[Help Thread] FMC Discussion

[blah]

I've only had that many moves cancel a couple of times. But then, you're clearly already a lot better than me. Frustrating.

I guess it makes sense that you're good at this, considering all the cube theory you've been interested in all along. Congratulations on some good solves.

Oh man blah, this is your new thing. I see it. You did BLD before if I'm not mistaken, and BLD wasn't really my thing, and neither is cube theory. My grasp on commutators is minimal. I think I'll just leave this event alone.

Oh I confess! I was cheating! So far I've only had two or three solves done under an hour I take 1:30 to 2:00 most of the time Hope this'll improve as I get more practice. I have absolutely no interest in no-time-limit FMC because I think it's silly.

But thanks anyway guys

@Mike: Hmm, I don't claim to know a lot about cube theory (only slightly better than the average speedcuber at best), but I'm quite surprised at how little my knowledge on cube theory has helped me in FMC, actually. If by "cube theory" you mean conjugates and commutators, then I guess most would agree with me that you don't need to understand commutators at all to be able to do insertions, all you have to do is memorize a couple of standard 8-movers for corners and the 6-mover for edges. Other than that, I guess the only other "cube theory" I know is edge orientation and permutation parity, which most can understand without trying too hard. That's all, really.

Don't mean to disappoint you, but I honestly think what makes me better than most other beginners is because I turn fast and scramble fast, seriously With only one cube, scrambling fast is a must for me (never timed it, but I can probably sub-5 a 20-move scramble easily) Within the same time limit, I get many more opportunities for trial and error than most others, and so I almost always manage to get lucky :rolleyes: Take a look at my signature: My PLL TA says something about my turning speed, and being good in team BLD means I process cube notation in my head pretty quickly, hence the fast scrambles. My partner and I only had 3 or 4 codes for F2L cases when we got the UWR single and average, the whole of F2L was done using basic UDFBRL notation. Maybe working on these two aspects will help you improve? I'm serious

@Sneaky: Yeah I was once sub-world class-ish in 3x3x3 and 4x4x4 BLD But then the rest of the world caught up and I gave up Don't give up on FMC! As I've already mentioned, commutators aren't a must for FMC, Arnaud didn't know about commutators when he got his competition PB of 33, and look at where he is now! I believe...

1. having started cubing with my version of PBP (Piece By Piece) Petrus for a year,
2. then switching to Fridrich,
3. then practicing ZZ for OH for half a year,
4. then playing around with Roux for about a week just for fun,
5. and having once been good at BLD and had many ideas about it,
6. and having had LOADS of ideas and done lots of research for LS + LL methods,

...have all been beneficial to FMC for me

 

[Mike Hughey]

Hmm, I don't claim to know a lot about cube theory (only slightly better than the average speedcuber at best)

Actually, when I mentioned cube theory, I was also thinking about your explorations into different methods to solve the cube. (Meaning mainly your numbered points that you gave.) That essentially amounts to cube theory of sorts - perhaps not particularly mathematical, but it means you've really studied solving from many different angles. I think studying all those different approaches to solving the cube give you a really big advantage.

Don't mean to disappoint you, but I honestly think what makes me better than most other beginners is because I turn fast and scramble fast, seriously

But yeah, I forgot about that. That's one of my biggest disadvantages - I usually get to explore about 1/3 of the possibilities of most of the really good fewest moves solvers in a given attempt. So when I get lucky, I get one that's almost competitive. But unfortunately, my odds aren't very good for that because I'm so slow. I'm slowly getting faster; it's just a slow and gradual process for me. Like Ton - I'm really only a little behind him in 3x3x3 speed progress. He's just got 4 years on me is all.

