• 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!

Thoughts on 3x3 improvement progression (CFOP)

Status
Not open for further replies.

Smiles

Member
Joined
Apr 22, 2012
Messages
575
YouTube
Visit Channel
Note: The following is subject to my opinion and you may not agree.

CFOP solvers anywhere from beginners to sub-20 and beyond often come up with the same questions regarding progression.
"My average time is ___, should I learn ___?"
In my opinion, your times shouldn't dictate when you learn things and what you learn. Everyone has different strengths and weaknesses. And if you're anything like me, you would have known full PLL before even starting to learn F2L, and be sub-35 with good look ahead. Obviously that's really bad and not even close to ideal, so I would never have classified that as the average sub-35 CFOP cuber. If I asked for general advice, I'd get "practice", "turn slow and look ahead", and "learn full PLL". I'd get other things too, but just being sub-35 doesn't mean that that advice would have been useful to me.

The point is, cubers should be able to identify what stages they're at in terms of improvement (especially because I'm making lists about it). And instead of grouping people into categories based on sub-x, I think it'd be better for everyone to move onto the next step of improvement in each part of their solve (cross, F2L, OLL, PLL, and extra things) after they believe they have completed the previous milestone.

Since CFOP is split into multiple steps, I think that we can work on something different for each step all at once. For example, you can practice planning your cross while you practice lookahead in F2L, and at the same time learn some new PLL algorithms, since every solve will have all of these components. Obviously that doesn't mean to focus on something new for every step all at once every day, but each cuber can decide for themselves how much improvement they should pursue all at once.

The following are lists of things that I think CFOP cubers should achieve for each step of their solve, in approximately that order.
They're categorized by CFOP step, not by how fast you can solve the cube.

tl;dr: I think CFOP progression shouldn't be based on times, but instead based on how advanced a CFOP cuber is in each step of CFOP.

Cross:
1. Find all 4 pieces during inspection. Before you start your solve, you should have a general idea of what you're dealing with.

2. Do cross on the bottom.

3. Be able to identify the difference between oriented and bad edges. This means, for example, a cross piece going from UF to DF can either be inserted by F2 (good edge) or U' R' F R (bad edge). That's just one example.

4. Plan out your cross before you start your solve. During practice, don't worry about how long it takes to inspect.

5. Be able to finish your cross in 8 moves or less, or really quickly, every single time. This involves seeing piece relationships that may require watching some example solves.

6. Track an F2L pair while making the cross. This means your inspection is good enough that you don't have to think about your cross as you make it.

7. Be able to build an x-cross when the opportunity is there. Watch example solves to get the idea.

8. Anticipate your first F2L case during inspection, almost every solve.

F2L:
1. Learn the F2L concepts that allow you to solve almost every case in a logical way by corner-edge pairing. There are exceptions.

2. Learn to minimize rotations by using both hands and back slot inserts such as L U L' or R' U R, etc.

3a. Be able to look at a corner-edge case and solve it without looking. This means getting each case out of your thoughts and into your muscle memory. Since there are so many possible cases during F2L, this step takes a long time.

3b (optional, but I recommend it). Drill F2L cases into your muscle memory by increasing your turning speed during F2L. While you do this, try NOT to look at the pair you're solving.

4. Decrease your turning speed so that you can still see the other pieces moving around. You may have to turn extremely slowly (less than 2 moves per second while solving F2L cases). The purpose of this is to look around at other pieces and track/anticipate your next F2L pair while solving the current pair. This is a more difficult step to arrive at if you did not do step 3a.
If you have pauses longer than 1 second between F2L pairs, you're turning too fast.

5. Increase your turning speed slightly, but still do not pause. If you must pause a little, that is okay and you will improve to stop doing that. If you have significant pauses, then again, you're turning too fast.
Step 5 never ends. Always continue to do this, no matter how fast you are.

6. Notice which F2L cases seem to result in slower solves or breaks your rhythm during F2L. Learn algorithms for some of the less desirable F2L cases, but ensure that these algorithms do not hinder look ahead. The possibilities are enormous, so this requires a lot of experimenting and exploring.
Note: R2, F2, L2, B, wide turns, and x/z rotations are not recommended in F2L algorithms, as they hinder lookahead. You can use them if necessary, but if you go out and find a bunch of algs like that, your times will go up. Keep it simple. See the Advanced section below.

7. Recognize edge orientation to anticipate F2L cases, and to know in advance whether or not a cube rotation is necessary.

Advanced:
If you decide to learn these optional techniques, you don't need to go about them in any particular order.

Know an efficient solution to each F2L case. This means that the solution:
- is low in move count OR can be executed quickly (like (R U R' U')3)
- does not use many moves that significantly hinder look ahead (i.e. try to avoid wide turns, B turns, z/x rotations etc.)
- can be done with 0-1 cube rotations
For example, set-up with R' U' R U R' U R F' U F and solve the back slot with R' F R F' R' U' R.

