alg is hard to master but can be very fast. Start with right thumb on bottom to do F2. Make sure your left thumb is not in the way of doing an r2. The second F2 is done like a right hand D2.
D
1/15
R U2 R2 U2 R2 U2 R
Wa
1/15
D' M' U2 M U M' U2 MThere is a 13-move <R, U> alg for the W perms, but it's slower than just doing D' then <M, U>.
Wb
1/15
D' M' U2 M U' M' U2 MSame as Wa except for the U' instead of the U.
Qa
1/15
R U R' U R' U' R' U' R2 U2 RThe first 6 moves are the same as the Ua perm. Make sure not to regrip after the third R'.
Qb
1/15
R' U' R U' R U R U R2 U2 R'Mirror of Qa. The (U R U) is tricky: Do the first U as an OH flick and the second as a left index push. You can reset your right hand while doing the U2 to make the R' easier.
Oa
1/15
R' U' R U' R U R U' R2 U2 RThe first 7 moves are the same as the Qb perm.
Ob
1/15
R U R' U R' U' R' U R2 U2 R'The first 7 moves are the same as the Qa perm. Make sure not to regrip after the third R'.
Also, is Devagio still practicing this method? If so, what progress has he made?
 

L1meDaBestest

Member
Joined
Oct 21, 2020
Messages
2
The original post has been updated.

Since the major steps have been frozen, interested people can begin transitioning to the method via the beginner variation. Do let it be known if you’re trying it out, whether you intend to continue to the full version, etc; this will motivate others to try it out and subsequently the method will develop quicker.

For those who wish to promote it, you could make YouTube tutorials or talk to YouTubers about it; since the method is fairly straightforward, it shouldn’t be difficult for anyone to make a tutorial.
This method looks incredible and I would really like to see how good this can get with fast recognition speeds.

I’ll learn the full method because with only 200 algs and that predicted move count you have definitely sold me on this (also the other reasons).

I average 10 with CFOP for reference and when I learn the whole method I’ll start posting progress if you would be interested.

I’m so glad I saw this and I think it will be really fun!
 

Devagio

Member
Joined
Apr 21, 2020
Messages
216
What if you did L5EP with the DR edge instead of the DF edge? That way you could have <R, U> L5EP algs instead of <M, U>, and 6CP algorithms would improve. I expected this to only be viable for one-handed, but many of the algs are so good that they outperform the <M, U> algs for even two-handed. Both P perms, both O perms, and the Qa perm are especially good compared to their respective <M, U> algs. Also consider that switching from <R2, U, D> turning to <M, U> turning requires a grip shift.

Here's a full list of L5EP algs with the DR edge unsolved.
CasePictureAlgNotes
I
1/15
S' U2 SLeft index leaves the cube during U2. Use a left index push for S.
Ua
1/15
R U R' U R' U' R2 U' R' U R' U RThis alg is relatively long but it flows well. To memorize, watch the pairs move around.
Ub
1/15
R' U R' U' R3 U' R' U R U R2Start with the right thumb on top. The R3 is intentional and is done as a rolling R3. It's there to prevent a regrip.
Pa
1/15
R U R' U' R' U' R' U RVery easy to memorize: Take out a pair, R', insert the pair. Make sure not to regrip after the third R'.
Pb
1/15
R' U' R U R U R U' R'Mirror of Pa.
H
1/60
M2 U M2 U2 M2 U M2
Z
1/30
M2 U2 M U M2 U M2 U M
Ca
1/15
R U R' U R' U' R2 U' R' U2 R U R U' R'The first 9 moves are the same as the Ua perm. A long alg, but it flows well. Do the 3rd to last move as a rolled R.
Cb
1/15
F2 U' r2 F2 R2 U' F2 r2This alg is hard to master but can be very fast. Start with right thumb on bottom to do F2. Make sure your left thumb is not in the way of doing an r2. The second F2 is done like a right hand D2.
D
1/15
R U2 R2 U2 R2 U2 R
Wa
1/15
D' M' U2 M U M' U2 MThere is a 13-move <R, U> alg for the W perms, but it's slower than just doing D' then <M, U>.
Wb
1/15
D' M' U2 M U' M' U2 MSame as Wa except for the U' instead of the U.
Qa
1/15
R U R' U R' U' R' U' R2 U2 RThe first 6 moves are the same as the Ua perm. Make sure not to regrip after the third R'.
Qb
1/15
R' U' R U' R U R U R2 U2 R'Mirror of Qa. The (U R U) is tricky: Do the first U as an OH flick and the second as a left index push. You can reset your right hand while doing the U2 to make the R' easier.
Oa
1/15
R' U' R U' R U R U' R2 U2 RThe first 7 moves are the same as the Qb perm.
Ob
1/15
R U R' U R' U' R' U R2 U2 R'The first 7 moves are the same as the Qa perm. Make sure not to regrip after the third R'.
Also, is Devagio still practicing this method? If so, what progress has he made?
So I did consider this possibility back when I had very little experience with the method, but discarded it because
1. I thought maybe at the top level people would transition from L5EP to L5E (I don’t anymore, this was around when only the first page of this thread existed) so MU is much more natural.
2. I just feel MU EPLL are better than RU EPLL except U perm, and the L5EP MU algorithms were so good that I (possibly incorrectly) reasoned MU L5EP is better than RU L5EP.
Later on, I just went on practicing with the method to get faster times than actively thinking about developing it.
But now that you mention it, yes, one of the biggest pauses I have is going from RUD grip to MU grip, and if the RU L5EP is this amazing, it’s definitely worth looking into.
While I don’t think 6CP on average will get any better, but yes, the faster algs will get even faster. Plus there will not be any need of learning two L5EP sets (DF and DB) so that’s less algs and more importantly a much cleaner way to do things.

