Why It's Almost Impossible to Solve a Rubik's Cube in Under 3 Seconds | WIRED

[TNL Cubing]

If you accept that it's possible to perform R U in under three seconds, then you have to accept that it is possible. There is a small, but non-zero chance of a two-move scramble appearing at a competition, which makes it, by definition, possible.

By the way, I like the video a lot

I didnt think competition legal scrambles would allow a two move solve?

 

[Mike Hughey]

I didnt think competition legal scrambles would allow a two move solve?

It actually is allowed.

4b3) Specification for a scramble program: An official scramble sequence must produce a random state from those that require at least 2 moves to solve (equal probability for each state). The following additions/exceptions apply:

4b3a) For blindfolded events, the scramble sequence must orient the puzzle randomly (equal probability for each orientation).
4b3b) 2x2x2 Cube: The (random) state must require at least 4 moves to solve.
4b3c) Skewb: The (random) state must require at least 7 moves to solve.
4b3d) Square-1: The (random) state must require at least 11 moves to solve.
4b3e) 5x5x5 Cube, 6x6x6 Cube, 7x7x7 Cube, and Megaminx: sufficiently many random moves (instead of random state), at least 2 moves to solve.
4b3f) Pyraminx: The (random) state must require at least 6 moves to solve.

In my opinion this is something we should really change. It might be a little difficult to add some sort of solver to TNoodle to check for more moves to solve with the more complicated puzzles, but it should certainly be possible - we should have some reasonable minimum greater than 2 moves. But I realize there is a technical barrier to this (although I don't think it's really all that difficult) - someone would have to write the software to check for legal scrambles, even for puzzles like 7x7x7 and megaminx. In any event, there's no real barrier to increasing the minimum number of moves for 3x3x3. If we have minimum moves for 2x2x2, I don't see why we wouldn't have some minimum number for 3x3x3 as well. Back when we didn't have minimum moves for any puzzles, I agreed with those who argued for allowing 2 move scrambles, since choosing a minimum number seems arbitrary. But now that it's already been done for other puzzles, I see no reason to not apply it to all puzzles.

 

[xyzzy]

In my opinion this is something we should really change. It might be a little difficult to add some sort of solver to TNoodle to check for more moves to solve with the more complicated puzzles, but it should certainly be possible - we should have some reasonable minimum greater than 2 moves. But I realize there is a technical barrier to this (although I don't think it's really all that difficult) - someone would have to write the software to check for legal scrambles, even for puzzles like 7x7x7 and megaminx. In any event, there's no real barrier to increasing the minimum number of moves for 3x3x3. If we have minimum moves for 2x2x2, I don't see why we wouldn't have some minimum number for 3x3x3 as well. Back when we didn't have minimum moves for any puzzles, I agreed with those who argued for allowing 2 move scrambles, since choosing a minimum number seems arbitrary. But now that it's already been done for other puzzles, I see no reason to not apply it to all puzzles.

I wrote a thing about this in another thread (where it was off-topic and maybe I shouldn't have written it then):

I mean, my personal opinion is that the move count floor should be raised from 2 moves to something like 13 or 14 moves, just to make it more consistent with the faster events that have a floor above 2 moves. That still leaves room for crazy stuff like just a straight up V perm (skipping F2L and OLL altogether), and then we might start arguing over whether optimal move count is really the metric to be filtering by… I don't really have a strong opinion on what the Correct way of handling this is, but I believe dismissing the problem as being too rare to care about (as the WCA has done) is a completely legitimate decision.

Big cubes also pose a computational challenge if you want a "comparable" move filter (i.e. something like 30 moves for 5/6/7/mega); it's just not possible with current hardware to even implement such a move filter and have the running time be measured in seconds rather than months/years/decades/universe-lifetimes. (Remember, people already complain about 4×4×4 random-state scrambles taking forever to generate, and those take only seconds.)

 

[Fukuoka Kengo James]

Ironically, the UWR is now 2.82 seconds.

Who have done that?

 

[AbsoRuud]

Who have done that?

Kim Roger Høyland Larsen

 

[efattah]

I agree that the 2-move scramble is a problem, but the problem goes much deeper. As the years go by, singles will be ruled more and more by luck until, theoretically, 1000 years from now, if the Earth still exists and cubing still exists, it would take billion-to-one luck to get a single record. Sometime between now and then, either single records will be ignored/eliminated, or a new definition of a single solve will have to be created. For example, the person's single time, in seconds, divided by the minimum number of moves that it takes to solve the scramble. So if you pull off 3.01 on a scramble that has a minimum solve of 20 moves, your 'score' is 3.01 / 20 = 0.1505 seconds/moveminimum. If someone did 2.82 seconds on a scramble that could be solved in 4 moves, their score is 2.82 / 4 = 0.705 seconds/moveminimum, with lower 'score-times' being better.

Another method would be to 'normalize' the time by the ratio against God's number, so if the scramble needs 20 moves to solve, no adjustment is done, but a scramble with a 10 move solution has its time doubled.

 

[FJT97]

It actually is allowed.... Text.

I think before we talk about such unlikely scrambles, we should talk about a timer that is actually good. The stackmat timers have proven to be unaccurate timer fails happen way too often. Pinging @Petro Leum to rant about this

 

[Ronxu]

I agree that the 2-move scramble is a problem, but the problem goes much deeper. As the years go by, singles will be ruled more and more by luck until, theoretically, 1000 years from now, if the Earth still exists and cubing still exists, it would take billion-to-one luck to get a single record. Sometime between now and then, either single records will be ignored/eliminated, or a new definition of a single solve will have to be created. For example, the person's single time, in seconds, divided by the minimum number of moves that it takes to solve the scramble. So if you pull off 3.01 on a scramble that has a minimum solve of 20 moves, your 'score' is 3.01 / 20 = 0.1505 seconds/moveminimum. If someone did 2.82 seconds on a scramble that could be solved in 4 moves, their score is 2.82 / 4 = 0.705 seconds/moveminimum, with lower 'score-times' being better.

Another method would be to 'normalize' the time by the ratio against God's number, so if the scramble needs 20 moves to solve, no adjustment is done, but a scramble with a 10 move solution has its time doubled.

Optimal solution length is not a good indicator of the difficulty of a scramble with a speedsolving method. EO and blocks are much better, but have fun quantifying them in a fair way. Or, you know, just look at the time instead of trying to fix a problem that doesn't even exist.