Throwing out scrambles in competitions?

[Ranzha]

Magic never requires thinking. How often would 2x2x2 not require thinking? So far there has been a single case out of every single round of every single event that has ever been held.

If 2x2 scrambles that can optimally solved in at most two moves are considered trivial, such cases statistically occur 0.0017% of the time.

 

[qqwref]

Magic never requires thinking. How often would 2x2x2 not require thinking? So far there has been a single case out of every single round of every single event that has ever been held.

Yes, it's very uncommon. But it clearly does happen, and since it's unfair, it should be prevented. It's about fairness and having good rules, not about trying to spend our time only dealing with the most common cases.

I don't see how this luck is unfair. Getting lucky is not a crime.

Also, this would only happen once ever, then the playing field would be level.

Getting lucky isn't wrong because it isn't your fault. But if a delegate sees a 2-move 2x2x2 scramble and makes the decision to let everyone in that group/round get an official time on it, that is certainly their fault - and they should be penalized, even if they don't personally get a WR out of it (cough clock cough).

And in what sense would having a 2-move scramble "level" the playing field? Everyone at that competition would have times that are almost unbeatable by everyone not at that competition, and almost unimprovable too. I would not consider entering a competition run by any organization that considers that a "level" playing field.

 

[Kirjava]

Yes, it's very uncommon. But it clearly does happen, and since it's unfair, it should be prevented. It's about fairness and having good rules, not about trying to spend our time only dealing with the most common cases.

I do not believe it is unfair. People keep saying it is. I still don't see how.

Getting lucky isn't wrong because it isn't your fault. But if a delegate sees a 2-move 2x2x2 scramble and makes the decision to let everyone in that group/round get an official time on it, that is certainly their fault - and they should be penalized, even if they don't personally get a WR out of it (cough clock cough).

Getting lucky isn't wrong because luck is an aspect of cubing.

And in what sense would having a 2-move scramble "level" the playing field? Everyone at that competition would have times that are almost unbeatable by everyone not at that competition, and almost unimprovable too. I would not consider entering a competition run by any organization that considers that a "level" playing field.

2x2x2 single would then remain a combination of luck and skill. The scenario you're describing is essentially the one we have right now.

 

[Ranzha]

If the WCA is looking to yield an accurate representation of the skill levels of competitors for the 2x2 event, 2x2 single should be omitted from the records sheet. It is undoubtedly evident that luck significantly overpowers skill when it comes to 2x2 single.

 

[Zarxrax]

Getting lucky isn't wrong because luck is an aspect of cubing.

Luck is certainly an aspect of cubing and there is no removing it completely from the equation. However, I don't see what's wrong with trying to at least reduce the influence of luck, so that there is at least some amount of skill involved in the outcome.
Going to a competition shouldn't be like going to Vegas and hoping to hit the jackpot. It should be about proving yourself based on the hundreds or thousands of hours that you have dedicated towards becoming skilled at something.

 

[Cubenovice]

Originally Posted by Cubenovice
The difference in triviality between 3 and 4 move scrambles is negligible.
The difference in triviality between 4 and 5 move scrambles is negligible.
The difference in triviality between 5 and 6 move scrambles is negligible.
The difference in triviality between 6 and 7 move scrambles is negligible.
The difference in triviality between 7 and 8 move scrambles is negligible.
The difference in triviality between 8 and 9 move scrambles is negligible.
The difference in triviality between 9 and 10 move scrambles is negligible.
The difference in triviality between 10 and 11 move scrambles is negligible.
The difference in triviality between 11 and 12 move scrambles is negligible.

Not sure if all those statements are true.
However! If they were, that doesn't mean that the difference in triviality between 4 and 12 move scrambles is negligible.

That is exactly my point, adding up negligible differences will eventually result in a significant difference with the starting state.
But at what number of repetitions?

 

[qqwref]

I do not believe it is unfair. People keep saying it is. I still don't see how.

It's unfair because someone who doesn't happen to go to one of the few competitions with the 2-move scramble (or whatever) will be behind someone who did go to one of those competitions in the rankings, no matter how skilled they are. That problem already exists, but 4-move scrambles are relatively common, so many competitions have had them; 3- and 2-move scrambles are even rarer but have even more potential to create unreasonably good times. I think that rather than accept this unfairness we should try to minimize it.

