Memorizing Algorithms

I'll make a video on this someday but not yet (I have my reasons).

Obviously there are memory tricks you can use like grouping common triggers (R U R' etc.) but that should just come naturally with enough experience and enough algorithms memorized in the past.

What I'm talking about today is the feeling of "I know this alg, I practiced it a ton, but I have no clue how it starts and I can't recall it right now" which then ruins your solve.

I learned a chunk of ZBLL a long time ago, and to this day I only remember some of the easy cases that were part of it. Even while I was learning and practicing them every day, I just had so much trouble recalling them quickly in solves. This is obviously worse for ZBLL than it is for a more common set like PLL or OLL, but some of those cases are extremely rare as well, or are just hard to memorize for other reasons.

G perms are very common (4/72 per G perm, 16/72 chance of any G perm) but they are all so similarly executed and recognized that they are hard to memorize. N perms are 1/72 each, the triple bar OLL is 1/108, and the X-OLL is 1/216.

Let's quickly talk about how memory works. Ironically, I learned about this before and forgot it a bit so this is just from my memory and may not be 100% how memory works, but it should be good enough.

Memory kind of works like a computer's hard drive + RAM, where you have long term and short term memory. Short term memory is important because it's easy to put something there and easy to throw it out as well, to make sure your brain isn't clogged with useless stuff you only needed once. When you memorize a G perm, your brain doesn't care how important the algorithm is. Even if you practice it for 10 hours straight and become a G perm god, if you go to bed and get it in your 10th solve the next day you won't recall it right away. There's a good chance you'll completely forget it as well. That's short term memory at work.

There are 2 key aspects I've noticed that help with this. The first has to do with converting to long term memory. The way to convince your brain that something is important in the long term is to frequently encounter it and force your brain to RECALL it. The important thing is recall. Doing an algorithm for 10 hours straight doesn't count as recall, because it never leaves your short term memory. You have to go do something else to discard the info from your short term memory, and come back to it frequently so that your brain realizes this is actually important.

The second thing is having an association trigger your memory. Certain blocks or patterns of an algorithm should be associated with the first move (or first few moves) of the algorithm, which quickly reminds you what the rest of the algorithm is. For example, I remember that for one of the G perms, I regrip to hold a block in my right hand and do R U R' U' D on that block. You can't develop long term visual triggers without frequently having to recall an algorithm from seeing the case. If you just do the algorithm over and over without looking, again, that's not going to help. You have to see the case and associate that with the start of the algorithm.

So there's long term memory, and there's visual associations.

Putting it all together, here's how I practice algorithms.

I find a trainer (which will be my website once that's done, but for now just google OLL trainer or something like that) and then if possible, pick a small subset of algorithms to work on. The most I can learn at once is usually around 3-4 without feeling exhausted or counterproductive. Practicing 1 algorithm only is fine if you're relatively new to learning algorthms, but if you're at the level where putting an algorithm into short term memory is easy, then doing a few more at once is actually better.

I learn a few algorithms and practice them each so that I can do them relatively fast. Then I start training those algorithms. I see the case on my cube and then recall the whole algorithm before starting (trying to do this as fast as I can, which is difficult).

Once I'm decent at recalling the whole algorithm for the cases I'm working on, I take a break and come back later. Then I repeat this until I'm pretty good at it.

Then the better I get at it, the more I can just begin the algorithm by recalling the first move rather than recalling the entire algorithm. Again, I take breaks often to force the algorithms out of my short term memory to recall again later. This builds long term memory quite quickly.

I do this for at least 1 session every day (possibly 5-10 mins?) or more if I have time, just to keep recalling it. I'm doing this to convince my brain that these algorithms are important and should be remembered.

Lastly a disclaimer: this doesn't actually create stable long term memories until you have done it for a long time. You can't really cheat time here, but as long as you don't stop using these algorithms, then you essentially simulate long term memory right from the start. The actual super solid long term memory that you'll have a hard time forgetting won't come for a while, but I don't think there's really a way to achieve that level of confidence so quickly. The goal of my method is just to be able to use these algorithms as soon as possible. If you do this perfectly for 2 weeks and then stop for a month, you'll definitely forget the algorithms again.

The most important thing is that 5 segregated sessions of 5-10 mins is way more effective than 1 session, no matter how long that 1 session is. You have to take breaks where you fill your short term memory with unrelated things. Sometimes I'll do homework because that requires a lot of short term memory, and then interrupt my homework with a short training session.