 

[AvGalen]

To become really good at FMC you need to:
1) Now about different methods (Layer-by-Layer, BlockBuilding, Corners First)
2) Now many algorithms
3) Be able to think outside of the box (don't be afraid to do F2L with one or 2 bad pieces if that makes F2L very short and it gives you an "almost" good last layer)
4) Get lucky
5) Read about and understand all the techniques I mentioned in The FMC thread

I have always known about commutators, but I am extremely slow at finding corner 3-cycles (30 minutes in a 20 move partial solution is normal for me). I am very good with edge-cycles though. And I never did an official 31, I am still stuck with a 2 year old 33 . I used to be extremely good at "outside of the box", but I have mostly lost that ability since I started using more "mainstream" methods

 

[blah]

@Sneaky: Given my minimal experience with FMC, I don't know if I'm in a right position to say this, but I'm pretty sure EJ pairs are a HUGE no-no in FMC. The worst thing I dread in FMC is ending with 1 or more pieces correctly permuted but misoriented.

The beauty about FMC is that you're not doing speed-blockbuilding, and you can take back moves. I don't know if you've realized this yet, but speed-blockbuilding and FMC-blockbuilding are actually quite different. So you don't have to do a 2x2x2 followed by an expansion into a 2x2x3. In fact, for me, most of the time (which is actually less than 10 times ) I prefer to start with a 1x2x3 because it's so easy to be built efficiently.

And one last thing, I think OLL/PLL is only good when you get an OLL with say, 8 moves or below, followed by A, U, T, J, or if you get cancellations, G. Or skip one of the steps. Or get a REALLY short F2L, say, 15 moves

I'd only use COLL if I get an EPLL skip, firstly because they're generally longer than OLL, and secondly because I hate inserting edge 3-cycles.

@Arnaud: My mistake I thought I read somewhere that you didn't know about commutators when you got your 33, guess I remembered wrongly. I've already provided the link to that ooold FMC thread in an earlier post, kinda inconspicuous though

Any tips on inserting edge cycles? I only know the 6-mover. Are there any others?

 

[blah]

Scramble: F2 D' B' R' D U' R F D2 B F R2 B2 F' D' R D U' B2 D' U R B2 R' U2

It looked sooo easy at first sight - sooo many nice blocks! But I couldn't find anything nice

Premove + scramble: F' R2 B' + F2 D' B' R' D U' R F D2 B F R2 B2 F' D' R D U' B2 D' U R B2 R' U2

Pseudo-2x2x3: D F2 (2/2)
Triple x-cross: L U' L' B U2 B (6/8)
Last F2L edge: U2 B' R B R' (5/13)
Solve LL edges: R' U' F' U F R (5/18)
AUF: U2 (1/19)
Undo premove: F' R2 B' (3/22)
Corner 5-cycle remaining.

or

Pseudo-2x2x3: D F2 (2/2)
Triple x-cross: L U' L' B U2 B (6/8)
Last F2L edge: U2 B' R B R' (5/13)
I don't know what to call this: R' U' F' U F U' R U' (7/20)
Undo premove: F' R2 B' (3/23)
Corner 3-cycle and edge 3-cycle remaining.

Ran out of time

----------

Edit: Premove + inverse scramble: U F' + U2 R B2 R' U' D B2 U D' R' D F B2 R2 F' B' D2 F' R' U D' R B D F2

2x2x2: D' L2 U' L' (4/4)
2x2x3: R F' U2 F' (4/8)
F2L: B' R2 F R B F' U2 (7/15) (Sweet!)
Solve LL edges: F D2 B' D' B D' F' (7/22) (This is actually just an Sune conjugated with an M move, i.e. one of the double-layered Sune OLLs)
AUF: R (1/23)
Undo premove: U F' (2/25)
Corner 3-cycle remaining. Could've just done a Niklas after solving LL edges, but there's probably a better insertion with one more cancellation, too tired to think now, it's 5:15 in the morning here, I'll leave it for when I wake up.

 

[JLarsen]

Thanks blah for that tip on EJ pairs, and explaining speed blockbuilding and FMC blockbuilding. At least I'm good at one of those lol.

 

[blah]

Thanks blah for that tip on EJ pairs, and explaining speed blockbuilding and FMC blockbuilding.