By the way, I’m cubing very little nowadays, but any amount of cubing I do, it is to get faster with this method. I currently average 16-19, very inconsistent because 6CO can be a pain if we don’t know the algs.

I personally will continue using the MU version of the method because I really don’t want to drill more algorithms for a few more months. Once the pandemic subsides a little (or schools get wise enough to not increase workload to “catch up” -_-) I’ll most likely shift.

To anyone using the method, I guess the RU L5EP looks objectively better, so give it a shot. If I ever get to finally making the full website for the method, I’ll most likely keep the final step RU gen if it gives more promising results, plus to help with OH of course.
This method looks incredible and I would really like to see how good this can get with fast recognition speeds.

I’ll learn the full method because with only 200 algs and that predicted move count you have definitely sold me on this (also the other reasons).

I average 10 with CFOP for reference and when I learn the whole method I’ll start posting progress if you would be interested.

I’m so glad I saw this and I think it will be really fun!
Hey! Welcome aboard! We would all really appreciate sharing or any progress or ideas regarding the method, and it’s really cool that you’re considering trying it out. It most definitely is fun.
I’ve been getting quite a few emails and DMs regarding this off late, so I’m considering making a discord group to talk about the method, take help, share progress, let non-users join to see if it’s fun, etc. I imagine it wouldn’t be to hard to set it up so I’ll do it this weekend.
 
Last edited:

Devagio

Member
Joined
Apr 21, 2020
Messages
216
I’ve been getting quite a few emails and DMs regarding this off late, so I’m considering making a discord group to talk about the method, take help, share progress, let non-users join to see if it’s fun, etc. I imagine it wouldn’t be to hard to set it up so I’ll do it this weekend.
Click here to join the discord!
Everyone is welcome, especially those who haven't gotten around to try the method out but are interested to explore something new.
 

L1meDaBestest

Member
Joined
Oct 21, 2020
Messages
2
Click here to join the discord!
Everyone is welcome, especially those who haven't gotten around to try the method out but are interested to explore something new.
I think a Discord is a great idea! It will be very good for talking about the method/progress.

Also I think the link you sent may have been temporary (I don’t really know how it works) because it doesn’t work and says that it has expired.
 

TheSlykrCubr

Member
Joined
Aug 13, 2020
Messages
643
Location
37.8270° N, 122.4230° W
YouTube
Visit Channel
I think a Discord is a great idea! It will be very good for talking about the method/progress.

Also I think the link you sent may have been temporary (I don’t really know how it works) because it doesn’t work and says that it has expired.

here's a permanent one
 

Andreas5204

Member
Joined
Jul 29, 2020
Messages
8
Good news! Mehta now has a complete algorithm sheet, with visuals!
Created by Andreas5204 and Cuberstache, with help from Devagio.
 

Nir1213

Member
Joined
Aug 31, 2020
Messages
807
Location
https://www.youtube.com/watch?v=FveF-we6lcE
Good news! Mehta now has a complete algorithm sheet, with visuals!
Created by Andreas5204 and Cuberstache, with help from Devagio.
where all the algs?
edit:nvm found them
 

Devagio

Member
Joined
Apr 21, 2020
Messages
216
Good news! Mehta now has a complete algorithm sheet, with visuals!
Created by Andreas5204 and Cuberstache, with help from Devagio.
So here there's a slightly updated beginners' version of the method which is easier to learn and in general faster to solve the cube with. I'll break it down here so it's easier to understand the beginners' section of the spread sheet.

