My journey on Speedcubing
One year ago, I decided to train to solve the Rubik’s cube as fast as I could. This article is about how I improved my solving time from over 2 minutes to 40 seconds and my discovery of the speedcubing world.
Let’s make it clear right from the start: I don’t pretend to be an excellent speed cuber. Solving the Rubik’s cube in 40 seconds is very far away from best performances, but I thought nonetheless that I could share my experience from one year of training.
When I was a kid, I learned how to solve the cube in about 2 minutes, using a minimal set of formulas (I used 5 basic ones and their symmetrical). I was happy with that for many years.
One year ago, I read a series of article from a guy who spent a year trying to complete 12 challenges, one per month, on various domains (like landing a standing back-flip or drawing a realistic self-portrait). One of the challenges was about solving the Rubik’s cube in less than 20 seconds, which is considered a very competitive time.
I decided to try to push a bit my limits and see how much I could improve my own time.
The world of Speedcubing
While searching for methods on how to improve, I discovered the speed cubing community.
Like any popular discipline, there are speedcubing associations and clubs, national and international competitions, champions (like Feliks Zemdegs, who can solve the cube in 4 seconds), various categories of solving (3x3x3 cube solving, 4x4x4, 2x2x2, blindfolded, one-hand solving,etc.).
There are many resources on the internet that describe methods and you’ll easily find lots of YouTube channels to help you learn algorithms and finger tricks.
If you want to discover the speedcubing world, I recommend you to read “Cracking the cube” by Ian Scheffler, who explains his own experience of competitive speedcubing.
The CFOP method
The most commonly used advanced method for solving the cube is called the CFOP method. The four letters of its name stand for each of the 4 steps of the method:
- Cross
- First 2 Layers (F2L)
- Orient Last Layer (OLL)
- Permute Last Layer (PLL)
Mastering this method is a huge work, that requires years of intensive training. You have to memorize lots of formulas (that look like this: F-RU’R’U’-RUR’F’-RUR’U’-R’FRF’, each letter representing a rotation of some part of the cube) and training hard to execute them at the speed of light (the above formula would be executed in just a second by top champions).
Happily for me, there are simplified CFOP methods that require a bit less memorization and work, at the expense of speed of course.
My first improvements
During my first 2 months of training, I improved my time by 30 seconds, from 2 minutes to 1 minute and 30 seconds.
I achieved this by 2 means: using a faster cube and learning the 2-look CFOP formulas.
Fast cube
It may seem surprising, but if you want to solve the Rubik’s cube fast, then you must not solve a Rubik’s cube! In speedcubing competitions, nobody is using a Rubik’s cube!
You’ll have to buy other kind of cubes, that turn much faster than the original Rubik’s cube, as they are designed differently. You’ll find easily lots of speed cubes on the market.
I remember the great sensation of fluidity I felt when I tried my first speed cube. I can’t imagine going back to using a Rubik’s cube again now!
2-Look CFOP method
In the CFOP method, you need to learn 57 formulas for the OLL step and 21 for the PLL step. That’s quite a lot of work! Thanks to these formulas, you can solve each of the 2 steps using only one formula, which is very time efficient.
The 2-look CFOP method is a simplified version of the CFOP method, where you only have to learn a small subset of the formulas (10 and 7). But instead of having to execute 1 formula per step, with the 2-look method you have to execute 2 formulas for each step (and sometimes only 1 formula if you are lucky).
Learning these formula was quite some work. Actually, I only learned 4 of the 7 PLL formulas because I’m lazy :)
Despite not executing them in a perfectly fluid way, taking time to recognize which formula to apply, and sometimes even messing up with the formulas, I nonetheless was able to reduce my time to 1m30s.
Further improvements
After these 2 months of work, I wondered where I should focus my efforts to improve further. Should I just learn the complete set of CFOP formulas? I realized along the way that improving my time involved much more diverse efforts than just learning formulas.
An interesting way to study the problem is to look at the time spent in each step of the method, and compare it to the split of a reference solve, like a 20s solve.