 

[Stefan]

It's unfair because someone who doesn't happen to go to one of the few competitions with the 2-move scramble (or whatever) will be behind someone who did go to one of those competitions in the rankings, no matter how skilled they are. That problem already exists, but 4-move scrambles are relatively common, so many competitions have had them

Did some calculations out of curiosity, not necessarily to make a point:

4-movers are indeed "relatively common" to appear in all competitions overall, but for a particular cuber, they're still quite hard to get. Happens once in 1989 scrambles. Nobody has had nearly that many attempts yet. François has had the most with 685, far more than anybody else, which gives even him only a 29.1% chance to have gotten one (or more).
1 - ((1 - (1847 / 3 674 160))^685) = 0.291380125

There have been 1744 rounds of 2x2, almost all with five attempts. Assuming the average round has two groups, that's about 1744*5*2/1989 = 8.77 occurences, I would call that "a couple" rather than "many" competitions.

 

[qqwref]

Maybe not "many", but there are certainly more individual cases than I can recall off the top of my head. An active cuber would still have a reasonable chance of getting a 4-move scramble - it's not like winning the lottery or anything. But a 2-move scramble might pop up once somewhere in the world every decade or more (assuming perfect randomness). So if someone gets a scramble like that you can't just expect others to wait for their turn, because they almost certainly won't have one.

 

[Escher]

Luck is certainly an aspect of cubing and there is no removing it completely from the equation. However, I don't see what's wrong with trying to at least reduce the influence of luck, so that there is at least some amount of skill involved in the outcome.
Going to a competition shouldn't be like going to Vegas and hoping to hit the jackpot. It should be about proving yourself based on the hundreds or thousands of hours that you have dedicated towards becoming skilled at something.

Actually, cubing is quite a bit like no-limit hold'em...

If you follow my comparison, I honestly believe that removing certain scrambles is basically the same as banning pocket Aces, and similarly as pointless.

 

[Bob]

Actually, cubing is quite a bit like no-limit hold'em...

If you follow my comparison, I honestly believe that removing certain scrambles is basically the same as banning pocket Aces, and similarly as pointless.

Nowhere near.

P(Pocket aces) ~ 1/221
P(2 move scramble) ~ 1/68,000

This makes pocket aces more than 300 times more likely than a 2 move scramble.

Pocket aces still only gives you about an 80% chance to win a hand. What do you think the odds are that the WR can be broken on a 2-move scramble? What about the odds that it will (or even can) ever be broken again afterwards?

 

[Escher]

Nowhere near.

P(Pocket aces) ~ 1/221
P(2 move scramble) ~ 1/68,000

This makes pocket aces more than 300 times more likely than a 2 move scramble.

Pocket aces still only gives you about an 80% chance to win a hand. What do you think the odds are that the WR can be broken on a 2-move scramble? What about the odds that it will (or even can) ever be broken again afterwards?

Yes, I'm well aware of the odds of receiving pocket aces and their chances to win, and obviously, it would be stupid to compare numbers directly (that said, there are also situations in Pot-Limit Omaha where almost every card in the deck is an out to improve and presumably win the hand).

I meant more to say that both disciplines have starting positions with differing relative strengths, that are out of a players control, and thus a sample size of 1 in each is equally as meaningless... So we shouldn't worry about it.

 

[Bob]

Yes, I'm well aware of the odds of receiving pocket aces and their chances to win, and obviously, it would be stupid to compare numbers directly (that said, there are also situations in Pot-Limit Omaha where almost every card in the deck is an out to improve and presumably win the hand).

I meant more to say that both disciplines have starting positions with differing relative strengths, that are out of a players control, and thus a sample size of 1 in each is equally as meaningless... So we shouldn't worry about it.

There's a major difference that you're overlooking. It would be fairer to compare a 2-move scramble to a straight flush, which I believe has a probability of like 1/72,000 or so.

In cubing, we have competitions, just like poker has tournaments. Let's say that WSOP is equivalent to our WC. Within a tournament, somebody might get a straight flush and it might even happen in the same hand to a couple people at a table. In a cube competition, the 2-move scramble will be given to an entire group of competitors, ALL of whom will get the ridiculously easy scramble and get a good time. Will it affect the standings in the competition? Probably not. Since everybody got the scramble in that group, they'll all get a good time on it and that will be the dropped (best) time for each of them. In the competition, it doesn't really make a difference. However, we keep track of something that they don't in poker: world records. In a poker tournament, the benefits of that royal flush don't really extend much beyond that particular game. It might not even extend much past that hand. However, in cubing, that 2 move scramble creates a record that is virtually unbeatable. Regardless of the skill level of competitors, many people will never have the opportunity to break that record. Even the best 2x2 solver in the world will probably never even come close to breaking it. By filtering these scrambles out, we reduce the impact of luck and increase the amount of skill required to break the record.

