Proposal: Regulations to ensure scrambling accuraccy

[henrysavich]

With the 6.88 misscramble, I've gotten to thinking about possible regulation changes in order to increase accuracy, Im not going to write these in "WCA regulation ready form", but I'll try to explain clearly and concisely.

1) Scramblers must sign the scorecards

Very simple but it adds a layer of accountability. The purpose of this is not primarily to make penal consequences easier, but rather that I believe the psychological effect of having to sign will encourage people to take more responsibility in their role as scrambler. I really don't see any reason not to implement this, but i suppose the effectiveness of this system merits some discussion.

2) All scrambles must be checked by a person other than the scrambler, against the pictures provided by the scrambler program.

More effective at ensuring accuracy, but also more of a burden. Fortunately, people don't need to know how to cube to know how to match a cube to a picture. This regulation I feel would best be complemented by the additional regulation that delegates exclusively are allowed to both scramble and check a cube, in order to not hinder efficiency in competition procedure.

More than promoting these singular ideas, I want to encourage a discussion, because I think that mis-scrambles are much more of a pervasive issue than just these high profile occurences may lead us to believe.

 

[quickalt]

2) sounds like it would take up a lot of comp time.

 

[Kit Clement]

We've actually implemented both of these ideas at competitions without these kinds of regulations directly enforced, and surprisingly, there were still scrambling errors at these competitions. The fact of the matter is that scrambling mistakes still occur, and the only thing we can do is plan competitions to ensure that trustworthy and accurate scramblers are available in each round/group of each event. At the very least, we should probably always double or triple check scrambles when someone like Feliks is going to solve it

 

[Julian]

I just had an idea. The pictures to check against should be 3D images instead of cube nets. I think this might get more people to check.

 

[asierrayk]

We need a machine to scramble the cubes.

 

[cashis]

We need a machine to scramble the cubes.

Do you have the money for this?
Maybe at big competitions, but not every one.

 

[EMI]

I just had an idea. The pictures to check against should be 3D images instead of cube nets. I think this might get more people to check.

+1

And, while we're at it: Let's only permit the boy color scheme /s

 

[TyrantCuberKing]

We've actually implemented both of these ideas at competitions without these kinds of regulations directly enforced, and surprisingly, there were still scrambling errors at these competitions. The fact of the matter is that scrambling mistakes still occur, and the only thing we can do is plan competitions to ensure that trustworthy and accurate scramblers are available in each round/group of each event. At the very least, we should probably always double or triple check scrambles when someone like Feliks is going to solve it

I agree that misscrambles are going to occur unless ridiculously rigid measures are taken, but my guess is that the frequency of misscramble will at least be lowered. However, it would be extremely difficult to verify or discredit this because there is no rigorous data on the number of misscrambles per competition controlling for scrambling procedure type.

 

[AlexMaass]

Could we use a webcam hooked up to a computer to check that the scrambles are right?

 

[qqwref]

Could we use a webcam hooked up to a computer to check that the scrambles are right?

This kind of thing could work, and could probably be as fast or faster than a human carefully checking the cube. However, you'd need to hook the scramble program up to the webcam program so it knows which scrambles to look at, and you'd also have to have to be able to handle unconventional color schemes, imperfect stickers, and (at least) both white and black cubes. It would probably take some work to set up, but could be very useful once it's done.

Alex suggested a computer, but I'm thinking a smartphone app - compared to laptops or desktops they have better cameras, are very common and standardized, and are also easier to hold up in the air (point the camera at the cube, rather than trying to hold the cube in just the right place).

 

[AlexMaass]

This kind of thing could work, and could probably be as fast or faster than a human carefully checking the cube. However, you'd need to hook the scramble program up to the webcam program so it knows which scrambles to look at, and you'd also have to have to be able to handle unconventional color schemes, imperfect stickers, and (at least) both white and black cubes. It would probably take some work to set up, but could be very useful once it's done.

