How do I become a sub-30 cuber?

Question

I am a sub-50 cuber right now. I am currently using the Fridrich method, I take about 2-5 secs to solve the cross and I am learning the 21 PLL algorithms. I am learning one or two PLL every day. I am using the 2 look OLL that uses 10 algorithms in two steps.

With all these have only a record of 32 secs and a average of 46 secs. Do you have any tips on becoming a sub 30 cuber or even sub 20 or sub 15?

Answer 1

Practice the speed of your algorithms specifically.

Learning faster algorithms is great, but when people are practicing speedcubing, the single most important thing people overlook is that the majority of their time is spent executing the algorithms.

Increasing the speed with which you execute your algorithms will drastically increase the speed at which you solve the puzzle, regardless of the algorithms you're using. The beginner's method takes me about 30-40 seconds to execute - and I'm not using OLL and PLL. I did it simply by increasing my speed of execution. Then I learned OLL/PLL.

---

Take the algorithms and steps you know, and push the speed faster. Time just the execution of one algorithm (or, if it's too fast to time, instead time several iterations of it). Work on increasing the execution speed of the algorithms you're using. Determine which ones take the longest, and practice those.

For instance, here's an example using a couple of the slow algorithms I created (though you might want to do something like ten tests):

  Edge 3-cycle | 3 iters |  9.27 | 10.86 | 10.54 |  9.99 | 10.40 |
Corner 3-cycle | 3 iters | 12.76 | 10.79 | 11.64 | 10.64 | 10.34 |

Therefore, the edge 3-cycle takes me an average of 3.4 seconds, and the corner 3-cycle takes me an average of 3.7 seconds. (These times are very slow, but that's primarily because I'm seriously out of practice and using a poorly-lubricated, old cube.)

---

1. Index the steps and algorithms you know (for instance, make the cross a step and F2L a step, then time specific last layer algorithms)

2. Identify sequences that seem abnormally slow, or abnormally inconsistent.

   You can do this with a standard error. From the above, my edge-cycle error is 0.244, and my corner-cycle error is 0.392. This betrays the fact that corner-cycles are drastically less consistent than edge-cycles.

   A second run of 5 shows that I'm only a little out of practice, and that I really should relubricate and retension this cube. My average time for the corner-cycle dropped 0.5 seconds, and the error dropped to 0.273.

3. Practice those sequences' execution times until they're much more consistent.

4. Try to identify where you might be getting caught - is one of your rotations going too far? (Chances are, at least one rotation is consistently out of alignment.)

This method should help you identify the slow points of your solution and improve those specifically. Then, practice tying the sped-up methods together.

Your visual processing time can't easily be trained, and that's just something that's going to drop the more you solve the cube.

Answer 2

One thing that is EXTREMELY important is improving your look-ahead! Look-ahead is your ability to foresee future cases and therefore prepare the algorithm for those cases so you can coordinate a pauseless transition from your current algorithm to the foreseen one.

THIS is what separates an above 30 from a 20 and sub 20 and even more so a teen to a sub 10.

There are multiple ways to improve your look-ahead and the most common one is to use a metronome while solving. Basically you solve your cube to the beat of a metronome where you may only make one move per beat and that's it. The reason for this is so you can look at the rest of the cube while doing an algorithm, this is a necessity for increasing look-ahead skills. Eventually you can increase the speed of the metronome more and more at your own pace.

Basically you want to minimize the time you spend just looking at the cube and maximize the time spent physically turning its faces.

I'm a sub 20 cuber and look-ahead makes the difference between a 20+ solve and a 13.

Also may I ask what kind of f2l you are doing? Depending on your methods i may be able to lend some more help.