The current blindfold Rubik's Clock UWR is 1:36.88. This is using Stefan Pochmann's method, which relies on memorizing the initial states and then solving each clock separately. There is a youtube video of this solve; the solver takes under 40 seconds for memorization, and the rest is the solve itself.

I have an idea for a method that could be faster than the Pochmann method, but would require a lot more practice. The idea is to solve the Clock in a speedsolve-like way, that is, with as few turns as possible, since turns take a long time when you have to feel what you're doing.

Now here's the trick. You're still memorizing 14 numbers, but instead of memorizing the clock faces you memorize specific sums and differences of the clock faces. Then, you turn (view the back), solve the back cross in 5 moves, turn to the front, solve the front cross in 5 moves, and finally do the corners in 2 moves each (because doing them the speedsolve way is too much arithmetic to do in a minute). That makes a total of 18 moves, as opposed to the Pochmann method's 46 (I believe).

Out of the 14 numbers, the first 10 are the exact amount you turn the first 10 moves, and then the last 4 are the corners and are solved by the Pochmann system (and, of course, you can do this first for added speed, as even though the corners won't look solved they will be by the end).

I haven't quite worked out the arithmetic, but I'm pretty sure it's reasonably simple (it's all modulo 12, so if you want you can do it by imagining a clock's hand moving). Although it will take a while when you start out it will be a lot easier when you get used to it. Memorization is a little tricky too, since you don't want to do a calculation more than once, so you have to go through the sequence and memorize it slowly rather than try to memorize the whole thing in one go.

Would anyone be interested?

Hi

Stefan Pochmann's method is already pretty much optimised.

The main problem as i see it is to perform the exact amount of turning the knob, not to overturn or underturn. With cube this problem is trivial (turn a layer not too much or too little), with clock it's not

Remembering differences is a good idea, but im quite sure this must be in use already?

Just my 2 cents (im not into any kind of blindsolving yet)

-Per

Well, I don't know if arithmetic is in use, because the top two people on speedcubing.com use Stefan's method. There might be a little arithmetic for them, but if you want to do it in half the turns you'll need a lot more than that. My method requires something like 30 additions or subtractions at the present, and I'm trying to figure out if that can be decreased.

Pochmann's method is more or less optimized; there are of course things that can be done a bit faster if you're willing to do some arithmetic. For example, if you memorize the sum of the numbers on the 4 corners as well as the 4 corner numbers, then instead of having to do 8 moves for the corners you just have to do 5 (because you turn all the clocks once instead of 4 times). You could do this on my solution too, if you want.

I agree that the main problem is to perform the exact amount of turning. It's like if someone asked you to do U7 on a well-lubed Eastsheen 2x2x2 blindfold, you'd probably take at least 2-3 seconds, because you can't tell when you've got U7 and not U6 or whatever. That's why I tried to lower the number of turns you do without increasing the amount of information you have to memorize. If you can do one sum or difference faster than one turn (including the selector pushing) then my method is probably faster than Pochmann's.

I haven't done any real blindsolving yet... and I don't have a clock yet... but I like thinking about it. I'm sure this is easier than the 3x3x3 anyway.