Let's assume that every competition held from now on has 3 rounds of 2x2 with 5 solves each. Last year, there were 265 competitions with 2x2. Assuming the same number of competitions each year, that's 3,975 scrambles per year. That gives about a 5% chance per year that a 2 move scramble will even occur. Assuming about 4,000 scrambles per year, that also means it would take 12 years for a 50% chance that a scramble like that will be given in a competition. Yes, the likeliness that it will even occur at all is really low, but if it does occur, it makes the single record that much more meaningless. At least a 4 move solve requires some skill. I don't think that we should make things worse by allowing these scrambles.

What about a 1 move scramble? If one of these ever occurred, it would be such a joke.

I feel this is one of those topics that people already have their opinions about. There's not much that somebody can say to me that would sway my opinion on it and I don't expect that there is much to be said that will change somebody's mind that already thinks we should not filter scrambles. Therefore, all discussion in this thread is about as useful as tits on a bull.

 

[Zarxrax]

The difference in triviality between 3 and 4 move scrambles is negligible.
The difference in triviality between 4 and 5 move scrambles is negligible.
The difference in triviality between 5 and 6 move scrambles is negligible.
The difference in triviality between 6 and 7 move scrambles is negligible.
The difference in triviality between 7 and 8 move scrambles is negligible.

I would actually disagree with this statement that the 1 move difference is negligible. I think it actually gets more and more difficult each time (up to about 5 moves). Maybe not exponentially more difficult, but its definitely not linear either.

Let's say there is a 1 move scramble. Any random drunk guy could solve it.
Now take a 2 move scramble. Any experienced cuber could do this no problem, but a person who has never touched a cube before might have difficulty getting it during the 15 second inspection.
With a 3 move scramble, most cubers will still be able to get this, but we are getting to the realm where a normal person wont be able to.
For 4 move scrambles, I think we are reaching the point where a lot of cubers simply can not recognize the situation in the time frame required. I mean seriously, do a random 4 move scramble that's not something like RUR'U'. Does it really look easy enough that you could recognize that its 4 moves and work out the solution in the inspection time? I think only really experienced cubers could do this for most scrambles.
Now moving on to a 5 move scramble, at this point, just about anyone is just going to go with their normal solving method, because its scrambled enough that someone just isn't going to be able to work out an optimal solution.

So, yes, there is a SIGNIFICANT difference between a 3 move scramble and a 4 move scramble, and then there is another significant difference between a 4 and 5 move scramble.

 

[Escher]

There's a major difference that you're overlooking...

Not going to quote/dissect/disagree with all of this individually, but...

- I don't see why 12 years of tests raises the probability of hitting a 1/68k to 50%? I'm no statistician but it sounds a lot like the Gambler's Fallacy.

- We keep track of the best 2x2 single time, but I don't see why keeping track of it suddenly makes it more important than the outcome of one hand of poker? I don't see any benefits arise for the competitor who gets the 2x2 single WR apart from that the WCA lists it on the website along with other records for different puzzles (of which the benefit is that everybody thinks they are awesome).

- The best 2x2er in the world is probably the one who is most aware that 2x2 is a game with a large amount of luck involved, especially when it comes to single times.

- Filtering scrambles would just create a different set of states that 2x2ers would consider lucky (a variety of 4-7 move states are almost as good as 3 moves, and I'm sure plenty are more common), and after a certain amount of time, we would inevitably reach the 'best state possible without being 3 moves or less' and the WR single would be equally as unbeatable as before you filtered scrambles - all you do is delay the 'problem'.

I actually don't think that 2x2 single should be recorded as any kind of record (even PR, let alone NR/CR), and that 2x2 format should at least be changed to avg7

The only reason discussion ITT is useful is to show the 'higher-ups' that the community is pretty divided on the matter, with pretty reasonable arguments from both sides, and therefore it probably isn't a great idea to implement it any time soon.

Not that the status quo helps my agenda or anything.

 

[qqwref]

- We keep track of the best 2x2 single time, but I don't see why keeping track of it suddenly makes it more important than the outcome of one hand of poker?

The difference is that the one hand of poker becomes irrelevant as soon as the next hand is dealt (the only lasting effects are a bit of redistributed money), but the 0.4x time some random noob inevitably gets will stay near the top of the rankings for many years.

- The best 2x2er in the world is probably the one who is most aware that 2x2 is a game with a large amount of luck involved, especially when it comes to single times.

Defeatist, no? You're saying that, if some luck is inevitable, we might as well give up trying to minimize the luck.

