Vincents
Veteran
Stefan, basically what happened was that there were identical scrambles for different 3x3 rounds and events, due to unknown reason (glitch or human error - not sure).
Throwing out scrambles in competitions?
- Thread starter StachuK1992
- Start date
Stefan, basically what happened was that there were identical scrambles for different 3x3 rounds and events, due to unknown reason (glitch or human error - not sure).
Kirjava
Colourful
One of the reasons doesn't count because the same thing will be achieved by not filtering.
The other is simply not a good reason.
Stefan
Legend
But which regulation was violated? Closest I found is "1h2) Groups must have different scramble sequences", but if you don't divide into groups...
Kirjava
Colourful
2 move scrambles.
Vincents
Veteran
4b) Puzzles must be scrambled using a computer-generated random scramble sequence or scrambled position, that must be kept secret for all but the scramblers.
They weren't a secret, because they'd already been used for a prior round. If this logic doesn't hold, 1h2 did, as there were 2 groups for OH Round 1 and 3x3 Round 1.
Muesli
Premium Member
I think too much emphasis is placed on single solves. The only people who are really naive enough to get away with caring about them are non-cubers. There's a reason we take averages.
Who's to say what's easy or not anyway? Perhaps the delegate doesn't spot the easy Xcross or 4 move roux block?
The only way I can see to remove all the outliers is to get everyone at the comp to try the scrambles first, throw some statistical maths at the results and remove the scrambles that consistently produce faster or slower solves. Obviously that is ridiculous.
I completely disagree with scramble filtering. Its unnecessary and, forward to Rowan's point about the poker comparison, why can't luck be a factor in cubing?
Who's to say what's easy or not anyway? Perhaps the delegate doesn't spot the easy Xcross or 4 move roux block?
The only way I can see to remove all the outliers is to get everyone at the comp to try the scrambles first, throw some statistical maths at the results and remove the scrambles that consistently produce faster or slower solves. Obviously that is ridiculous.
I completely disagree with scramble filtering. Its unnecessary and, forward to Rowan's point about the poker comparison, why can't luck be a factor in cubing?
Stefan
Legend
Again: only if we actually don't filter. You're talking about the theoretic ideal case, in reality it's more like "the same thing might be achieved by forbidding filtering".
Consistency is more likely be ensured by one computer program than by dozens of humans, so it's not quite the same.
Yeah, but why would you care?
Unless people knew they'd be given these scrambles again, I'd still call that secret. But ok, that's just opinion.
Um... that's why we're talking about *not* the delegate making the decision (the computer program doing it instead) and *not* about Xcross/RouxBlock/otherMethodSpecific filtering (only about short scrambles)?
Last edited by a moderator:
Kirjava
Colourful
We're talking about changing this from the judge's discretion to something else. There are two options for this - filtering and non-filtering. Both have the advantages of just being able to be produced by scramble programs with no intervention required by the delegate. This is why I don't think it's worth mentioning.
I would like to hear a good reason for filtering, I really would. Then I would be more open to considering it.
Because I would like people to be able to compete and have the possibility of getting those scrambles. If the scrambles are removed, this is not possible.
Last edited by a moderator:
Stefan
Legend
Do you mean that as "doesn't have to intervene" or as "must not intervene"? It sounds like the former but should be the latter, and the latter I do find worth mentioning.
But what would they lose if they couldn't get those scrambles? You're not really answering the question, you're just making us go in circles. Let me rephrase, maybe this works better: What harm would be done?
I would like to hear a good reason for *not* filtering, I really would.
cuBerBruce
Veteran
If an average-of-5 round has a 2-move scramble, practically all competitors would get their fastest time on that scramble. Therefore the time on that scramble would be moot except as far as records are concerned. If a round were best of two or best of three, then that really easy scramble would be pretty much the only factor affecting who advances to the next round. I think it would be really silly to base who advances to the next round on a 2-move scramble. I've made this point on the WCA forum before, but the WCA ignored it.