1. FB
2a. Entire belt
2b. EO (algs in sheet)
3. 2-look 6CO
4. 2-look 6CP
5. 2-look L5EP

FB: Solve a 1x2x3 block in bottom left such that 6 of the stickers of the block are on the D face. You could just solve 3 cross pieces followed by two corners, but I'd recommend learn direct blockbuilding ASAP. Try to be CN from the start.
Belt: Solve the 4 E slice edges. Note, you have a free R layer, so you will not have to worry about restoring anything in general, so be efficient here.
EO: Orient the last 5 edges using one of the 5 algorithms in the sheet.
6CO-1: The only difficult-to-understand concept. The idea is to have DFR and DBR stickers to be the top/bottom colour. There are 4 cases. If there are two adjacent corners in the lop layer which are oriented, simply use R2 U2 R2 to put them in D layer and move to the next step. Else, there will always exist a pair of opposite corners in the top layer that are unoriented (this can happen in 3 ways: N, Z or L). Simply use the corresponding algorithm given in the sheet.
6CO-2: Basically OCLL but with better algs.
6CP-1: Use an R2 U R2 type insert to solve the DBR corner.
6CP-2: Solve all of the last 5 corners using one of 8 algorithms in the sheet. If DFR is solved, it is either J or Y perm, else the case is recognised by looking at the top layer corners in F face and R face.
L5EP-1: Solve the final D layer edge using an M' U2 M type insert.
L5EP-2: Basically EPLL but with potentially an ADF as well.

Will put an example solve in the example solves thread.

This beginner version for single FB averages well under 60 moves. Moreover, 7 out of the 9 steps are algorithmic, i.e. no scope of inefficiency there; hence even an inefficient solver with 3 - 3.5 TPS would be sub-20 theoretically.

1604154338615.png
 
Last edited:

Devagio

Member
Joined
Apr 21, 2020
Messages
216
With most of the development now happening on discord (in a greatly accelerated pace than here previously), I will give an update of the major developments for the benefit of those not on discord.

The general structure of the method is:
1. First block (FB)
2. 3-Quarters belt (3QB)
3. last edge + edge orientation (EOLE) [56 cases]
4. orienting 6 remaining corners (6CO) [72 cases]
5. permuting 6 remaining corners (6CP) [48 cases]
6. permuting 5 remaining edges (L5EP) [16 cases]

The first 3 steps are collectively called EO-ledge.

The recognition systems for 6CO and 6CP have been formulated.

There has been a great deal of emphasis on option select due to the algorithmic nature of the method. The three major options to finish the solve after EO-ledge are:

a. 6CO + 6CP + L5EP: The original default route which will be the most efficient 3-step path most of the times.

b. 6CO + DR bar + PLL: Sometimes it takes only 3-4 moves to make and insert the DFR and DBR corners, as well as the DR edge. There are about 40 cases for DR-bar the algorithms for which are already generated. If the 6CP is not great, or the DR bar is very obvious, we could do this instead and follow it up by one of 21 PLL cases. This is not ideal to use as default due to PLL algorithms being much worse in recognition and execution than L5EP algs, and some DR-bar algs being not great.

c. DFR-DBR + COLL + L5EP: Sometimes it may be much easier to solve the DFR and DBR corners (3-7 moves) instead of recognising and doing 6CO. This has ~100 algorithms, many of which are not great, but can be used if the 6CO case is not great or the D-corners solution is very obvious. This can be followed up with COLL leaving L5EP.

There are many other option select possibilities (most notably 6CO + edges + corners which is extremely efficient but almost never worth it recognition-wise) but the above seem more than sufficient to always have a great finish to the solve. There are some trick algorithms sets (like 4 corners only, setups to PLLs, etc.) which can pop up a couple times in an ao100 and potentially help us skip a step.

For L5EP, I personally believe it is best to learn it from right, front and back (4+ 12 x 3 short algs) to save a move on ADF at the end. The current consensus is if only one of the sets is to be learned, it has to be L5EP on right.

There are a lot of other developments (like the OH variation, the big-cube variation, the squan variation, Mehta-1LLL, etc) because they aren't as relevant here.

The current plan is to work on developing EO-ledge to make it more ergonomic and flexible.
Many of us have started getting good times with the method, and the example solves forum and reconstructions forum have progressively great solves with the method if you haven't checked them out yet.

