2009: Megaminx, Square-1 and clock: change the format?

Pedro_S (2008-09-11 15:07:48 +0000)
We could make these events average of 5 instead of mean of 3, no? megaminx is not that hard to scramble anymore (thanks, Stefan :)) and the solving times are not much slower than 5x5 square-1 is faster to solve, and scrambling isn't really hard if you "practice" for a bit I don't know about clock, 'cause I don't have one, but the solving times are quite fast, and we see lots of DNF averages... of course, we can do combined rounds to save time, but I think having them as average of 5 would be good
MichaelPalone (2008-09-11 16:14:46 +0000)
I can't really add anything, I agree with you on everything already said - make them all average of 5. But what of all the current records?
Pedro_S (2008-09-11 21:10:10 +0000)
[quote="MichaelPalone":1hln3rjn]I can't really add anything, I agree with you on everything already said - make them all average of 5. But what of all the current records?[/quote:1hln3rjn] keep them until they get broken in the new regs this was done in the past with OH the big cubes
qqwref (2008-09-12 00:20:05 +0000)
Yeah, I think every event should be average of 5 unless it is completely impractical to do so (BLD events, or anything that's way longer than 5x5, such as 7x7). Mean of 3 is outdated and also includes too much luck (DNF = last place, and one good solve affects a mean a lot). The DNF problem is especially true with Clock because you can only see one side of the puzzle at a time and thus a small execution mistake can mean a DNF average. Since you will never get more than one Clock round per competition, at US Nationals I made sure to check my Clock each time after I solved it, and although I did waste about a second I did make sure that I would get a successful mean. I was talking about this with Arnaud at the US Nationals competition and he agreed that it made no sense to do mean of 3 on such a short and risky event (it's a lot easier to get a DNF on Clock than on any other actual puzzle, since you can only see one side at a time so making a small mistake can mess up the back without your knowledge).
Lucas (2008-09-12 06:40:13 +0000)
Indeed, average of 5 makes sense for these events, but we have to be realistic. Megaminx is a relatively consistent challenge, and mean of 3 is relatively descriptive. There are also generally fewer dramatic things like pops and accidental DNFs. Scrambling the extra minxes ist just a lot more work, and is not much more effective. Clock is very short, not terribly hard to scrambles, and accidental DNFs are very influential. I think that we should have avg5, and I think we can influence this change in the next regulations. Square-1 is debatable. Thing is, the more experienced scramblers at competitions are often the solvers, so that you have less experienced scramblers who make more mistakes, costing either time or fairness. Also, there are some funny reasons for time variations; is avg5 that much better?). I think Square-1 is an issue for later. We need better Square-1 scrambling first, I'd say.
BryanLogan (2008-09-12 11:14:27 +0000)
I'll re-iterate what I've said in various other places. If we have some sort of "Super-penalty" for the mean events, then we decrease the chance of a DNF for the average. The simplest way to do this is if the puzzle is a DNF and they have no received a super-penalty already for the round, is to record the time, reset the timer, and time the competitor to finish the puzzle. Add the original time, plus the new time, plus 10 seconds (or whatever number). This way, we don't have to determine if they're two turns, three turns, a turn and a peg push, etc... If you have a major failure on Square-1, Megaminx, or Clock, then your average won't be severely penalized to infinity. You can't skip Square-1 parity and just take a penalty. You have to be able to complete the puzzle.
Ron (2008-09-12 21:32:21 +0000)
I think we have to become stricter in the upcoming years. Instead of more kinds of penalties we should have fewer. Just DNF or a solve, nothing else. I do not like this poll. I think it is a similar poll to: "how many dollars do you want to receive for free?". A better poll would be: "how many dollars do you want to receive if you have to work one hour?". Who wants to do 'average of 5' for Megaminx, under condition that the 5x5x5 event will not be held in all competitions you are going to compete at? This may be an extreme example, but what I am trying to say is that we cannot just add all kinds of events and extra solves, without having impact on other things. Thanks, Ron
BryanLogan (2008-09-13 15:02:47 +0000)
[quote="Ron":32a3q9nf]I think we have to become stricter in the upcoming years. Instead of more kinds of penalties we should have fewer. Just DNF or a solve, nothing else.[/quote:32a3q9nf] OK, but the problem you have is that Megaminx, Square-1, and Clock still have the disadvantage that any error will ruin your chances of winning. If you drop the +2 penalty for everything else, then if someone has a misalignment on their 3x3x3 solves, then they're still in a position to win. Unless you said that for the "average of 5", if you get a DNF, even if it's the lowest one, you get a DNF for the average. That way, you're demanding perfection for all puzzles. Of course, no one is going to like that suggestion. You could have the major penalty as the only penalty, for all puzzles and get rid of the +2. So if someone got what would be a +2, it would be assessed as a major penalty, and they could only do it once per round. So that we are demanding better performance from all cubers, since someone who would get two +2's in a single round would then get a DNF for the average. And the problem with a mistake on the Mean of 3 puzzles would also be solved.
