Without practicing, how do you know if one alg is faster or not? Like, If I'm good with a specific Y perm for example, how would you determine if you should invest time into learning and practicing a specific perm that is shorter and doesn't feel as good if you're not 100% sure it'll be faster? Or is it more of a trial and error thing? Just learn it anyway so you know if it's faster?
I can only tell you what I did, I'm sure there are many other ways. Around this time I started timing my PLL's and taking averages of 12 for each PLL case. It was around this time that I discovered that algs that I thought were my fastest cases were sometimes not (like my 17 move Y perm), and other surprising ones were more consistently fast despite not "feeling" fast. You might even want to do 20-30 solves with a particular PLL alg and take a session average. Whenever you start to attempt to learn a new case, and are somewhat comfortable with it, start timing it. By timing I mean timing each individual execution of the alg as a separate attempt.
You can usually tell pretty quickly how good an alg will be by this point. If you perform the alg, lockup a bit at the end, pause a bit in the middle to remember, etc. but the time is still relatively close to the time you get on your old alg at full speed, then your old alg sucks and the new one is amazing.
If the new alg starts out slower than your old alg, and even as you get smoother and quicker with the execution you still have not gotten even one single time under your "average" time for your old alg, then I would probably not invest much more time into the new alg, unless you are still locking up a lot (and thus you just need to improve turing accuracy/recall/etc. for the new alg).
Again those are only things that helped me. Remember, your brain will find it's own prejudices to algs it thinks are fast. The timer, however, never lies as to which alg is faster.
Chris