So many people have combined efforts into developing, compiling and testing the above, that it would be extremely difficult for me to provide appropriate credit to everyone, which is why I highly encourage to check out their work on the discord server, where you will find all the algorithms (except DFR-DBR) and respective recog systems I mentioned above.

If you'd like to contribute towards the method in any way, or consider using it, join the discord server using the link in one of the previous posts.
 

Devagio

Member
Joined
Apr 21, 2020
Messages
216
An example solve using "Meta-Mehta"

U' L' F2 D2 L2 F2 R2 U' B2 R2 L' B U B F2 L' F R2

y2 // inspection
M' B2 U' L B2 // FB+1e (5/5)
u' R U R u R2 // +2e (6/11)
F' U F2 R F' R' // EOLE (6/17)
R' U' R U' R' U2 R' U' R2 // DR-block (9/26)
R2 D R' U2 R D' R' U' R' U R U2 R' // ZBLL (13/39)
E' // ABF (1/40)

This is basically the ZB-equivalent of Mehta, where after EO-ledge, we solve the DR-block with 1 alg (2-gen 9-mover, ~400 cases), and end up with ZBLL (~500 cases). HARCS gives sub-40 movecount. Better names coming soon.

Also note the way EO-ledge is done - FB + 1 belt edge is planned in inspection, and the 3QB is completed by solving 2 more edges simultaneously (will have under 90 cases, so just like F2L this can be both intuitive and algorithmic). This will standardise FB+3QB enough for us to minimize usage of u moves and facilitate lookahead. This is still under experimentation though.
 
Last edited:

Nir1213

Member
Joined
Aug 31, 2020
Messages
807
Location
https://www.youtube.com/watch?v=FveF-we6lcE
An example solve using "Meta-Mehta"

U' L' F2 D2 L2 F2 R2 U' B2 R2 L' B U B F2 L' F R2

y2 // inspection
M' B2 U' L B2 // FB+1e (5/5)
u' R U R u R2 // +2e (6/11)
F' U F2 R F' R' // EOLE (6/17)
R' U' R U' R' U2 R' U' R2 // DR-block (9/26)
R2 D R' U2 R D' R' U' R' U R U2 R' // ZBLL (13/39)
E' // ABF (1/40)

This is basically the ZB-equivalent of Mehta, where after EO-ledge, we solve the DR-block with 1 alg (2-gen 9-mover, ~400 cases), and end up with ZBLL (~500 cases). HARCS gives sub-40 movecount. Better names coming soon.

Also note the way EO-ledge is done - FB + 1 belt edge is planned in inspection, and the 3QB is completed by solving 2 more edges simultaneously (will have under 90 cases, so just like F2L this can be both intuitive and algorithmic). This will standardise FB+3QB enough for us to minimize usage of u moves and facilitate lookahead. This is still under experimentation though.
im curious, its sub-40 movecount?? Thats even lower than roux and zz, and is it still as/better than CFOP??? the ZB equivalent of metha might be one of the most fastest and/or efficient methods for speedsolving.
 

LukasCubes

Member
Joined
Aug 10, 2020
Messages
320
Location
West Virginia
An example solve using "Meta-Mehta"

U' L' F2 D2 L2 F2 R2 U' B2 R2 L' B U B F2 L' F R2

y2 // inspection
M' B2 U' L B2 // FB+1e (5/5)
u' R U R u R2 // +2e (6/11)
F' U F2 R F' R' // EOLE (6/17)
R' U' R U' R' U2 R' U' R2 // DR-block (9/26)
R2 D R' U2 R D' R' U' R' U R U2 R' // ZBLL (13/39)
E' // ABF (1/40)

This is basically the ZB-equivalent of Mehta, where after EO-ledge, we solve the DR-block with 1 alg (2-gen 9-mover, ~400 cases), and end up with ZBLL (~500 cases). HARCS gives sub-40 movecount. Better names coming soon.

Also note the way EO-ledge is done - FB + 1 belt edge is planned in inspection, and the 3QB is completed by solving 2 more edges simultaneously (will have under 90 cases, so just like F2L this can be both intuitive and algorithmic). This will standardise FB+3QB enough for us to minimize usage of u moves and facilitate lookahead. This is still under experimentation though.
im gonna learn mehta-mehta or whatever this is called.

Edit: is there any youtube wideo or whatever this method is in?

Another edit: 37 moves. 9.25TPS to get a sub-4 on this solve. I average 3-4 TPS so this for me would probably be a sub-13 if I knew all this
 
Last edited:
Top