Pedro_S (2008-09-13 17:48:54 +0000)
[quote="Ron":pgv9mwxz] I do not like this poll. I think it is a similar poll to: "how many dollars do you want to receive for free?". A better poll would be: "how many dollars do you want to receive if you have to work one hour?". [/quote:pgv9mwxz] sorry, Ron, that wasn't my intention but I think the poll is valid. Some people may agree to change just the clock format, some may not agree at all. And, as I said, we can still make combined rounds for those events to save time. The difference would be 2/5 solves instead of 1/3.
Shelley (2008-11-12 17:01:57 +0000)
I don't know enough about Clock or Megaminx to comment on those, but I've noticed that Square-1 has become more popular lately (more people registered for Square-1 than 3x3 BLD at both the upcoming Caltech and Berkeley tournaments). I get the impression that Square-1 was Mean of 3 because it was a "fringe" event that not many people did. As it gets more popular I don't see why we shouldn't make it Average of 5, at least eventually. More people are learning how to scramble it and it doesn't take as long as Megaminx.
Ernesto (2008-11-13 13:41:39 +0000)
Hi! Rubik´s Clock, If you have a mistake you can not win, the other puzzles you can win. Example: Competitor X: (times) 10 minutos/20 minutos/30 minutes = 20 minutes Competitor Y: (times) 10 seconds/DNF/12 seconds = DNF Competitor X wins And the +2 should also be an option for small mistakes such as 3 * 3 * 3, 2 * 2 * 2 ... etc. [u:2art9ihj]we see lots of DNF averages....[/u:2art9ihj] IT´S TRUE [u:2art9ihj]it's a lot easier to get a DNF on Clock than on any other actual puzzle, since you can only see one side at a time so making a small mistake can mess up the back without your knowledge[/u:2art9ihj] IT´S TRUE Clock and square-1, average 5. Megaminx, combined final could be a solution Thank you very much!!!
anders (2008-11-13 18:18:26 +0000)
One reason, as I understand it, to have "Mean of Three" is to be more time-efficient. This is not the same issue as before since we already have too many events to house in one competition day. Even a two-days competition has trouble to fit in all events, in particular if no events are in parallel. Thus, I support the idea to promote these events to "Average of Five" (at least Clock and Square-1...) as we already have done for 3x OH, 4x and 5x. /Anders
Edouard Chambon (2008-11-15 18:13:01 +0000)
I'm really FOR a changing in clock. For 2 reasons, it's short and a DNF can be easily possible. I'm totally AGAINST in megaminx because it's already too long, and it would not change anything. And for sq1, I don't care.
Tyson (2008-11-18 18:40:10 +0000)
One of the main things here to consider is the impact of average of 5 vs. mean of 3. For the Megaminx, I do not think the initial state of the puzzle really impacts the solving much. There are 12 sides to a megaminx, and it really solves itself more like a 5x5x5 than a 3x3x3 in terms of variation. (Please correct me if I'm wrong.) However, Clock and Square-1 are far more variable in terms of their solving. A Square-1 can have parity cases, and Clock is prone to DNF and other things. I think average of 5 for these two events would really help illustrate a more meaningful time for the competitor. Scrambling in Clock and Square-1 is not too bad, and as long as competitors are willing to help, I do not believe it is a drain on resources. But Ron has a point about the poll, but I don't think he meant any offense. If we had a poll saying, "Should we have more competitions," wouldn't everyone say yes? When you can ask for something without doing any work for it, it's not quite the same thing. People who vote on this might not be considering the impact on competition resources. But that being said, I don't think Square-1 and Clock are a considerable drain on resources. For Megaminx, I don't see that there is significant enough variation in the solve.
cada (2008-11-19 00:16:49 +0000)
[quote="Tyson":3q1cd20n]But that being said, I don't think Square-1 and Clock are a considerable drain on resources. For Megaminx, I don't see that there is significant enough variation in the solve.[/quote:3q1cd20n] For megaminx I think the concern would be pops. What about defining a +2 state for clock? One already exists for square1 and megaminx, right? -Chris Krueger
Tyson (2008-11-19 17:58:35 +0000)
For Clock, how do DNFs mostly occur? By being one-click away?