Influence LL EO by using different ways to insert pairs. This includes possibly inserting using rotations, sledgehammers, etc.

Multislotting: Influence other pairs by using different ways to insert pairs. The most basic examples are U R U' R', U2 R U2 R', and R' F R F'.
It is not necessary to learn how to insert 2 pairs with one algorithm.

Last Layer:
1. Learn 2-look OLL. 3 algs for edge orientation, 7 algs for corner orientation. (10)

2. Learn 2-look PLL. 2 algs for corner permutation, 4 algs for edge permutation. (6)

3. Learn 1-look PLL, 21 algs total (including the 6 from 2-look).

4. Learn 1-look OLL, 57 algs total (including the 10 from 2-look). This is an advanced step and may happen over a long period of time.

5. Drill all of your OLL and PLL algorithms and gain confidence in executing them quickly during solves.

6. Experiment with new algorithms for both OLL and PLL, if any alg feels sub-optimal. It is highly unlikely that the first algorithm you learned for each case is the one that is best suited for you. Experiment with different finger tricks for each algorithm, and try to reduce regrips.

Advanced:
If you decide to learn these optional techniques, you don't need to go about them in any particular order.

- Learn additional PLL algorithms for different angles (particularly if you would have to do U2 + alg + U2).
For example, R2 u' R U' R U R' u R2 B U' B' = y2 R2 F2 R U2 R U2 R' F R U R' U' R' F R2

- Regularly use CLL recognition during OLL to help predict PLL.
This implies that you should know which CLL case each of your OLL algorithms will either solve, or produce an adjacent corner swap, or produce a diagonal corner swap.
For example, if I see a 2x2 square solved when I arrive at PLL, it can either be A perm or V perm. Using prior CLL recognition during OLL, I can quickly recognize this PLL case without checking any other stickers, based on whether I predicted my corner swap to be adjacent or diagonal.

LS + LL Extensions:
If you decide to learn these optional techniques, you don't need to go about them in any particular order.

VHLS: Vandenbergh-Harris Last Slot
Condition: Corner edge pair can be inserted in 3 moves.
Effect: Solves LS + EO.
Advantage: Become familiar with concept of edge control, OCLL is fast, connects to other subsets like ZBLL, COLL.
Disadvantage: VHLS + OCLL/COLL is often slower than LS + OLL.
I recommend: At least learn the sledgehammer case.

Winter Variation + Summer Variation
Condition: EO is solved, corner edge pair can be inserted in 3 moves.
Effect: Solves LS + CO, and therefore OLL.
Advantage: Usually fewer moves than LS + OCLL, can be done 2-gen if desired (useful for OH).
Disadvantage: Can only be used when EO is done.
I recommend: Learn some of the easy cases, such as L' U R U' R' L.

OLS: OLL + Last Slot (includes Winter Variation and Summer Variation)
Condition: Corner edge pair can be inserted in 3 moves.
Effect: Solves LS + OLL.
Advantage: Usually fewer moves than LS + OLL.
Disadvantage: High algorithm count.
I recommend: Check out VLS and RLS (2 subsets of OLS) and find easy cases such as M' U R U' r'.

Pure LL Extensions:
If you decide to learn these optional techniques, you don't need to go about them in any particular order.

COLL: Corners of the Last Layer
Condition: EO solved
Effect: Preserves EO, solves CO/CP. What remains is EPLL.
Advantage: Easy PLL recognition and quick execution.
Disadvantage: Can only be used when EO is done, sometimes COLL + EPLL is slower than OCLL + PLL.
I recommend: Don't learn the Sune/Antisune sets.

OLLCP: OLL + Corner Permutation
Condition: none
Effect: Solves EO/CO/CP. What remains is EPLL.
Advantage: Easy PLL recognition and quick execution.
Disadvantage: High algorithm count, sometimes OLLCP + EPLL is slower than OLL + PLL.
I recommend: Learn some easy OLLCP cases, especially if your OLL alg will leave you with a diagonal swap, learn the case that leaves you with EPLL instead.
For example, if you arrive at OLL and you would normally solve the OLL case as F (R U R' U')2 F', but it results in a diagonal swap, the corresponding OLLCP algorithm is U R' U' R' F R F' R U' R' U2 R.

How to use these lists:
If you are good at 1, look at 2.
If you are good at 2, look at 3. Continue until you find something you're not good at. Stop. Practice it. Repeat.
Do the above for each of the lists.

For extensions, there are definitely far more than are listed here. I listed the ones I felt are either important or easy to learn.
 
Last edited:

JemFish

Member
Joined
Jan 27, 2015
Messages
338
Location
Sydney, Australia
Hey, great post. I've bookmarked this thread and will practice based on this. I've seen all of this advice before, but it's nice to have it all in once place, especially with the practice suggestions numbered. Again, great post!

EDIT: And thanks for adding in the extensions.
 
