Practicing Modern Quasi-rotationless F2L
Quote Originally Posted by FinnGamer View Post
I'm trying to stop Cube rotation during my F2L, but it's really awkward to do some moves without y, can somebody give me tips?
If the edge is correctly flipped, solve it using <R,U,L>. If the edge isn't flipped correctly, flip it with an M', solve it <R,U,L>, then put the cross edge back in place
Is that really all there is to it? Because that seems to add a lot of moves. E.g. if the edge is at UR and needs flipping, U takes it to UF, then M' flips it, then you solve the resultant f2l case, then you often have at a U/U' and an M to finish (a move that in itself can be awkward), and thats hoping that the cross edge didnt become trapped in a slot whilst the pair was being placed! (can happen if you do an insertion that takes advantage of empty slots) Thats at least full four moves more than y + f2l case (furthermore this makes the assumption that both f2l cases are the same length, and does not account for situations where the act of flipping the edge makes for a longer/shorter case. This will however probably average out over all possible eventualities so it is a reasonable assumption). This makes sense if you can do four moves (one of which is M) in less time than it takes to do a y/y'. I did some rough trials and reckon an unplanned y move is roughly a quarter of a second. (I mean unplannd as opposed to y moves, such as those in a G-perm, that are obviously drilled in by repetition.)
In the simplest possible example: take (R U' R' y)
I could solve this with U M' (U2 L' U L) U M as theZcuber suggested (beginning with M'), be a little smarter about where the pair ends up by using M, as in: U' M (L' U L) U' M, turn the f face with F U F' (although F U tends to be awkward and thus slow): or I could just do y' R U R'. I think the latter is probably fastest.
If you can do U/U' + M' + U/U' + M in less than a quarter of a second (16 TPS) then the approach given has benefits over simply rotating, otherwise why not just rotate?
FinnGamer, I suggest you take a look at Macky's site
http://cubefreak.net/speed/cfop/f2l_list.php where it has specific algs for pretty much every case. But I don't think it is necessary to learn every single one, just use this as a reference if you find a particularly baffling case.
Just to clarify, an oriented edge is one which, if it were brought to the U face using only [L,U,R] moves,the uppermost colour would be that of F or B, and the side would be L or R. if the edge is misoriented, F/B/wide turns/M turns need to be involved to place the corner edge pair, in oriented cases it can be done just with [L,U,R]. Obviously you can identify orientation of an edge without actually bringing it to U with a little practice.
In my opinion youy should start by learning the wide turn insertions for the standard matched pair and separated pair cases which so many f2l cases reduce to (basically f R' f' and the left equivalent f' L f for the matched case in the rear slots, sledgehammer for matched case misoriented in FL or FR, and f R f' or f' L' f, with a rotation for the unmatched case misoriented in FL or FR as explained earlier. Also worth knowing is r' U' R U M' if you just hate f turns, but it is effectively a move longer. The unmatched pair equivalent: (U') M' R' U R U' M adds too many moves over the f R f' solution in my opinion), you can drop your average cube rotations per solve down to below 2, and never have to turn the B face (which seems to be the basic objective here right?).
The maths of this: on average half of f2l cases are misoriented, so solving all oriented cases with L,U,R would mean two cube rotations per solve, (actually 1 if you are so good at look ahead that you did all the oriented pairs, then rotated and did all the misoriented ones - an interesting exercise for your slow solving practice). If you know a couple of ways of dealing with the most common misoriented cases, as given earlier, then you further reduce the need for rotation, leaving you averaging somewhere below 2 rotations.
Your only particularly difficult cases become those that the pairing phase is awkward if there is not a front slot free, perhaps requiring the rotation of B for the pair up (eg setup R U R' U' R U' R' y). Many of these cases can be navigated by using empty slots, especially if they come up early in the f2l where there are lots of slots to use, but if not you can learn a specific alg like U2 f U R2 U' f' (solves given case) from Macky's site, or just rotate, its really not a big deal.
There are of course a few F2L cases that you may have learned specific algs for. If these algs are of the pair up and insert variety, you can pair up as normal, turn U face as necessary to position the pair, then insert using one of the shortcuts given, or if not, learn the alg from a different angle / learn a different alg. By slow solving and thinking carefully about each case as it comes up, you will soon drop rotations.
So how to train this to get it under your fingers:
For each alg just run it backwards and opposite (U' becomes U, R' becomes R etc) to set up each case.
1) Learn the four matched pair oriented cases: R U' R', L' U' L (front), R' U R and L U' L' (rear). I would be surprised if you were not already good with these.
2) Learn the four matched pair misoriented cases F R' F' R, F' L F L' (front sledgehammers) f R' f' and f' L f (rear insertions with pair at left or right)
If you hate f turns try: r' U' R U M' and l U L' U' M' with the pairs along the B side.