- Filtering scrambles would just create a different set of states that 2x2ers would consider lucky

So? Obviously there will always be some Best State. The problem with 2-move scrambles is that that state is significantly easier, to every solver, than the relatively much more common 4-move scrambles. By contrast the best 4-move scramble isn't really all that much better than a bad 4-move scramble (or an awesome 7-move scramble, see Erik). A bad time by an unskilled solver on the best 4-move scramble is still beatable with a more common type of scramble. But a bad time by an unskilled solver on a 2-move scramble may not be.

 

[Escher]

The difference is that the one hand of poker becomes irrelevant as soon as the next hand is dealt (the only lasting effects are a bit of redistributed money), but the 0.4x time some random noob inevitably gets will stay near the top of the rankings for many years.

So what? Keeping track of it still doesn't give it any meaning, I think the community as a whole is pretty decided that 2x2 single rests almost entirely on luck. All it does is give the random noob some bragging rights to his friends.

Defeatist, no? You're saying that, if some luck is inevitable, we might as well give up trying to minimize the luck.

More that 2x2ers should work on maximising the luck they are given (practising optimally solving 4-6 move scrambles, learning multiple methods), rather than the community minimising the variance between puzzle states.

So? Obviously there will always be some Best State. The problem with 2-move scrambles is that that state is significantly easier, to every solver, than the relatively much more common 4-move scrambles. By contrast the best 4-move scramble isn't really all that much better than a bad 4-move scramble (or an awesome 7-move scramble, see Erik). A bad time by an unskilled solver on the best 4-move scramble is still beatable with a more common type of scramble. But a bad time by an unskilled solver on a 2-move scramble may not be.

Even in the case that an 'unskilled' solver receives a 2 move scramble, in order to break even the current WR one would have to be pretty experienced at using a stackmat. Everybody in that round would receive the same scramble, so the best person in that round would presumably get the lowest time, and thus could be said to be somewhat 'skilled', even if they aren't the best in the world - not that 2x2 single often reflects skill in any case.

All of this arguing seems to be based on the idea that the 2x2 single is valuable in some way, which I thought we all grew out of years ago. If the community has the problems it seems to with 2x2 single, then the solution that resolves it the best would probably be to simply stop awarding a rank to single 2x2 results, rather than use an 'official subset of states', with its own particular Best State.

 

[Stefan]

- I don't see why 12 years of tests raises the probability of hitting a 1/68k to 50%? I'm no statistician but it sounds a lot like the Gambler's Fallacy.

You're just misinterpreting what he said. If with one attempt (year) you have a 5% chance, then with 12 attempts (years) you have about 50%:

1 - (1-0.05)^12 = 0.459639912
or
log(0.5) / log(1-0.05) = 13.5
(obviously he didn't use exactly 5% in his calculation)

I don't see any benefits arise for the competitor who gets the 2x2 single WR apart from that the WCA lists it on the website along with other records for different puzzles (of which the benefit is that everybody thinks they are awesome).

1. That benefit is not good.
2. It takes the benefit from a more skilled/deserving person.
3. Cubing looks less respectable if we have to admit that the WR was set on a "scramble" that everybody can solve easily.

Filtering scrambles would just create a different set of states that 2x2ers would consider lucky (a variety of 4-7 move states are almost as good as 3 moves, and I'm sure plenty are more common)

As said before, two moves is not just lucky, it's trivial. To the point that even non-cubers can easily see the solution. Not the case with 4-movers or even 7-movers.

Not that the status quo helps my agenda or anything.

Yup, the status quo *opposes* your agenda, as the status quo is that all scrambles less than four moves *have* been filtered out.

 

[Erik]

3. Cubing looks less respectable if we have to admit that the WR was set on a "scramble" that everybody can solve easily.

Which isn't the case now? With all due respect for the top 3 in 2x2 single, but they aren't really world-top. All 3 of them have a best average of 3,50+ (3,64, 3,97 and 3,82). They all did it on the same scramble.

 

[Stefan]

Which isn't the case now?

No, I don't think that's the case now. Give the current WR scramble (apparently U'F'R'U) or a harder 4-mover (including half turns) to non-cubers and I'm convinced they'll have trouble with it that they won't have with a 2-mover. Don't you think so?

With all due respect for the top 3 in 2x2 single, but they aren't really world-top. All 3 of them have a best average of 3,50+ (3,64, 3,97 and 3,82). They all did it on the same scramble.

Indeed. But I don't see what that has to do with my point? They're not representative of everybody, not even representative of cubers (since their averages are already pretty good) but just representative of already pretty good cubers.