• 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!

Another Scramble Dependent Method (SDM) for CFOP

ottozing

Platinum Member
Joined
Aug 13, 2011
Messages
3,260
Location
Canberra, Australia
WCA
2012MCNE01
YouTube
Visit Channel
Recently I posted an idea for a CFOP based Scramble Dependent Method. I'd recommend at least reading the 2nd point on "the aspect of method neutrality that gets overlooked" since that section will be relevant here

Honestly, I already think the initial idea I came up with is mostly a bad idea. Any scramble that has easy blocks, whether it's a 1x2x3 or 2x2x2 etc really isn't that hard to deal with using CFOP so long as you have some level of experience with efficient block building & piece placement

Maybe the idea I posted (along with updates in the BFOP server) will eventually become useful if PSF2L recognition gets improved, but even if that happens, I don't see why anything related to blockbuilding needs to be categorized as a secondary method


As such, I think it makes more sense to find approaches that deal with patterns that are easy to spot, but at the same time patterns that aren't necessarily easy to deal with using blockbuilding or cross solving


Which brings me to my new idea that I believe is less trash (Name TBD). I can already see a few downsides, but I think that they're mitigated so long as the cuber is using the method as a scramble dependent method

Here are the general steps

Step 1 - Cross + 1 solved corner, one LL corner inserted & oriented in an adjacent slot (4 options), and two F2L edges an R2 U2 R2 U2 R2 away from solved within the same two slots (an easy pattern to recognize if you set it up, but not something you can deal with intuitively the same way you can deal with easy blocks for other 2H methods)
Step 2 - Last two pairs
Step 3 - OLL
Step 4 - Finish F2L with R2 U R2 U' R2 or D R2 U' R2 U R2 D'
Step 5 - PLL

There are some interesting properties that you can take advantage of by following the first two steps which I'll get into in a moment, but first I want to talk about the drawbacks

Firstly, if you're a right hand dominant solver and want to avoid rotating unnecessarily, you're going to have to solve at least one pair on the left. Basically, there's a decent chance that you're going to rotate more than you want to, so for this SDM to be worthwhile you need to account for potential added rotations and figure out if it's justified

Secondly, you're adding between 5-7 HTM turns or 8-10 QTM turns, so again, you need to be very careful if you're going to choose to go for this in a solve

anyway, with that out of the way, here's some cool stuff you can do with this method without learning a bunch of algs that wouldn't help a normal CFOP solve

Cool stuff section

Firstly, by switching between R2 U R2 U' R2 and D R2 U' R2 U R2 D', you can always avoid diag PLL and will get EPLL's much more often, meaning you can always do the best OLL in terms of speed without worrying about getting screwed by an E perm. Not bad

Secondly, another way you can solve this unique F2L multislot is by using R F R2' F' R' which flips the UF and UR edges. If you notice before OLL that you have two adjacent flipped edges between the oriented F2L corner, you can do this instead and get ZBLL (the alg itself also preserves a lot of LL blocks which can absolutely help with recognition). By knowing ZBLL, this SDM becomes a little bit better

Finally, by switching between R2 U R2 U' R2, D R2 U' R2 U R2 D', and R F R2' F' R' before OLL, you can always preserve any 1x2x2 in the U layer and get a Tripod LL case. By knowing Tripod LL, this SDM becomes even better than before (this can also be applied to preserving any 1x1x2 pair, which would make this idea even better if you know 1LLL for any case with a solved 1x1x2 pair)

The odds of the latter two outcomes aren't particularly high assuming you don't try and influence the F2L corner during LS, but by using R U' R' and R U2 R' (which a lot of LS cases naturally reduce to), you can force an oriented corner roughly half of the time (7/12 cases) . If you also include R' F R F', you can orient the corner 2/3rds of the time (8/12)

Knowing good OLS tricks will also increase your overall odds of getting the F2L corner oriented without wasting time. The best example I can think of are the F2L cases with the edge solved and the corner twisted in place. By using either R U R' U' R U2 R' U' R U R' + inverse or R U' R' U' R U R' U' R U2 R' + inverse, you can always orient the F2L corner using AUF and nothing else