TMOY (2008-11-20 07:48:27 +0000)
[quote="cada":h2yra2pt] For megaminx I think the concern would be pops. [/quote:h2yra2pt] A normal pop (say not more than three pieces) can be fixed in about 10-20 seconds, which at megaminx does not destroy an average as much as it can do at 3^3. I have popped twice in competitions so far (once in Brussels and once in Geneva) and have still gotten decent times for both solves.
Laetitia (2008-11-20 13:06:07 +0000)
Pops can destroy an average at megaminx. Watch my times, or worst, Yann Lenhardt's times at Lyon Winter Open. But on the other hand, it's the competitor's fault if (s)he pops. I'd like to see megaminx avg of 5, it does not take more time than 5x5x5 (neither to scamble nor to solve).
Tyson (2008-11-20 19:59:42 +0000)
Is it necessary though? I feel that times on a Megaminx have less variance than times on a 5x5x5. But I am not a Megaminx solvers so I will have to defer to other's opinions on my previous statement.
StefanPochmann (2008-11-20 20:08:41 +0000)
Mean-of-3 for megaminx is fine for me.
Tyson (2008-11-20 23:23:07 +0000)
Yes, and a nice delicious moist cake would be fine for me :-P Can we get some numbers to be a bit more objective maybe? Like, standard deviation compared as a percentage to the mean time?
Pedro_S (2008-11-21 01:16:11 +0000)
[b:3rldvizv]5x5x5[/b:3rldvizv] Dan Cohen, 1:29.40, 3.70, 4,1387% Hsuan Chang, 1:30.72, 3.45, 3,8029% Michal Halczuk, 1:30.79, 2,13, 2,3461% Yu Nakajima, 1:32.47, 3.07, 3,3199% Lee Seung-Woo, 1:32.85, 0.34, 0,3662% Average percentage: 2,79476% [b:3rldvizv]Megaminx[/b:3rldvizv]** Erik Akkersdijk, 1:07.70, 5,66, 8,3604% Stefan Pochmann, 1:23.79, 1,62, 1,9334% Bence Barat, 1:25.73, 3,89, 4,5375% Adam Polkowski, 1:37.77, 1,42, 1,4524% Laetitia Lemoine, 1:37.80, 1,74, 1,7791% Average percentage: 3,61256% *I didn't use Erik's average because it had a DNF, so CCT couldn't give me the SD **I used the SD from CCT, which I think counts only the middle 3 times, hence his SD being so low, even with a 2:10 time those are the top-5 (2-6 for 5x5x5) for both...I think megaminx has a bigger deviation, indeed :)
Tyson (2008-11-21 13:46:45 +0000)
Thanks Pedro, Can you do the same for Clock and Square-1?
Pedro_S (2008-11-21 14:10:22 +0000)
[quote="Tyson":k72sze9z]Thanks Pedro, Can you do the same for Clock and Square-1?[/quote:k72sze9z] yeah (comparing with some other event? or just the 2?)
Tyson (2008-11-21 14:21:08 +0000)
Well, once you have the percentages, we can take a look at it against Megaminx and 5x5x5, which you have already done. And I bet someone smarter than me, and possibly blond, and a bit taller, could probably write a script connecting to the database to do this calculation and calculate for the top 20 results, or perhaps across competitions, and not just personal bests. Oh... I think there's a flaw in this calculation right now. A bias, which we must take into account. If you count a person's BEST mean of 3, naturally it will have a standard deviation. If you have a typical mean of 3: average solve, average solve, average solve, the standard deviation will be less. If you have a good mean of 3, perhaps your personal best, it's probably more something along the lines of: average solve, average solve, HOLY CRAP GOOD SOLVE! which would hurt the standard deviation. So whereas this calculation is good for reference, we would expect the mean of 3 percentage deviation to be higher for megaminx than 5x5x5, even though this might not be necessarily so if both were done by average of 5. (I hope I made sense.)