Say you build a 2x2x2: A B C D E (5), but can't find a good expansion into a 2x2x3, and even Johannes solver tells you the optimal solution to expand into a 2x2x3 is say, 7 moves. You get depressed

Here's a nice trick I use very often: Look for a point (which may or may not exist) within the first 5 moves where all 4 2x2x2 pieces are not on any one (or more) face(s), do an X/X'/X2 turn on that face and proceed with your original solution. Now you have 9 more opportunities to expand into a 2x2x3 efficiently.

This is, in my opinion, the main difference between speed-blockbuilding and FMC-blockbuilding. Tell me if my explanation isn't clear enough.

 

[JLarsen]

Thanks blah for that tip on EJ pairs, and explaining speed blockbuilding and FMC blockbuilding.

Say you build a 2x2x2: A B C D E (5), but can't find a good expansion into a 2x2x3, and even Johannes solver tells you the optimal solution to expand into a 2x2x3 is say, 7 moves. You get depressed

Here's a nice trick I use very often: Look for a point (which may or may not exist) within the first 5 moves where all 4 2x2x2 pieces are not on any one (or more) face(s), do an X/X'/X2 turn on that face and proceed with your original solution. Now you have 9 more opportunities to expand into a 2x2x3 efficiently.

This is, in my opinion, the main difference between speed-blockbuilding and FMC-blockbuilding. Tell me if my explanation isn't clear enough.

No I perfectly understand what you are saying thank you very much for your help.

 

[MistArts]

What's a good number of moves for the first two 1x2x3 blocks for Roux? I did two solves and got 18 (7+11) and 17 (4+13), and I don't think it's good at all.

I'd attempt Roux if I get a nice 1x2x3 (<7 moves) but couldn't get a short sliced 1x1x3 (doesn't matter if the colors are matched or mismatched), is this a good approach? What would you guys do in this case?

My FM solve for weekly-28:

Scramble: D' B2 R2 B2 D U B2 U F2 L2 F2 R B' U B' F2 D' F' D L U'

FMC: U D' F' D2 U' B L2 U' L2 U B U B' U' R2 L F' L R' D' U' B2 R2 F2 D' L2 F2 R L B' L R' (32)

1x2x3: U D' F' D2 U' (5)
Opposite 1x2x3: B L2 U' L2 U R' (11)
CMLL: R B U B' U' R' U (18-2)
EO: U' L R' F L R' D' . L R' B' L R' (28-3)
Insert at .: U' B2 R2 F2 D' L2 F2 R2 (33-1)

Cancellations: R' R at the beginning of CMLL cancels to nothing; R' U U' L R' at the beginning of EO cancels to R2 L; R2 L R' cancels to R L at the end of the insertion.

I think Roux has the potential to use in some cases, but not always.

 

[AvGalen]

...
And one last thing, I think OLL/PLL is only good when you get an OLL with say, 8 moves or below, followed by A, U, T, J, or if you get cancellations, G. Or skip one of the steps. Or get a REALLY short F2L, say, 15 moves

I'd only use COLL if I get an EPLL skip, firstly because they're generally longer than OLL, and secondly because I hate inserting edge 3-cycles.

@Arnaud: My mistake I thought I read somewhere that you didn't know about commutators when you got your 33, guess I remembered wrongly. I've already provided the link to that ooold FMC thread in an earlier post, kinda inconspicuous though

Any tips on inserting edge cycles? I only know the 6-mover. Are there any others?

OLL+PLL is indeed a big no-no for FMC. Basically you want to get a skip on one of them (lucky) or manipulate the last layer during F2L so you end up with a 3-cycle (that you can insert) after F2L.

Inserting edge 3 cycles is easier because edges only have 2 orientations. If there is an 8 mover, there are probably many (not just 1 or 2 like with corners) so you have a bigger chance on cancellations. In this thread, you can find information about a really useful class of edge 3-cycles. Basically edge 3-cycles are less "reliable" than corner 3-cycles because they can be very short (6) or very long (12). Short edge-ones are less common then short corner-ones

 

[blah]

Scramble: L' B2 D2 L' F2 L2 B2 D2 B2 L' D2 B' U' R2 D' R F' R' D L2 R' (21f)

2x2x2: D F' D2 R2 (4/4)
2x2x3: L2 U' F U F (5/9)
Triple x-cross: F L U L' F' (4/13)
LS + LL control: U' L' U2 L U' L F' L' F (9/22)
2 misoriented corners remaining.

