Speedcubing
/ˈspiːdkjuːbɪŋ/
-noun
1. The art and sport of solving a Rubik's Cube in least time possible.
Welcome to the obsession. My current speedcubing method for Rubik's Cube is the Fridrich method with the following extra techniques:
- Extended cross and other block approaches to F2L when possible.
- F2L cross on bottom or left. Frequent use of empty slots and miscellaneous tricks.
- Multiple algorithms for corner-edge pairs to minimize pause and regripping.
- Avoiding 4 wrong LL edges when fixing the fourth slot (but only when it flows nicely from the rest of the F2L)
- The nicer patterns from COLL when all LL edges are correctly oriented
- PLL recognition by blocks of continuous colors
With this method I set multiple world records between 2004 and 2006 in both single solve and average of 5 for 3x3x3 speedcubing.
Menu
How should I learn the Fridrich Method?
Fridrich and Beyond
Other Speedcubing Methods
Memorizing Algorithms
How should I learn the Fridrich Method?
You just bought a Rubik's Cube and can't wait to start learning the Fridrich Method? Not so fast. The Fridrich Method is quite advanced and is meant for cubers near 30 seconds who want to average under 20. Since the full Fridrich method requires a large number of algorithms, it is best to learn it in several steps.
- Step 0: Preliminaries
- Before anything, you should select a color scheme and learn the notation.
- Step 1: Simple Layer-by-Layer
- Learn Leyan's Beginner Solution. The Fridrich Method is based on the Layer-by-Layer (LBL) approach, which solves the three layers of the cube in order: first, middle, then last. Although full Fridrich actually combines the first two layers, the simple LBL method in Leyan's guide will get you started in the right direction.
Make sure to keep the cross on either bottom or left during the first two layers. Since this method requires about 110 moves, only 2 turns per second is needed to get a sub-minute average.
I realize that Jasmine's beginner page is the standard among layer-by-layer beginner's method guides. I recommend Leyan's over Jasmine's because Leyan teaches to do the first two layers with the cross on bottom. Later on, this will make looking ahead much easier and will be crucial for those sub-20 times."Algorithms" needed: 11 total. Of these, only 4 are for the last layer. The rest are short and mostly intuitive "algorithms."
Moves on Average: ~110 moves on average
Possible average:~35 seconds
Goal for Average before Advancing: 60 sec, which comes out to a little less than 2 tps (turns per second) - Step 2: 4-step last layer (2-step OLL and 2-step PLL)
-
The method described in Leyan's guide often solves each of the four steps of last layer (edge orientation, corner orientation, corner permutation, and edge permutation) using more than one algorithm. Our goal in this step is to combine these sub-steps so that each of the four steps, in the same order, is now solved with just one algorithm. The list of required algorithms, which is a subset of full OLL and PLL, are given on this page.
Algorithms needed: 16 last layer algorithms (3 edge orientation, 7 corner orientation, 2 corner permutation, 4 edge permutation)
Moves on Average: ~85 moves
Possible average: ~25 seconds
Goal for Average before Advancing: 40 sec (~2 tps)Once you have memorized the algorithms, consult Getting Faster with an Easy Method for tips on how to get down to 40 seconds.
- Step 3: Fridrich for the Lazy Cuber
This is almost full Fridrich. Work on the following substeps simultaneously:
- 3a: 3-step last layer (Full PLL)
Combine the 2-step PLL into one by learning the complete PLL.
Algorithms needed: 31 last layer algorithms (the same 10 orientation algorithms from Step 2 and 15 new permutation algorithms for a total of 21 PLL)
- 3b: Standard F2L
The method described in Leyan's guide solves the first two layers by first solving the corners and then the edges. F2L allows us to solve a corner and an edge at the same time, thus combining some of the steps within the first two layers and reducing the number of algorithms applied from a maximum of 8 to 4. Learn the standard F2L and the basic techniques described on the F2L page.
Algorithms needed: 41 F2L "algorithms," many of which are intuitive and short (only the first 3 or 4 moves are essential).
Although memorizing algorithms is of course important, at this stage looking ahead is much more crucial to getting faster times. Once you are comfortable with the algorithms, start praticing to look ahead. After learning both Steps 3a and 3b:
Moves on Average: about 63 moves
Possible average: 15 seconds
Goal for Average before Advancing: 25 sec (about 2.5 tps)- Step 4: Full Fridrich (Full OLL)
As in Step 3(1), we combine the two-step OLL into one by learning the complete OLL. This step is simultaneously the least important and the one that requires the most memorization. For cubers who are already very good at two-step OLL, learning these algorithms might only improve their time by 2 seconds or so. Of course, it is a necessary step for becoming a world-class cuber.
As you are learning OLL, continue to practice looking ahead. To achieve a sub-20 average, a good goal is 3 tps.Algorithms needed: 78 last layer algorithms (47 new orientation algorithms)
Moves on Average: 56 moves
Possible average: 11 seconds ??
