Humphrey Wittingtonsworth IV | Tran Algorithm Update @HumphreyWittingtonsworthIV | Uploaded September 2018 | Updated October 2024, 8 hours ago.
I've tested a ton of different variations and iterations such as RUF, RUD, Minimising, 2phase with no L /B moves, and the best result was with RUD with a filter.
I just ran my program for several minutes, constantly spitting out 3 phase mostly RUD scrambles, and only outputted scrambles 25 moves and below. Even though mathematically the AVERAGE was higher than 25 moves, there are millions, if not hundreds of millions of random states that can be achieved at 25 moves and below.
So then I did an average of 50 with normal scrambles, and then with Tran scrambles, and I found that my scrambles took 0.8 seconds slower.
The reasons are pretty simple:
1. The moves were too ergonomic, and I turned faster than my eyes could read the scramble. With normal scrambles, it's slow enough that I can think ahead and prepare the next move.
2. Certain sequences were still NOT ergonomic such as (U2 D) or (D2 U), which led to a slowdown of turning. It was nice when (U D') or (U' D) happened, but it wasn't enough to make up for the difference.
3. I've been doing 2-phase scrambles for years, and most people are already super good at scrambling normally, in order for the maximum benefit to occur, people would need to train RUD scrambles for years, and I don't think people are willing to change.
4. Ergonomic algs are fast because people practiced them, but with scrambles, you're going in blind. Most cubists aren't good at doing the moves and sequences on the fly.
Even with minimising, which is testing a longer phase 1 to get a shorter phase 2, this never led to more than 1 move of difference with three phases, and 90%+ of cases couldn't be minimised.
It was a fun project, but intensely frustrating that it didn't work out.
Time to work on other projects now! :)
I've tested a ton of different variations and iterations such as RUF, RUD, Minimising, 2phase with no L /B moves, and the best result was with RUD with a filter.
I just ran my program for several minutes, constantly spitting out 3 phase mostly RUD scrambles, and only outputted scrambles 25 moves and below. Even though mathematically the AVERAGE was higher than 25 moves, there are millions, if not hundreds of millions of random states that can be achieved at 25 moves and below.
So then I did an average of 50 with normal scrambles, and then with Tran scrambles, and I found that my scrambles took 0.8 seconds slower.
The reasons are pretty simple:
1. The moves were too ergonomic, and I turned faster than my eyes could read the scramble. With normal scrambles, it's slow enough that I can think ahead and prepare the next move.
2. Certain sequences were still NOT ergonomic such as (U2 D) or (D2 U), which led to a slowdown of turning. It was nice when (U D') or (U' D) happened, but it wasn't enough to make up for the difference.
3. I've been doing 2-phase scrambles for years, and most people are already super good at scrambling normally, in order for the maximum benefit to occur, people would need to train RUD scrambles for years, and I don't think people are willing to change.
4. Ergonomic algs are fast because people practiced them, but with scrambles, you're going in blind. Most cubists aren't good at doing the moves and sequences on the fly.
Even with minimising, which is testing a longer phase 1 to get a shorter phase 2, this never led to more than 1 move of difference with three phases, and 90%+ of cases couldn't be minimised.
It was a fun project, but intensely frustrating that it didn't work out.
Time to work on other projects now! :)