Questions:
1. Is this a good ending? (I know it's acceptable in this case because I only took 22 moves so far, but what if I took, say, 25 moves to reach here, would this still be a decent ending?)
2. How would YOU twist the last 2 corners with an insertion in this case?
3. What are all the optimal algs (14 HTM) that twist any two corners?

By the way, interesting observation: No B moves in the solution so far

----------

This would've been pretty decent for me if I knew optimal tripod:

X-cross: L2 R2 U2 B2 U' B' (6/6)
2nd F2L slot: L' U L F U2 F' (6/12)
3rd F2L slot: F' U' F (2/14)
Tripod: U B' U B U (5/19)

 

[cuBerBruce]

3. What are all the optimal algs (14 HTM) that twist any two corners?

14 HTM is not optimal.

The basic alg you should know is:

R' B D2 B' R U2 R' B D2 B' R U2 (12f*)

For other adjacent corners or centrally opposite corners, use the same alg with a setup move such that you get a cancellation between the setup move (or undo setup move) with the alg itself.

EDIT (appending):
For the example, you could insert the following at the end:
F' R D2 R' F U2 F' R D2 R' F U2

The first move cancels the last move of the skeleton, giving a 32-move solution. There may be a better insertion elsewhere, though.

 

[blah]

I've come to the conclusion that I can almost always find relatively good starts, but really, really suck at the end. So I decided that I could use some help from the pros here. How would YOU end each of the following scrambles with the start I provided? I'm hoping to see some sub-30 solutions here

This is from this week's forum competition scramble.

Premoves + Scramble: B2 F D' + F R2 F2 R2 B' D2 R2 B' L' B2 F' D F U2 F U' R' F U2 L'
1x2x3: L' U' B' R D2 L2 (6/6)
Triple x-cross: F' D' F' D' B2 (5/11)
11 + 3 = 14-move triple x-cross. How would you solve the rest of this?

Scramble: F R2 F2 R2 B' D2 R2 B' L' B2 F' D F U2 F U' R' F U2 L'
1x2x3: L' U' B' R D2 L2 (6/6)
Triple x-cross: F' D' F' D' F' B2 U F2 B2 (9/15)
How would you solve the rest of this?

----------

After I produced a poor solution, just for fun, I gave myself another hour for the inverse scramble.

Premove + scramble: F' + L U2 F' R U F' U2 F' D' F B2 L B R2 D2 B R2 F2 R2 F'
1x2x3: R B2 D R F D2 * (6/6)
2x2x3: R' F' (2/8)
8 + 1 = 9-move 2x2x3, which is pretty good for me. How would you solve the rest of this? (B/B'/B2 move can be added at * for more opportunities.)

My best triple x-cross was this:
2x2x3: R B2 D R F D2 *B' R' F' (9/9)
Triple x-cross: L' B2 L U (4/13)
(Undo premove: F' (1/14))

My best F2L was this:
2x2x3: R B2 D R F D2 *B R' F' (9/9)
F2L: L' B L2 F U F' U L' U (9/18)
(Undo premove: F' (1/19))

But I couldn't find good continuations to either of these.

 

[blah]

Further proof that I'm terrible at FMC endings Can anyone provide nice solutions for these too?

I found 2 8-move 2x2x3s in the first 15 minutes. And I still DNFed 45 minutes later

Scramble 1 (for 3x3x3 speed) of this week's forum competition: D2 L2 U2 L2 D F2 U2 R2 B2 U' F' R' B L D2 F2 R D' R2 B2 U

No premoves.
2x2x2: F' D' R B (4/4)
2x2x3: L F2 L2 D' (4/8)

Premove: B2 (1/1)
1x2x3: D' R B' R (4/5)
2x2x3: B2 D' B2 (3/8)
EO: L' U' L R U R' (6/14)
Triple x-cross: U2 F U2 F2 (4/18)
Remaining edges: U' F' U2 F (4/22)
Didn't have time to look for insertions.