Last edited:

cmhardw

Premium Member
Joined
Apr 5, 2006
Messages
4,117
Location
Orlando, Florida
WCA
2003HARD01
YouTube
Visit Channel
I like your breakdown of the skill sets necessary to really improve at each step of CFOP. One suggestion would be to add a section about extensions. For example COLL, OLLCP, MGLS and all its variations, etc.

By your list I am at stage 7 of cross, stage 4.75 or somewhere between 4 and 5 for F2L, and stage 4 for LL. You've inspired me to try to predict my first pair more often during inspection of cross :)
 

Smiles

Member
Joined
Apr 22, 2012
Messages
575
YouTube
Visit Channel
I like your breakdown of the skill sets necessary to really improve at each step of CFOP. One suggestion would be to add a section about extensions. For example COLL, OLLCP, MGLS and all its variations, etc.

By your list I am at stage 7 of cross, stage 4.75 or somewhere between 4 and 5 for F2L, and stage 4 for LL. You've inspired me to try to predict my first pair more often during inspection of cross :)

yeah i thought about extensions. i'll add that now.

and that's awesome that you're benefiting from this haha. if people actually like this then i guess it could be a guide or something.
 

AlphaSheep

Member
Joined
Nov 11, 2014
Messages
1,081
Location
Gauteng, South Africa
WCA
2014GRAY03
This thread should be stickied and a link to it should automatically posted to every "How can I be sub-xx?" thread...

As someone who turns really really slowly compared to others my speed (I peak at ~3-4TPS doing the sexy move) , but has no problem with lookahead or learning lots of algorithms, I couldn't agree with this more.

But to add my own thoughts on progression...

People often look at their best solves, and focus on beating their PBs, which I feel is a big mistake. To improve, you have to look at your weaknesses, and weaknesses are most apparent in your slowest solves.

I started to get some good insight into my solves when I started reconstructing some of them. It's difficult at first, but like everything else, reconstructing becomes easier with practice. I don't do it with all my solves, but once or twice in 100 solves, I'll come across a solve that made me think it should have been faster, so I go to alg.cubing.net and reconstruct it.

Every solve is different. Sometimes I'm slow because I spent too long searching for an F2L pair. Sometimes I made a mistake by inserting the wrong piece somewhere and had to fix it, or had slow LL recognition, or mistook one case for another, or had a tricky EOLine, or missed a turn that was planned in inspection, or just had an unnecessarily high move count. There are so many reasons a solve goes wrong, but I can usually figure out exactly what it was that caused the solve to be slower than average.

Then I spend 2 or 3 minutes looking at what I did, and trying to think of ways to do it better. If I couldn't find a piece, I look at what other options were available. I look for patterns so that maybe I'll recognise the case next time it comes up. If I messed up an alg, I go back and try it over and over. If I frequently mess up the same alg, I look for an easier one that suits my style. If it's bad case recognition, I look for easier ways to spot the case.

I've been doing this for two months, and I've been improving steadily ever since, despite my practice time being less than half of what it was before.
 

Seanliu

Member
Joined
Jul 29, 2014
Messages
330
YouTube
Visit Channel
This thread should be stickied and a link to it should automatically posted to every "How can I be sub-xx?" thread...

As someone who turns really really slowly compared to others my speed (I peak at ~3-4TPS doing the sexy move) , but has no problem with lookahead or learning lots of algorithms, I couldn't agree with this more.

But to add my own thoughts on progression...

People often look at their best solves, and focus on beating their PBs, which I feel is a big mistake. To improve, you have to look at your weaknesses, and weaknesses are most apparent in your slowest solves.

I started to get some good insight into my solves when I started reconstructing some of them. It's difficult at first, but like everything else, reconstructing becomes easier with practice. I don't do it with all my solves, but once or twice in 100 solves, I'll come across a solve that made me think it should have been faster, so I go to alg.cubing.net and reconstruct it.

Every solve is different. Sometimes I'm slow because I spent too long searching for an F2L pair. Sometimes I made a mistake by inserting the wrong piece somewhere and had to fix it, or had slow LL recognition, or mistook one case for another, or had a tricky EOLine, or missed a turn that was planned in inspection, or just had an unnecessarily high move count. There are so many reasons a solve goes wrong, but I can usually figure out exactly what it was that caused the solve to be slower than average.

Then I spend 2 or 3 minutes looking at what I did, and trying to think of ways to do it better. If I couldn't find a piece, I look at what other options were available. I look for patterns so that maybe I'll recognise the case next time it comes up. If I messed up an alg, I go back and try it over and over. If I frequently mess up the same alg, I look for an easier one that suits my style. If it's bad case recognition, I look for easier ways to spot the case.

I've been doing this for two months, and I've been improving steadily ever since, despite my practice time being less than half of what it was before.

Awesome! How to reconstruct? Do you have to film the solve?
 
Status
Not open for further replies.
Top