Throwing out scrambles in competitions?

[StachuK1992]

It's been noted a few times now that scrambles have been thrown out in competitions because they were 'unfair' in various ways.

I'd like to move the main discussion here.

-What's 'unfair'?
-Should we continue doing this? If so, shouldn't we have an official rule stating this?

-A list of previous deletions would be neat.


On a personal note, deleting 'unfair' scrambles is silly; if they happen, they happen.
Different judges throwing out scrambles at different takes of the idea of 'unfair' in itself is quite unfair.

[Vincents]

I don't believe scrambles should ever be thrown out, except in cases where there is a violation of WCA regulations. E.g. this week at Stanford Fall 2011, Jeremy threw out all the 3x3 scrambles because they were all the same thing.

If you start throwing scrambles out, you have to have some definable standard by which you are basing your judgement on. Such a standard does not exist. Every single scramble can be easy or difficult, depending on what method, event, or puzzle is being scrambled. I believe WCA scramble generators generate random states. Cutting off certain scrambles because of perceived ease-of-solution basically makes this a nonrandom state; instead of solving cubes from random places, we would be solving "challenges."

[kinch2002]

I was actually thinking of making this myself so thanks

One thing I am sure of: A clear ruling needs to be made one way or the other.

Clearly there is some confusion about whether scrambles should be thrown out or not. Today I thought about it, and have decided that nothing should be thrown out. It's too difficult to draw a line somewhere, so we'd better not draw one anywhere. Most people realise that 2x2 singles (and pyra single to some extent) mean very little so I don't think it's really a problem at all to let some people get insanely lucky and set crazy times.

One problem I can think of is this: What happens if a cube is already solved (and I mean actually solved, not +2)?

[Cubenovice]

I am not in favor of delegates "manually" sorting out scrambles.

I suppose scramble programs can be set up so the scrambles are rejected (and thus a new one generated) when the scramble length is below a certain threshold value.

2x2x2: 5 moves (yep, you will still have lucky cases like Sunes and other OLL's)
3x3x3: 13 moves
etc etc
The treshold values would have to be set by the WCA.

edit for clarification:
the programs should offcourse not even show the scrambles below the treshold length.

[Mike Hughey]

I think that for random state generators (which we should use as much as possible - we really need to go to one as soon as possible for square-1, for instance), a minimum distance from solved would be acceptable. I personally don't care what that minimum distance is, as long as it's:
1. greater than zero (so as not to allow Daniel's case of a cube already solved), and
2. less than or equal to the smallest minimum distance that we know to have already existed in a competition - if this is done, then it's reasonable to maintain the current records as they are reasonably.

That means that, for instance, with 2x2x2 I'd be fine with minimum distance being set anywhere from 1 to 4 moves. (That was a 4 move WR, wasn't it?)

And I agree that the scramblers simply shouldn't generate any scrambles with shorter distances from solved than the minimum - if they do, they can always silently generate another until an acceptable one is found. So the person generating the scrambles is never even aware such a decision was made.

[asportking]

I think that for random state generators (which we should use as much as possible - we really need to go to one as soon as possible for square-1, for instance), a minimum distance from solved would be acceptable. I personally don't care what that minimum distance is, as long as it's:
1. greater than zero (so as not to allow Daniel's case of a cube already solved), and
2. less than or equal to the smallest minimum distance that we know to have already existed in a competition - if this is done, then it's reasonable to maintain the current records as they are reasonably.

That means that, for instance, with 2x2x2 I'd be fine with minimum distance being set anywhere from 1 to 4 moves. (That was a 4 move WR, wasn't it?)

And I agree that the scramblers simply shouldn't generate any scrambles with shorter distances from solved than the minimum - if they do, they can always silently generate another until an acceptable one is found. So the person generating the scrambles is never even aware such a decision was made.

This would work for smaller sized cubes, but what about, say, a 7x7? It could have an entire center solved and the minimum distance would still be pretty close to the minimum distance of a totally scrambled 7x7.

[Mike Hughey]

This would work for smaller sized cubes, but what about, say, a 7x7? It could have an entire center solved and the minimum distance would still be pretty close to the minimum distance of a totally scrambled 7x7.

I think that in the extremely unlikely case that such a thing happened, the appropriate thing to do would be to simply allow the scramble. Honestly, I don't expect this to even apply to something as big as a 7x7x7 (or even a 5x5x5), since random state scramblers are not likely to become practical anytime soon (if ever). For puzzles too big for a random state scrambler, we should simply allow any scramble.

[Kirjava]

Say no to scramble filtering

This seems like a nice solution. However, I'm not against allowing 1 move scrambles either.

I see these as the only valid solutions to this 'problem'.

[Stefan]

See also: http://worldcubeassociation.org/foru....php?f=4&t=954

I don't believe scrambles should ever be thrown out, except in cases where there is a violation of WCA regulations. E.g. this week at Stanford Fall 2011, Jeremy threw out all the 3x3 scrambles because they were all the same thing.

How did that violate the WCA regulations?

Say no to scramble filtering

Why? Why do you want to keep easy/short scrambles?

[Kirjava]

Why? Why do you want to keep easy/short scrambles?

It's too hard to draw a line and say how easy is too easy and how short is too short. I would prefer to be given a solve from a pool of all states, not a subset of them. I don't like it because David Salvia suggested it.

None of this is important.

I don't see why we should be trying to avoid having anomalous results in the first place. You can get lucky - that's speedcubing.