Premove: B2 (1/1)
1x2x3: D' R B' R (4/5)
2x2x3: B2 D' B2 (3/8)
EO: R' F2 R2 U R' (5/13)
Triple x-cross: U2 F U2 F2 (4/17)
3/4 F2L + ELS in 17 moves. But stuck

For some reason I always get stuck at the end and my mind just blanks out...

 

[blah]

Noob question about insertions: Do you guys really use post-its? I tried it for the first time just now and they fall off after every two turns... *helpless*

 

[Mike Hughey]

Noob question about insertions: Do you guys really use post-its? I tried it for the first time just now and they fall off after every two turns... *helpless*

I've used post-its a lot. And yeah, definitely I have the same problem. Sometimes I add tape to the back of the post-its, and that helps for a little while. But I think the best thing would be to get some extra Cubesmith stickers or something, and use those. I keep meaning to do that, but I never get around to it.

 

[blah]

I've used post-its a lot. And yeah, definitely I have the same problem. Sometimes I add tape to the back of the post-its, and that helps for a little while. But I think the best thing would be to get some extra Cubesmith stickers or something, and use those. I keep meaning to do that, but I never get around to it.

I've had an idea, but I don't know if it's practical, I'd like to know what you guys think of it since you've definitely had more experience with other weird methods to track insertions.

Since I'm allowed three cubes, and I've been doing fine with only one so far, I thought I could leave the third cube blank, i.e. unstickered, and always have a set of Cubesmith stickers ready. The simplest example I can think of to describe this is an A-perm: sticker the three U positions yellow, the three clockwise positions blue, and the three counterclockwise positions red, now write 1, 2 and 3 on each of those stickers (which denotes their respective orders in the cycle). In case it wasn't clear enough, each sticker on the same piece must share the same number.

Since every interchangeable sticker is of same color and labeled with numbers, it's very easy to see which two pieces are interchangeable instantly (especially on a blank cube) because you'd have two same colors on the same face. My problem with insertions is that I take a while to realize that 2 pieces are interchangeable if the specific stickers I'm tracing are not on the same face. E.g. FRU and LFU are interchangeable, but I take a while to notice this, it'd be much easier to notice this if I see that UFR and ULF share the same color.

So with this approach, at any point in the solution when 2 same-colored stickers are not on the same face, you know instantly that an 8-mover is not possible. Obviously, when 3 are on the same face, an 8-mover is also not possible.

Besides, with three "angles" to look at the commutator, you'd see it more quickly than with only one "angle".

So, what do you think?

As for preparation, obviously you could have a blank cube with 5 corners (and edges) labeled and numbered prepared before competition.

P/S: In case you were wondering, I'd use the second cube for the inverse scramble

 

[Mike Hughey]

I've come to the conclusion that I can almost always find relatively good starts, but really, really suck at the end. So I decided that I could use some help from the pros here. How would YOU end each of the following scrambles with the start I provided? I'm hoping to see some sub-30 solutions here

This is from this week's forum competition scramble.

Premoves + Scramble: B2 F D' + F R2 F2 R2 B' D2 R2 B' L' B2 F' D F U2 F U' R' F U2 L'
1x2x3: L' U' B' R D2 L2 (6/6)
Triple x-cross: F' D' F' D' B2 (5/11)
11 + 3 = 14-move triple x-cross. How would you solve the rest of this?

R U R2 F R F' . (6/17)
B' R' U' R U B U (7/24)
insert at .: R B L B' R' B L' B' (8/32)
B' B' become B2, so one move cancels; 31 moves; with the premoves, 34 moves.

I guess it's not very good (nowhere near sub-30), but I would be happy with it, because I'm not very good. I wish I had found this start - I would have been very happy!


Oh, and I think your idea for stickers seems pretty good - it would certainly give you an advantage on doing insertions. But I'm not sure I'd be willing to work that hard to prepare for it; I think it would require some practice with that to get good at it, just because of the mechanics of it all.