Another simple extension that doesn't demand you to learn a bunch of algs for an obscure algset you're never going to use, is using R S' U2 S R' and D R' S' U2 S R D' to complete F2L if you don't like your OLL case. This to me feels pretty trash in all honesty, especially since after OLL you can always avoid diag and get disproportionate amounts of EPLL, but hey, it's an idea I guess lmao (it would also never lead to an OLL skip using this method which is the other reason I don't like it)

With that all outlined, feel free to discuss, & also share some scrambles where you think this method might be good (I tried already but everything was too easy for CFOP or blockbuilding lol)
 

kubesolver

Premium Member
Joined
Nov 15, 2019
Messages
318
So this is a special case of "don't solve a cross perfectly and fix it later with a short sequence" similar to solving a cross off by M2 U2 M2. (which is less useful than what you propose because it doesn't do anything to the corners)

If you do it you have an advantage of:
- you can finish this cross in two different D rotations, which gives you more options for first pair planning
- sometimes you can fix the cross during F2L to get a free pair
- sometimes you can fix it before OLL to get a better case
- you can influence PLL by choosing which pair of opposite edges to switch

This probably can be explored in any situation where the cross isn't solved, but the way it's unsolved doesn't affect solving later steps. IIt can be easly fixed - preferably by at different steps by different sequences that do something else to remaining unsolved pieces.

I think the best case scenarios are those where not solving a cross immediately allows you to see far into the solve in inspection (e.g. if the cube is one move away from being off by M2 U2 M2 and after that there two easy pairs set-up)/
 
Last edited:

abunickabhi

Member
Joined
Jan 9, 2014
Messages
2,955
Location
Yo
WCA
2013GHOD01
YouTube
Visit Channel
Recently I posted an idea for a CFOP based Scramble Dependent Method. I'd recommend at least reading the 2nd point on "the aspect of method neutrality that gets overlooked" since that section will be relevant here

Honestly, I already think the initial idea I came up with is mostly a bad idea. Any scramble that has easy blocks, whether it's a 1x2x3 or 2x2x2 etc really isn't that hard to deal with using CFOP so long as you have some level of experience with efficient block building & piece placement

Maybe the idea I posted (along with updates in the BFOP server) will eventually become useful if PSF2L recognition gets improved, but even if that happens, I don't see why anything related to blockbuilding needs to be categorized as a secondary method


As such, I think it makes more sense to find approaches that deal with patterns that are easy to spot, but at the same time patterns that aren't necessarily easy to deal with using blockbuilding or cross solving


Which brings me to my new idea that I believe is less trash (Name TBD). I can already see a few downsides, but I think that they're mitigated so long as the cuber is using the method as a scramble dependent method

Here are the general steps

Step 1 - Cross + 1 solved corner, one LL corner inserted & oriented in an adjacent slot (4 options), and two F2L edges an R2 U2 R2 U2 R2 away from solved within the same two slots (an easy pattern to recognize if you set it up, but not something you can deal with intuitively the same way you can deal with easy blocks for other 2H methods)
Step 2 - Last two pairs
Step 3 - OLL
Step 4 - Finish F2L with R2 U R2 U' R2 or D R2 U' R2 U R2 D'
Step 5 - PLL

There are some interesting properties that you can take advantage of by following the first two steps which I'll get into in a moment, but first I want to talk about the drawbacks

Firstly, if you're a right hand dominant solver and want to avoid rotating unnecessarily, you're going to have to solve at least one pair on the left. Basically, there's a decent chance that you're going to rotate more than you want to, so for this SDM to be worthwhile you need to account for potential added rotations and figure out if it's justified

Secondly, you're adding between 5-7 HTM turns or 8-10 QTM turns, so again, you need to be very careful if you're going to choose to go for this in a solve

anyway, with that out of the way, here's some cool stuff you can do with this method without learning a bunch of algs that wouldn't help a normal CFOP solve

Cool stuff section

Firstly, by switching between R2 U R2 U' R2 and D R2 U' R2 U R2 D', you can always avoid diag PLL and will get EPLL's much more often, meaning you can always do the best OLL in terms of speed without worrying about getting screwed by an E perm. Not bad