Pedro_S (2008-11-21 14:23:56 +0000)
sorry, I had the megaminx values wrong I've corrected them now. The average deviation is 3.61256%, for the top-5 doing square-1 now... EDIT: [b:29q1k7xp]Square-1[/b:29q1k7xp] Kang Ji-Jon, 17.77, 2,49, 14,0124% Lee Seung-Woo, 18.50, 6,52, 35,2432% Grzegorz Prusak, 19.57, 5,36, 27,3889% Michal Robaczyk, 20.28, 2,78, 13,7081% Martin Zahradnik, 22.99, 1.01, 4,3932% Average percentage: 18.94916% as I expected, the % is much higher, since the times are much lower, compared to megaminx and 5x5x5 I'll do clock in a minute, after lunch EDIT2: [b:29q1k7xp]Clock[/b:29q1k7xp] Matyas Kuti, 8.60, 0.52, 6,00575% Oliver Perge, 9.50, 1.24, 13,07991% Stefan Pochmann, 9.74, 0.41, 4,23690% Dan Cohen, 9.84, 0.65, 6,64725% Ernesto Fernandez Regueira, 9.94, 0.80, 8,01174% Average percentage: 7.59631% Note: I used my brain a bit and made an excel sheet, instead of doing the maths by hand. I can do more persons now, if you'd like
BryanLogan (2008-11-21 15:06:47 +0000)
I think one of the reasons people want an average of 5 is to solve the DNF issue. Even is the standard deviation is low and the mean of 3 is a good representation of your skill, a single DNF will basically disqualify you. Can we get the total number of DNF's in these events?
Pedro_S (2008-11-21 15:43:38 +0000)
I did more calculations, for the top-10 on each Average % Square-1: 15.50846 Clock: 7.18452 Megaminx: 4.97658 5x5x5: 2.57012 3x3x3: 3.74853 Square-1 is by far the "worst" one. Clock has a high relative SD, but the absolute SD is pretty low.
Pedro_S (2008-11-21 15:56:28 +0000)
[quote="BryanLogan":10xkext1]I think one of the reasons people want an average of 5 is to solve the DNF issue. Even is the standard deviation is low and the mean of 3 is a good representation of your skill, a single DNF will basically disqualify you. Can we get the total number of DNF's in these events?[/quote:10xkext1] I can't get you the totals, but for the [u:10xkext1]top-10[/u:10xkext1] people: Clock: 17% averages are DNFs Square-1: 3% DNFs (Lars has 28 averages, 5 being WRs) Megaminx: no DNFs but these are for the top-10 people. I have no idea of the whole situation
StefanPochmann (2008-11-21 19:12:19 +0000)
[quote="Pedro_S":1s76qxaq]Clock: 17% averages are DNFs but these are for the top-10 people[/quote:1s76qxaq] Well, I'm still among those top-10 (I'm #3 in both single and average) and I had a single DNF in 49 attempts so far. And that was recently when I was out of shape, didn't care, forgot my own clock and borrowed someone else's that I wasn't used to. Also, my overall average is 11.4 seconds and would be 10.8 seconds without my terrible 37 seconds solve last year (when I was also out of shape and made several horrible mistakes). So I'd say flawless clock-solving at high speed is quite possible and people have to blame themselves for their DNFs because they're careless. It's not a flaw of the puzzle. That said, I don't like a single mistake practically disqualifying someone. As I can't imagine good rules for determining a penalty situation (can anyone?), I guess I'd like it to be average-of-5.
Pedro_S (2008-11-21 19:44:45 +0000)
[quote="StefanPochmann":2qjsemdr][quote="Pedro_S":2qjsemdr]Clock: 17% averages are DNFs but these are for the top-10 people[/quote:2qjsemdr] Well, I'm still among those top-10 (I'm #3 in both single and average) and I had a single DNF in 49 attempts so far. And that was recently when I was out of shape, didn't care, forgot my own clock and borrowed someone else's that I wasn't used to. Also, my overall average is 11.4 seconds and would be 10.8 seconds without my terrible 37 seconds solve last year (when I was also out of shape and made several horrible mistakes). So I'd say flawless clock-solving at high speed is quite possible and people have to blame themselves for their DNFs because they're careless. It's not a flaw of the puzzle. That said, I don't like a single mistake practically disqualifying someone. As I can't imagine good rules for determining a penalty situation (can anyone?), I guess I'd like it to be average-of-5.[/quote:2qjsemdr] I counted just (DNF averages)/(total averages) I agree that a single mistake disqualifying is not good
BryanLogan (2008-11-21 20:28:39 +0000)
[quote="StefanPochmann":3g7wk785]As I can't imagine good rules for determining a penalty situation[/quote:3g7wk785] I proposed an idea a while back that involves basically taking anything that isn't solved as a penalty, and the time would be some time plus the time it takes the competitor to get to a solved state. And you could only do this once a round. EDIT: Doh, it was in this thread before the thread was renamed and bump.