I could see that I spent half of the time on the F2L step. I also noted that I was spending as much time on the cross as I was on each of the 2 final steps. Maybe it would be easier to improve my Cross rather than learning tens of algorithms to improve my OLL and PLL time.
After months of work, here’s how my average 50s solve looks like:
Yes, 50s, not 40s as I mentioned earlier. My average solve time is 50s. My best time is 40s, that is when I execute perfectly and when I have favorable positions where I can skip one formula or two.
Here’s below what I did to improve my solving time on each step.
The Cross
Most of the work I did on the Cross was to improve my planning during the inspection phase. Planning? Inspection phase? What?!
In competitions, before you start turning the cube and the timer starts, you have 15 seconds to inspect the cube and analyze what your first moves will be. The best speedcubers are able to plan how they will solve the Cross and part of the F2L during this inspection phase.
To improve on that part, I watched examples of cross solving to understand how it worked. This step is rather intuitive and really not about learning formulas. I had to learn though how to solve this step with the cross on the bottom side (meaning I don’t see the cross when I’m solving it).
Today, I’m able to plan almost the entire cross during the inspection phase, but I still have to check where are some part of the cross when I’m solving it, making me loosing a bit of time.
To improve further, I would need to be able to always find a plan for the cross with 6 moves during the inspection phase and execute it without having to look at the cube. I probably could gain 3 seconds if I focused on this part.
First 2 Layers
F2L is the longest step of a solve. It’s also where I had the biggest opportunity to improve my time.
Solving F2L efficiently requires a combination of intuition, formula memorization, planning of your next moves while solving (which is called “look-ahead”) and finger tricks (how to best position your fingers to make specific moves).
To improve on F2L, I watched videos to understand the intuition behind the F2L formulas and learned a few of them on cheat sheets. This made me reduced the number of turns I needed to execute. I also forced myself to look at other parts of the cube while solving, in order to spot where the next edge and corner that I would solve were.
If I ever want to reach sub-20s solve, I really have to improve a lot on F2L. My main improvement area is on the look-ahead. I have to learn to plan for the next formula while executing the current one, and stop rotating the cube to find what my next move will be or to ease the execution of the next moves. I also have to learn more efficient formulas than the ones I’m using, to reduce the number of moves I make for this step.
Orient Last Layer
This step may be the one I enjoy the most. I feel I execute it fluidly.
To improve from my 1m30s solve, I didn’t learn new formulas compared to the 10 I already knew. But I trained a lot so that I could recognize more quickly which formula I had to apply. By training, I develop my muscle memory, meaning that I would execute the moves of a formula automatically, without thinking about each of the move. I also learned finger tricks to execute the formulas faster, like the “sexy move”, which is called that way because it is really satisfying to execute it:
I feel that the main way to improve on OLL from here would be to learn the full set of 57 algorithms, which seems a huge work to gain just a few seconds!
Permute Last Layer
The last step before completing the cube!
To go from 15s to 10s, I learned 2 more formulas in addition to the 4 I already knew, and I practiced a lot, like for OLL, my finger tricks and my muscle memory to execute faster.
From time to time, I encounter a situation where I have to apply 3 formulas instead of 2 to solve the PLL. To avoid that and gain around 4 seconds in these situations, I would need to learn one more formula. To improve my PLL time more consistently, here again it seems that learning the full set of 21 PLL algorithms is the only way.
Conclusion
So that was the summary of nearly one year of speed cubing practice.
It is really an addictive activity. I always want to practice, improve my time, discover new finger tricks or more efficient formulas.
I realized along the way that reaching sub-20s solve was far much harder than what I thought initially. I also realized that learning formulas was only a limited part of the work required to improve at speed cubing: inspection, look-ahead, memory muscle and finger-tricks are at least as important to develop.
I feel that I can still improve a lot my personal best time, especially by working on the F2L step, but at the same time each second of improvement requires more and more efforts. I’ll probably keep on turning the cube, but I won’t put as much energy as I did throughout the past year, hoping that I’ll be able to bring my PB time to less than 30s within a year.