Bob
Premium Member
I believe there should be a minimum scramble length in which scrambles shorter than that are automatically NOT generated by the software. Ideally, 2x2 scrambles would be a minimum of AT LEAST 3 or 4 moves, for example.
asportking
Veteran
This problem really only applies to the 2x2 and maybe the 3x3 (possibly pyraminx as well, I don't know much about that). The easiest way I can see is to just accept anything that isn't solved or 1 move away. If we started throwing away lucky scrambles, would the 2x2 WR be invalid? Because I don't see how anyone would be able to beat .69 seconds with a non-lucky scramble.
Stefan
Legend
Oh really?
Depends on what you throw away. If we only throw away scrambles luckier than the WR, then certainly not.
jskyler91
Veteran
This is an important point. Nothing that has happend before should be unachievable now. Also, we need to consider how harshly we want to limit speeds of our competitors. By this I mean: do we think that a 2 second solve for the 3x3 should be achievable or considered legitimate? Personally, I say not if we don't filter and yes if we do: Without some form of filters one could have a 5 or 6 moves solve which would be almost impossible to beat if say FAZ performed it at full speed. If, however, we filter the solves such that no 13 or less move solve is legitimate then I feel any speed achieved should be accepted because the scrambled state is so great that it would require some skill to solve it, not just luck. Just some food for thought...
Last edited:
Bob
Premium Member
There is of course the alternative I have suggested in the past: ignoring the 2x2 single (ie - no WR for 2x2 single). There is too much variation in scrambling to make the 2x2 WR meaningful.
qqwref
Legend
- Joined
- Dec 18, 2007
- Messages
- 7,834
- Location
- a <script> tag near you
- WCA ID
- 2006GOTT01
- YouTube
- Visit Channel
We'd also have to ignore Pyraminx single, Clock single, Square-1 single, Skewb single, and potentially 3x3x3 single if someone gets lucky enough. I don't like the idea of just throwing out records for part of an event just because we don't have an appropriate way to deal with exceedingly lucky cases.
I stand by my previous proposal - when doable, we should ignore positions that are "too close" to solved. We can choose something like the closest 1/million or 1/thousand scrambles in order to provide as little effect to normal competition as possible. IMO there is definitely such a thing as a scramble that is so easy that people will get times they could never achieve due simply to getting an easy scramble (e.g. sune on 2x2) or having the luck to choose a very fast speed-solution.
I stand by my previous proposal - when doable, we should ignore positions that are "too close" to solved. We can choose something like the closest 1/million or 1/thousand scrambles in order to provide as little effect to normal competition as possible. IMO there is definitely such a thing as a scramble that is so easy that people will get times they could never achieve due simply to getting an easy scramble (e.g. sune on 2x2) or having the luck to choose a very fast speed-solution.
masterofthebass
Premium Member
to be honest... I don't think a scramble like that should be thrown out. That solve was 22 moves in the same metric that the 25 move scrambler uses (I think. Each face+slice = 1 move) It just happened to be easy if you did it a certain way. If I had gotten that in competition, I would have gotten an Aperm on top and solved on D, which isn't the amazing solve that Arnaud and the others happened to get. I do agree that lots of square-1 scrambles shouldn't be allowed in competition, but that particular one doesn't fit.
Kirjava
Colourful
Must not.
The difficulty level is artificially increased.
You didn't reply to these - maybe you didn't see them?
Godmil
Premium Member
Wasn't it the case in the old days that people had to say what colour they started with, so the scrambler would make sure they didn't have an easy start... which lead to competitors lying about their starting colour so that they weren't being deliberately given difficult scrambles. I'm sure everyone would agree that the current 3x3 rules are much better for removing this. Is it valid to apply this reasoning to the 2x2? I think so, but it may not be a direct comparison.
Cubenovice
Forever Slow
On average: yes.
But no "more difficult" scrambles will be created by filtering.
I think also the effect on average would be minimal. I'm sure Bruce could calculate by how much the average optimal solve/scramble length would increase by eliminating scrambles up to a certain treshold. (assume 4 on 2x2 and 13 on 3x3).
And what about a more emotional argument:
What "value" would you attach to a 0.78s 3x3x3 solve where only 4 moves are needed?
And what would the general non-cuber public think of a 4 move WR solve?
As if a darts player would just walk upto the board and stick his darts into the triple 20...
And even with a threshold of 13 moves people can get lucky: PLL...