StefanPochmann (2008-11-22 15:03:25 +0000)
[quote="BryanLogan":1ieblu2w]taking anything that isn't solved as a penalty, and the time would be some time plus the time it takes the competitor to get to a solved state.[/quote:1ieblu2w] And how do you count the time it takes to determine that the back side wasn't solved? And how do you account for the competitor seeing what is wrong and thinking how to fix it while the timer isn't running? Imagine I could quickly compute "God's algorithm" in my head, but only for one side. So I do that during inspection for the first side. Then I start the timer, solve the first side, turn the puzzle over, stop the timer saying "oops I forgot the second side let me continue", but have a good look at the second side and plan my solution to it before restarting the timer. Cool.
BryanLogan (2008-11-22 20:01:51 +0000)
[quote="StefanPochmann":j0kedgj4][quote="BryanLogan":j0kedgj4]taking anything that isn't solved as a penalty, and the time would be some time plus the time it takes the competitor to get to a solved state.[/quote:j0kedgj4] And how do you count the time it takes to determine that the back side wasn't solved? And how do you account for the competitor seeing what is wrong and thinking how to fix it while the timer isn't running? Imagine I could quickly compute "God's algorithm" in my head, but only for one side. So I do that during inspection for the first side. Then I start the timer, solve the first side, turn the puzzle over, stop the timer saying "oops I forgot the second side let me continue", but have a good look at the second side and plan my solution to it before restarting the timer. Cool.[/quote:j0kedgj4] Well, I guess is you did have this Rainman-like ability, you might be able to get a WR for a single solve, assuming that the additional time isn't going to harm you (instead of just adding the two times, you also add something else, like 2 or 5 seconds, in there also). So unless you're able to shave off 2 or 5 seconds of your time by having it split, it's not going to help. Also, you would be able to only do this one during a round, so if your regular solve is so much more slower than this solve, your average isn't going to work.
qqwref (2008-11-23 08:49:52 +0000)
I don't like the 'stop and continue' idea, it lends itself much too easily to doing it on purpose (i.e. cheating). I'd prefer to just say, if you stop the timer yourself and the puzzle is together but not solved, you get a DNF. It's your fault. Square-1 and Clock are very short events though, much shorter than 4x4 and 5x5 (which have been average of 5 for a very long time), so there's no reason not to make them averages. And as has been said before Square-1 has a ridiculously high SD compared to its average - easy cases (adjacent-adjacent on EP, say) are MUCH faster than hard cases (W/Z or something awful like that). Or someone can spend 10 seconds on a cubeshape normally, but if they have a two-turn one it might be 3 seconds instead. Just because of the inconsistency of the event, it's a bad idea to do a mean of 3, because one really good solve will make your average awesome (and one really bad solve will make your average terrible), and the high BLD that very good solves and very bad solves happen much more often than in things like 3x3 and 2x2. In fact fully half of the single WRs (25.09, 23.65, 19.46, 16.45, 13.56) were also achieved in world record means! I think this is a pretty good indication that we need average of 5.
jbcm627 (2008-11-23 22:21:10 +0000)
I really agree that Sq1 and Clock should be avg of 5. As qq said, they really don't take much time anyways, etc, etc. I do like the start/stop idea, but I do agree it could be hard to regulate. You could cover back up the puzzle after the timer has been stopped or something similar, but this could become a messy procedure. To that end, what would be nice would be timers that could take splits, or could keep running in the background until reset. This way, even if a competitor stops the timer, they can still rectify any mistakes as the timer will keep going.
qqwref (2008-11-23 23:34:37 +0000)
Ah, that would be a cool idea. So the timer just keeps running, and the time is just the first time you stop the timer after the cube is solved, but with a (say) +5 penalty for each time you stopped it during the solve? That could work, although it would still change the spirit of a competition a bit.
Ron (2008-12-21 18:30:30 +0000)
For 2009 draft 1 I changed the following: Clock: average of 5 Square-1: average of 5 Megaminx: mean of 3 Ron