Expert advice needed on algorithm learning

[efattah]

For those solvers with very high TPS, I am wondering how long does it take you to learn a new algorithm to the point that you can execute it with perfect (ingenious) finger tricks at maximum TPS. Bindedsa talks about learning an algorithm in 10 minutes. While I am capable of memorizing the algorithm in that length of time, I am nowhere near being able to execute it ultra fast, and on complex algorithms just deciding which finger tricks are the fastest can take an hour or more, FOR A SINGLE ALGORITHM. Even then I might find better finger tricks months later causing a frustrating re-learning process.

Is it simply a matter of the number of years of speed cubing? Perhaps after many years so many patterns are familiar to the fingers that they can adopt a new algorithm as a 'combo' of previous triggers?

An example of these challenging to finger-trick algorithms:
M2 U' r' R' U M U' r2 U = M2 U' M R2 U M U' r2 U

 

[Tao Yu]

I'm not surprised Bindedsa can learn an algorithm in 10 minutes. I've learned about 300 of his ZBLL algs, and they're all really good - there are hardly any that are awkward to fingertrick. There are also a lot of familiar fingertricks to be found in them. Tons of the algs are just two OLL algs put together, a sune that cancels into an OLL, a sune that cancels into another ZBLL, taking out a pair and inserting it with WV/VLS and so on. I would bet that most of the time is spent on recognition.

The easy algs of Bindedsa's ZBLL probably take me less than 5 minutes to drill, and the slightly weirder ones up to 10 minutes.

I don't have a lot of experience with awkward algs, but I find it helps to learn more fingertricks for each move. For example to do an F move, I am able to do all the following fingertricks: left thumb, right thumb (push UFR to RFD), right index flick, left fourth finger (flick DLF to LFU), right fourth finger (push RDF to DLF). For U/U' moves, I also do a lot of OH flicks, OH pushes and sometimes push the top layer with my four fingers if a U move comes after an R2' move. This way, I have more options when a U or F move comes my way. Other than that though, I would just experiment a lot, try as many fingertricks as I can think of, and if possible, gen a better alg - after all, some algs are simply not very good.

 

[efattah]

Thanks for the reply! So I guess it comes down to a large arsenal of finger tricks, as well as the simple fact that some algorithms are 'easy' to learn and master and others are much harder.

Some of your finger tricks are pretty neat-- I wonder is there a reference somewhere for really advanced finger tricks?

 

[obelisk477]

For those solvers with very high TPS, I am wondering how long does it take you to learn a new algorithm to the point that you can execute it with perfect (ingenious) finger tricks at maximum TPS. Bindedsa talks about learning an algorithm in 10 minutes. While I am capable of memorizing the algorithm in that length of time, I am nowhere near being able to execute it ultra fast, and on complex algorithms just deciding which finger tricks are the fastest can take an hour or more, FOR A SINGLE ALGORITHM. Even then I might find better finger tricks months later causing a frustrating re-learning process.

Is it simply a matter of the number of years of speed cubing? Perhaps after many years so many patterns are familiar to the fingers that they can adopt a new algorithm as a 'combo' of previous triggers?

An example of these challenging to finger-trick algorithms:
M2 U' r' R' U M U' r2 U = M2 U' M R2 U M U' r2 U

I think the problem you're having is that you're learning algs for (presumably) LMCF, and most of them are new and don't belong to the standard RUF or RUD subets that CFOP is accustomed to using. If you're only ever using R, U and F moves, you're bound to get enough repeats of triggers to remember them across a large enough set of algorithms.

To expand on that, the number of different types of triggers of length, say, 3 or 4, increases with the 'gen' of the algorithm type you are learning. So for RUF or RUD, there are only (3*3)*(2*3)*(2*3)=324 sets of 3 move triggers that are possible, and even less since people tend to avoid F moves when choosing an algorithm. But with RrUM (4-gen) which is necessary for LCF, that increases to (4*3)*(3*3)*(3*3)=972, which is triple the number of three move triggers as a 3 gen alg, which makes it that much more difficult to memorize and repeat the same trigger.

 

[Tao Yu]

Thanks for the reply! So I guess it comes down to a large arsenal of finger tricks, as well as the simple fact that some algorithms are 'easy' to learn and master and others are much harder.

Some of your finger tricks are pretty neat-- I wonder is there a reference somewhere for really advanced finger tricks?

I mostly just picked them up over a long time, from watching videos and experimenting with stuff myself, but Chris Olson and J perm's videos look pretty good. There are also many other interesting looking videos if you search "Rubik's cube fingertricks" on youtube. Be open minded, and remember that some fingertricks may be hard to learn at first, but worth it in end.

Daniel Lin has some interesting looking fingertricks for edge 3-cycles on his channel. I'm not sure how useful they would be for LMCF, but it's probably worth having a look at.