3) Learn the four unmatched pair oriented cases: R U R', L' U' L (front), R' U' R, L U L' (rear). Again I am sure you know these already.
4) Learn the four unmatched pair misoriented cases: do a rotation if FR or FL, f R f' or f' L' f (rear). If you hate f turns then just do a rotation with these too, as the M slice version is too long really.
5) Learn/translate any specific algs. (RUR'U')x3 works out really nicely when translated to all slots, e.g. in BL it is: z (RUR'U')x3 z', or (LUL'U')x3 if you are ambidextrous and follow my translation rule: Here is a translation of a useful F2L alg into another slot: U2 R2 U2 R' U' R U' R2 (FR) becomes U2 L2 U2 L' U' L U' L2 when translated to BL. notice the similarities, it turns out that keeping U rotations in the same direction and turning Rs to Ls again maintaining direction turns any [L,U,R] FR slot alg into its BL equivalent, or FL into BR.
6) Practice all of this by doing (very very) slow solves, actively only rotating if you come across a case for which you have given yourself permission to rotate. (like y' R U R'). Look carefully at each case as you come across it, and identify if the edge is oriented or misoriented (learn to do this prior to bringing the edge up to the U face).
7) One more thng is to try and use a d turn rather than a full rotation. Consider R U' R' U' y. This can be solved with d R U R' quickly and efficiently.
If oriented: you can match and insert the pair just using just L U and R, alhough depending on thecase you may have a fingertrick/shortcut/alg that solves the pair more efficiently than an intuitive {L,U,R} solution. If that is the case, perform your shortcut by all means. The idea here is to train ourselves to solve F2L in a move efficient/minimal rotation way and part of that is always treating each case in the same manner unless we have a very good reason to do otherwise (such as multislotting). If it still feels awkward or too move heavy because of the position of the slot, consider looking up the case on Macky's site and see if there is an alg that suits you. Note that the principle illustrated by one alg may be generalisable to other cases, so its always worth having a look.
If misoriented: Try to match up the pair on the U face in the normal way, slots permitting. If you can match them up, insert using the misoriented insertions given above (like f R f') If you can't match them up, consider looking up the case on Macky's site and see if there is an alg that suits you. Note that the principle illustrated by one alg may be generalisable to other cases, so its always worth having a look. Also play with adding a d/d' turn into the solution somewhere, this may orient your pieces at a convenient point.
If you can't find an alg that you get on with, and d doesnt seem to make life easier, add the case to your list of "permission give to rotate", but the idea here is to keep that list as short as possible, and slow solving is where you ingrain these ideas onto your brain prior to full speed solves. Return to this list from time to time and try to eliminate more cases with your more experienced eye.
A good further exercise is to limit yourself to one rotation (or d turn) per super slow solve. In this case you complete the cross and them look at all four pairs and their positions on the cube. You will have on of the following: all oriented, three oriented, one misoriented, two and two and so on. Solve the oriented ones, then do a single rotation then do the misoriented.
As an alternative, force yourself to do all the misoriented ones by pairing and using an insertion technique after the oriented ones are placed. The provisor that B is forbidden may mean you have to ruin a full slot to pair up as you want to, and then refill it. This is a sign that you may want to look up that case as there will probably be a more efficient way on Macky's site. You will find that you are able to almost always do a rotationless solve in this way after a short while.
As you go through these practice methods, every now and again check your movecount, and make sure it isnt going above 28-32 moves for F2L (not including cross) as almost all F2L pairs are easily solvable in 8 moves or less, regardless of slot. The only exception which will trash your movecount is RUR'U' x 3 (or its underappreciated brother (URU'R')*3), but its fast so we let it off the hook.
If you continue to introduce ideas in your slow solving, and use some of these practice idetas to essentially trick your mind into really thinking about what it is doing (as that is what exercises with arbitrary restrictions basically do) you will soon find yourself with a very short list of cases where you need to rotate, and a bunch of tricks and skills that can be executed as smoothly in any slot as they can in FR, and a practically rotationless F2L which is as efficient as it is fast.
Sorry this is so long, but it is something I have been thinking about for ages. To pose a question myself, we keep referring to "Modern F2L" but what do we really mean by that? Is it the "very few rotations, awareness of partial edge control, multislot a bit where you can" approach I strive for, or is there a more subtle evolution of style that I am missing? What is "proper" F2L technique at the highest level? In a great post Escher made some time ago, he referred to Breandan Vallance's old style and new style with regard to F2L. How did it change, and what made him change it? Precisely how much hassle are top cubers willing to go through to avoid a y/y' turn? How does all this tie in with other systems like WV? Do people really know Winter/Summer variation in all slots? Thats more than one question, let me try again: I think I will start a thread on these issues, where should I put that?.
FinnGamer - I hope this helps.