Of course, it's not necessary to strictly follow this outline. Many people start learning OLL earlier, and it's perfectly possible to mix the OLL algorithms you do know with two-step OLL for cases you don't. Several cubers have learned the entire Fridrich method in less than three months. Even with the high level of speedcubing in the last year years, because of the increased availability of optimized algorithms and videos it's still possible to become a world-class speedcuber in less than one year.
Memorizing Algorithms
Yes, the Fridrich Method does require you to memorize a large number of algorithms. Yes, it's difficult to memorize and keep memorized the algorithms for many OLL cases that happen only rarely. There are, however, things that can make the process less painful.
Relate patterns to one other, by shape, mirroring, inverses, finger tricks, etc. Breaking an algorithm down into smaller substeps or two other patterns is also useful. For example, #12 on my list is #45 twice with a y in the middle. While learning the algorithms, always keep a print-out page handy for reference. Feel free to edit mine by putting in your own algorithms.
One good technique for memorization is to find key shapes or blocks of colors while excuting an algorithm. For example, in doing RUR'URU2R', look at the block formed by the FRD corner and the FR edge. As you excute the algorithm, this block moves up to the top layer and is rotated clockwise 90 degree at a time before being placed back by a 180 degree turn and into the starting position. It becomes much harder to find these things for a longer algorithm, but knowing something peculiar about an algorithm help you reconstruct it as you go.
Another basic method is to "chunk" each algorithm into smaller pieces that can be excuted in one move. Each chunk should be easy enough to memorize, and then all you need to do is to piece them together to reconstruct the entire algorithm. Chunks can be any finger trick like RUR', R'UR, or B'R'U.
Mnemonics can be useful for getting an algorithm to at least stick in your mind before getting it internalized with muscle memory. Some people associate each move to a sound so that an algorithm becomes a "word." While they can be helpful, these systems should only be thought of as training wheels. Speedcubers need to be able to react instantaneously to any learned case, so the memory needs to be in the muscle more than as letters or sounds. Also, splitting up the moves by sound usually gives a different result than the grouping based on finger tricks.
Here's an analogy I like: mnemonics for a foreign language. I often use creative connections to learn new words, and that fine for taking vocabulary tests in class. But it's not as useful in conversation, where I have to come up with words very quickly and put them together in a natural manner--kind of like finger tricks in speedcubing.
Of course, there's always repetition. This could be simplified with the help of a training program like the last layer trainer.
Other Speedcubing Methods (to be edited)
Although most top speedcubers use a Fridrich-based method, there are a handful of cubers who use very different methods. Sub-15 average has been shown to be possible with each of the methods in this section.
Besides Jessica Fridrich, there are two other cubers who competed in the 1982 World Championship in Hungary and has come back to compete after 2004. Each of them invented their own method.
- Razoux Shultz Method (CLL/ELL)
-
After completing the first two layers, this method finishes off the last layer in two steps, as in Fridrich, but using CLL, which permutes and orients the corners, followed by ELL, which permutes and orients the edges while preserving the corners. Since the Fridrich F2L is usually used for the first two layers, this approach is used not as an entirely separate method but as an alternative to Fridrich last layer. In this context, it is usually simply called CLL/ELL.
- Petrus Method
- Invented by Lars Petrus, this method relies less on algorithms and more on building blocks; the first step, for example, is to build a 2x2x2 block. At the same time that this is more intuitive than Fridrich F2L, it usually takes longer to master all the various block building techniques.
- Waterman Method
- Marc Waterman invented this advanced corners-first method during the first cube crazy. For many years, his 17 second average remained one of the fastest in the world using any method. Watermaan has not competed since the 80's. A description of this method can be found at Rubikscube.info.
- Roux Method
- Gilles Roux invented this method in 2003 partly in reaction to the growing popularity of the Fridrich Method, which relies on a large number of algorithms. Extending the block approach of the Petrus Method even further, this method starts off with a pair of 1x2x3 blocks.
Beyond Fridrich(to be edited)
(hmm note the recurring idea of combining steps by adding more algorithms)
When I entered the world of speedcubing in 2002, sub-20 average was the holy grail achieved only by a handful of speedcubers around the world. Over the next few years, this barrier dropped to 17, 15, 14, and 13--times that many believed neared the limit of the Fridrich method. While Fridrich has remained mainstream, cubers have developped a number of new system in search for faster methods. Among these, what seemed the most logical extension of Fridrich was Zborowski-Bruchem method, or ZB, which combined another step in the last layer by dramatically increasing the number of algorithms.
Although several cubers attempted to learn ZB, or at least ZBF2L (the ZB version of F2L), as far as I know there is no one as of yet who has memorized the full method. What's more, I, like many, now doubt that ZB would improve the solving time significantly. Although the number of moves used in ZB is very appealing, with such a large number of algorithms, it would take a tremendous amount of time to optimize and master every single pattern to the extent that top cubers have done for the Fridrich method. Now with the fastest cubers recording sub-12 averages with Fridrich, there is no longer any motivation to learn ZB.
So where will speedcubing go? What's the true limit of Fridrich, and is there another method that can surpass Fridrich? These are the questions we're still trying to answer.