 

[blah]

I've come to the conclusion that I can almost always find relatively good starts, but really, really suck at the end. So I decided that I could use some help from the pros here. How would YOU end each of the following scrambles with the start I provided? I'm hoping to see some sub-30 solutions here

This is from this week's forum competition scramble.

Premoves + Scramble: B2 F D' + F R2 F2 R2 B' D2 R2 B' L' B2 F' D F U2 F U' R' F U2 L'
1x2x3: L' U' B' R D2 L2 (6/6)
Triple x-cross: F' D' F' D' B2 (5/11)
11 + 3 = 14-move triple x-cross. How would you solve the rest of this?

R U R2 F R F' . (6/17)
B' R' U' R U B U (7/24)
insert at .: R B L B' R' B L' B' (8/32)
B' B' become B2, so one move cancels; 31 moves; with the premoves, 34 moves.

I guess it's not very good (nowhere near sub-30), but I would be happy with it, because I'm not very good. I wish I had found this start - I would have been very happy!


Oh, and I think your idea for stickers seems pretty good - it would certainly give you an advantage on doing insertions. But I'm not sure I'd be willing to work that hard to prepare for it; I think it would require some practice with that to get good at it, just because of the mechanics of it all.

I'd always be happy with any sub-35 solve! At least for now and for the rest of the year I guess...

May I know what your thought process was for that solution? I'd never have thought of that! Was it really just an F2L insertion followed by the 6-move OLL as it appears to be? Despite having done lots of research on LS + LL methods, I realize that's actually my biggest weakness in FMC - I think it's actually worked against me, I've been spending too much of my cubing life thinking of braindead ways to solve LS + LL that I can't come up with short and easy solutions on the fly now

Actually I'd be happy to give you tips for good starts, but with your experience in the event, I'm not sure it'll be worth anything I use a very algorithmic approach, actually, which kind of insults the elegance of FMC, I regret to say. (Algorithm as in Euclidean Algorithm, not as in PLL algorithm, hope you get what I mean.) I'll try to polish up my 2x2x3 approach before I share it - I may have been just lucky enough to get good starts for the majority of my attempts so far

 

[Mike Hughey]

May I know what your thought process was for that solution? I'd never have thought of that! Despite having done lots of research on LS + LL methods, I realize that's actually my biggest weakness in FMC - I think it's actually worked against me, I've been spending too much of my cubing life thinking of braindead ways to solve LS + LL that I can't come up with short and easy solutions on the fly now

Actually, it's one of my main "techniques". Not much of a technique, but it's a quick and dirty way to get a solution about 1/3 of the time that at least isn't terrible. It's how I manage to almost never DNF, and usually come in sub-40. I just do the stupid simple thing and insert the fourth pair, exactly like I would with a normal Fridrich solution. If there are multiple ways that I know, I try them all. This one worked pretty good with just 6 moves. (When this doesn't work, sometimes I'll just try to solve all the edges and see if I can get 5 permuted corners. If so, then I can try 2 insertions.)

Then I start trying OLLs as fast as I can, to see if one can solve edges and leave just 3 corners to make a good insertion. This one was really lucky because a 6-move OLL actually gave me one. You can't beat that - any time I get one of those, I'm elated. If I were any good, I wouldn't go beyond 7-move OLLs with this, but I stink, so I often go as far as the 9 or 10 movers, figuring if I get a good insertion, it still at least won't be terrible. But I'm sure I waste too much time doing that - the 6- and 7-movers are probably all that are really worth trying.

Actually I'd be happy to give you tips for good starts, but with your experience in the event, I'm not sure it'll be worth anything I use a very algorithmic approach, actually. (Algorithm as in Euclidean Algorithm, not as in PLL algorithm, hope you get what I mean.) I'll try to polish up my 2x2x3 approach before I share it - I may have been just lucky enough to get good starts for the majority of my attempts so far

An algorithmic approach would be great - I'm an engineer at heart, not a scientist. I generally don't particularly care why something works; I just want to make it work.