How to execute algorithms quickly

[Loiloiloi]

This is not a question about how to execute certain 3x3 LLs quickly, or anything like that, which is why I created a thread seperate from: https://www.speedsolving.com/forum/threads/how-to-increase-execution-speed.3134/

Instead, I'm wondering about how people generally learn how to do movements/algorithms speedily. There's a huge transition for me between learning to do an algorithm from muscle memory, and learning how to do it quickly. For example, I've been able to do Ja for a long time, but it still takes me around 4 seconds (a lot of times it's quicker to just do Jb + U + Ab + U'), and now that I'm learning square-1 algorithms I find this issue arising again. Please share your personal experience of how you transition from slow recall and execution, to near-instant algorithm execution, and also any tips you might have.

 

[SolveThatCube]

When I learn a new alg I keep performing it slowly over and over again until I can remember it. Then I just keep doing it faster and faster until I've done it about 100 times and it's sub-1.5

 

[Rnewms]

Try different grips and fingertricks. Watch someone do Ja in slow motion; they probably do it differently. Sometimes I post written algs that are nice with a specific set of movements, but they're usually ignored if they look ugly/require odd turns. Clips of execution would probably give better reception than reading the algs.

 

[Loiloiloi]

Try different grips and fingertricks. Watch someone do Ja in slow motion; they probably do it differently. Sometimes I post written algs that are nice with a specific set of movements, but they're usually ignored if they look ugly/require odd turns. Clips of execution would probably give better reception than reading the algs.

This is usually how I end up learning to do them quickly, like the Feliks F' used in algorithms. Chris Olson's series is especially helpful. I was just curious how people do all their algorithms so quick without these resources available. There's obviously no videos dedicated to every single ZBLL algorithm or every EP square-1 algorithm, but people can still sub-3 these.

 

[Rnewms]

I was just curious how people do all their algorithms so quick without these resources available.

With experience, you know which combinations of moves flow nicely and how. If the popular alg is not obviously good, you try something like R' U' R F R2 D' R U R' D R2 U' F' with D' as left ring and F' as right ring.

Thought process like:
"It would be good if I could free up my right hand for the regrip before R U R'."
"The F' would be slow with left index after R2 or a right thumb after a slow R2', but it seems I can just treat R2 U' F' as R l F' D'."

 

[shadowslice e]

There are 3 rules when it comes to finding moves that work well:
1) Try not to regrip (maybe once per alg) the thumb trick F/F' is quite a good example of this.
2) have quite a few ways of doing the same move: this is often what helps you avoid regrips. For example, to do a U I have several options such as reverse flick with my left hand, standard flick with my right or push on the LBU sticker.
3) Learn which sequences move well together. For example, if I see the moves R' F, it's quite clear that a good way to do this is to do a standard R' followed by an index push like I would in U normally. Similarly, if I see an R B, I would be more likely to do an l U as that's where my hands would be.

 

[Hssandwich]

I like to imagine myself doing the algorithm when looking at it when I'm looking for algs, this is helpful when you are looking through lots of algs to find a good one (cube explorer, algdb, etc).