Secondly, another way you can solve this unique F2L multislot is by using R F R2' F' R' which flips the UF and UR edges. If you notice before OLL that you have two adjacent flipped edges between the oriented F2L corner, you can do this instead and get ZBLL (the alg itself also preserves a lot of LL blocks which can absolutely help with recognition). By knowing ZBLL, this SDM becomes a little bit better

Finally, by switching between R2 U R2 U' R2, D R2 U' R2 U R2 D', and R F R2' F' R' before OLL, you can always preserve any 1x2x2 in the U layer and get a Tripod LL case. By knowing Tripod LL, this SDM becomes even better than before (this can also be applied to preserving any 1x1x2 pair, which would make this idea even better if you know 1LLL for any case with a solved 1x1x2 pair)

The odds of the latter two outcomes aren't particularly high assuming you don't try and influence the F2L corner during LS, but by using R U' R' and R U2 R' (which a lot of LS cases naturally reduce to), you can force an oriented corner roughly half of the time (7/12 cases) . If you also include R' F R F', you can orient the corner 2/3rds of the time (8/12)

Knowing good OLS tricks will also increase your overall odds of getting the F2L corner oriented without wasting time. The best example I can think of are the F2L cases with the edge solved and the corner twisted in place. By using either R U R' U' R U2 R' U' R U R' + inverse or R U' R' U' R U R' U' R U2 R' + inverse, you can always orient the F2L corner using AUF and nothing else

Another simple extension that doesn't demand you to learn a bunch of algs for an obscure algset you're never going to use, is using R S' U2 S R' and D R' S' U2 S R D' to complete F2L if you don't like your OLL case. This to me feels pretty trash in all honesty, especially since after OLL you can always avoid diag and get disproportionate amounts of EPLL, but hey, it's an idea I guess lmao (it would also never lead to an OLL skip using this method which is the other reason I don't like it)

With that all outlined, feel free to discuss, & also share some scrambles where you think this method might be good (I tried already but everything was too easy for CFOP or blockbuilding lol)
The intellectual level of the forum has risen considerably this month with all the method branches proposal. Good work and idea Jay!
 

Athefre

Member
Joined
Jul 25, 2006
Messages
1,128
Solving pieces incorrectly at the beginning of CFOP is something that is often proposed in the New Method thread here. What we often see proposed are things like "Solve the cross but the edges don't have to be permuted correctly", "Cross with two swapped edges", "F2L pairs with unpermuted corners", or things like PEG. The hope is always that you can take advantage of these positions in scrambles to decrease the overall move-count or time that it takes to get the cross and or F2L stage complete. We can think of infinite ways of incorrectly solving F2L and say to use it when it's easy in a scramble. But which of these are actually good? This would probably require some kind of computer analysis.

Scramble dependent or not, intentionally placing pieces incorrectly has to be done in a way that there is a lot of freedom to fix them later. Also, it is often best to fix the messed up pieces simultaneously with a later step. EG, ACMLL, and TCLL (and its ROFL method origin) are all examples of methods which reduce move-count and save time versus what was used before them. Having a separate step at the end is pretty much the moves that it would have taken to solve the pieces correctly in the beginning. Possibly even more.

It's interesting to use the separate "correction algorithm" to influence the next steps. But if the solve relies upon the final correction algorithm being able to alter the final algorithm state, are there techniques that you could use in a normal solve instead? Such as edge influencing, corner influencing, alternate pair inserts, alternate pre-PLL algorithms, or whatever is applicable to the current method. You can develop and learn an arsenal of scramble dependent methods and try to spot them during inspection. But if you are using the same number or more moves to correct the imperfectly solved pieces later, are you saving time versus solving things correctly in the beginning and using other techniques to get your desired better final step cases?

The use of pseudo is a delicate thing. It has to be corrected eventually in the solve. But reducing the overall move-count, inspection time, thinking time, and the later step algorithm quality are all dependent upon the way that the pseudo state is corrected and even the type of pseudo used.
 
Last edited:
Top