Alex suggested a computer, but I'm thinking a smartphone app - compared to laptops or desktops they have better cameras, are very common and standardized, and are also easier to hold up in the air (point the camera at the cube, rather than trying to hold the cube in just the right place).

you could just put the cube in WCA orientation, and the software could easily handle nonstandard color schemes by which color is the U R L D B F color. I was thinking webcams because you can just mount them up on something. Its annoying to hold the phone up for the whole round imo. I was thinking two webcams, one pointed at the ULF corner and the other pointed at the URB corner, so the scrambler doesn't have to rotate the cube twice. The program could be made so it could just read the pdfs.

 

[josh42732]

Why we don't only let perfect cubers compete in competitions? That way, there would be very consistent scrambles, averages, and many problems that we usually have at competitions would just magically go away!?!?! lol jk

But for real though, piggy-backing on what henrysavich originally said, having one person do the scrambles is actually a good idea. I haven't personally been to a comp before, but what if we had selected scramblers, and not have the judges scramble, and then judge? The scrambler would scramble, look at the sheet for every side, then give it to the judge, then go scramble the next cube so that the person scrambling won't have to worry about judging as well, therefore getting more consistent scrambles. And the judge could have a picture at the cube and make sure that the cube is scrambles correctly.

 

[qqwref]

That's how it's already done - scramblers sit at a table and just scramble cubes, then the scrambled cubes are brought to judges.

 

[Animorpher13]

Ideally, we would be able to implement some sort of technology that is more accurate than use humans currently (in scrambling).

 

[Lucas Garron]

Incorrect scrambles are a big problem – probably the biggest "new" problem since 2014.

Several competitions have already used one of henrysavich's suggestions – the results are better, but not perfect.
Read this thread for the results so far.

I don't think else suggested in this thread currently sounds practical. Probably the most realistic is to use laptop/smartphone cameras.
I'd be very interested if someone could demonstrate a system that worked under less-than-ideal lighting conditions, with any color scheme, preferably on multiple puzzles, doesn't interfere with running competitions in practice, allows multiple scramblers to work efficiently with fewer devices, and ideally also doesn't require a human to identify the correct scramble to use (but also doesn't waste a lot of time scanning, or the use of fancy technology prone to breaking).

I agree that misscrambles are going to occur unless ridiculously rigid measures are taken, but my guess is that the frequency of misscramble will at least be lowered. However, it would be extremely difficult to verify or discredit this because there is no rigorous data on the number of misscrambles per competition controlling for scrambling procedure type.

See my link above; we do have some data.

I think the most practical thing is to require one of the two suggestions from the first post, unless the Delegate can convincingly vouch that there were almost no scrambling errors (e.g. very experienced cubers were asked to scramble – but even they can make mistakes).

 

[adimare]

Incorrect scrambles are a big problem – probably the biggest "new" problem since 2014.

Do you think this is new? Could it be that it's just a lot more noticeable now because everyone uploads their solves to youtube for everyone to see?

 

[szalejot]

For me - option 2)
I think judge should check scramble when he/she picks up the cube.
Printing few pages of scrambles more (for judges) which will reside on scramblers table would not increase cost of competition dramatically and it will make second layer of check to scramble.

 

[mDiPalma]

i think we should have the scramble sheet written in qqtimer's "3x3x3 for noobs" notation.

 

[blade740]

This is not a problem that regulations can solve. There is already a picture of the cube state on the scramble sheet. If the cube goes out and it does not match that picture, than the scrambler was lazy. Period.

There simply needs to be a concerted effort on the part of competition organizers and staff to use more experienced scramblers, and to remind anyone scrambling to double- and triple-check each scramble against the picture on the sheet.

 

[Kyle™]

Would it be possible to have something similar to the 'cube solving robot' that we could bring to competitions to scramble cubes in under 5 seconds per cube? 0 mistakes guaranteed.