Hi guys, Should we filter optimal scrambles that are "too" short? And if so: for which events, and what would be a good lower limit of number of moves? In 3x3 the chance of an optimal scramble of less than 13 moves is extremely small. But in 2x2 and Pyraminx I have seen very short optimal scrambles. Have fun, Ron

As far as i experienced, both pyraminx and 2x2x2 had really short scrambles in competition without any spectacular results. For 2x2x2 a 4-5 move scramble came many times and as far as i know not many competitor found the optimal solution. The unofficial 2x2x2 times are scary at home because everyone scrambles the cube for themselfes. Without knowing the length of the scramble, it's very hard to find the optimal solution. In my opinion we should filter the "too" easy scrambles which in the case of the 2x2x2 can be 9-10 move long, and still very easy to solve. The scramblers of the round should ask the main judge if they found a very easy scramble.

[quote="Olivér Perge":f08zoz5u]In my opinion we should filter the "too" easy scrambles which in the case of the 2x2x2 can be 9-10 move long, and still very easy to solve. The scramblers of the round should ask the main judge if they found a very easy scramble.[/quote:f08zoz5u] The problem here is how you define "easy to solve". I personally don't know any special 2x2x2 stuff so if I'm scrambling then I will not catch any "easy" cases other than LBL stuff.

I just made a little experiment. I applied short handscrambles to a 2^3 and then solved them (I'm using Guimond, if that mattters). With 6-move scrambles I got mostly normal solves, only a few luckys and I couldn't see them at inspection. With 5-move scrambles, most of the solves were just the scramble done backwards. So IMHO 6 moves is a good lower limit, it still allows for lucky solves but not ridiculous ones.

5 moves is an easy number to brute force, or to make a precalculated table of forbiden scramble. 4 move solution are trivial to see. For 5 moves, it depend on the 5 moves.

[quote="deadalnix":37fdqu9j]4 move solutiona re trivial to see.[/quote:37fdqu9j] Maybe to you. Not all judges/scramblers can notice that, "Oh that one side has 2 faces of green and 1 blue, and the other corner has a blue to the edge". OK. I know enough that it's a first step of some solution. Can I "see" if that a LL skip if I do step 2? No.

Maybe it's not for everybody, but considering I'm 7 avg on 2x2x2, I guess it will be obvious for many cubers. The consequence is that single results for 2x2x2 just means nothing. It's mostly ramdom due to the luck people have with scrambles. And it have impact on the avg.

I don't think I quite support a minimum scramble, but here is a 2x2x2 scrambler that supports a minimum optimal scramble length setting: http://zandenonline.nl/22scrambler_minimum.html

[quote="MadsMohr":461qffbu][quote="Olivér Perge":461qffbu]In my opinion we should filter the "too" easy scrambles which in the case of the 2x2x2 can be 9-10 move long, and still very easy to solve. The scramblers of the round should ask the main judge if they found a very easy scramble.[/quote:461qffbu] The problem here is how you define "easy to solve". I personally don't know any special 2x2x2 stuff so if I'm scrambling then I will not catch any "easy" cases other than LBL stuff.[/quote:461qffbu] Exactly what the problem is. I have experienced optimal scrambles solvable in 8 moves that (due to lockups) only ended up in a time of like 1.90. At the same time I raced Derrick a couple of times with 5 and 6 move scrambles which were VERY hard to find a good solution for. I guess in the end we already covered the 'problem' by defining winners on the average of 5 solves and not for the best single time, plus otherwise we should rename speedcubing into 'solve a cube from a not too easy state to a complete state as fast as you can' ...

[quote="deadalnix":3fpr7hex]4 move solution are trivial to see.[/quote:3fpr7hex] Really ? 1.63 WR was a 4 move solution, yet nobody did find it. Why focusing on the single solve ? We all know that the current 2x2 single WR doesn't mean anything anymore.

I agree with Clément and as consequence I don't see a reason to filter short 2x2x2 scrambles. The current Top 10 2x2x2 single list mostly constists of such scrambles and even the current 2x2x2 average WR was supported by 2 of those. I think filtering scrambles takes a lot of legitimate chances to break these records in future competitions.

What is the matter of breaking record if it mostly based on luck ? We are not playing some lotery here. And I maintaing that 4 move solutions is really easy to see.

[quote="deadalnix":2fprdkwq]What is the matter of breaking record if it mostly based on luck ? We are not playing some lotery here.[/quote:2fprdkwq] The 2x2 event contains a part of luck, and we won't be able to change that. Remember that the current WR is 8 moves, so would you remove all scrambles of 8 moves and below ?

At the least, I don't think 5 or fewer moves should be allowed. I generated a bunch of 5-move scrambles and got a 2.38 rolling average of 12 (where my normal average is ~5) and almost every solve either had a skip or could be solved directly. I know there's supposed to be some luck in 2x2, that existing records may be very hard to break if we get rid of the shortest scrambles, and so on. But I still don't like the way things are going where someone (someone who isn't any better than me at 2x2) can be 200th in single, and then immediately drop to top ten in the world in one competition... without having improved at all, but because they got a very easy scramble. It's not fair because I didn't get that opportunity. Either we should remove lucky scrambles completely, or give everyone at least one attempt on them... why should I be ranked so high just because I am not lucky? Cubing isn't a game of luck.

To call it unfair that a slower competitor get's a lucky scramble is plain wrong. Yes, that lucky solve are not representative of the competitors skills, but that's why we don't decide the winner by a single solve. I can't see how we could regulate this in any way, and and I don't see why we should. The problem with regulating this is how we define a lucky scramble. If it's only defined by move count and putting the arbitrary limit at 5 moves, then we still could get scrambles like the WR scramble. I vote to keep using random state and masking short scrambles by using a fixed scramble length. This is to help reducing the risk of someone peeking at the scramblers and picking up on a possible lucky scramble.

Agreed with Mads. Single records don't mean anything and won't mean, that's why we use averages (except on bld events, of course). Changing the rules now is really unfair, because people who never competed before will have "harder" scrambles. Really, Michael, do you care that much about 2x2 single? What if it were [b:1pcx43gf]you[/b:1pcx43gf] who got a lucky scramble (and a WR, maybe)? Would you go and ask the delegate/main judge to disqualify your solve because you got lucky?

[quote="MadsMohr":cb79rnru]To call it unfair that a slower competitor get's a lucky scramble is plain wrong.[/quote:cb79rnru] Just because you think 2x2 single rank doesn't matter does not mean this situation is suddenly fair. In fact this situation is practically the definition of unfair. I can't say this enough, [i:cb79rnru]just because you don't care about an unfair situation does not mean it does not exist[/i:cb79rnru]. [quote="Pedro_S":cb79rnru]Changing the rules now is really unfair, because people who never competed before will have "harder" scrambles.[/quote:cb79rnru] I have competed before and have only gotten "harder" scrambles. Isn't it unfair to me that I didn't get one? Either everyone should get such a ridiculous scramble or nobody should. [quote="Pedro_S":cb79rnru]Really, Michael, do you care that much about 2x2 single?[/quote:cb79rnru] Normally I wouldn't. But I seem to be hovering around the 10th place mark for the sum of all single ranks, so this is an important thing to me, and I don't think my rank is a fair representation of my skill if someone who averages two seconds slower gets to do a solve that is two seconds faster than the best one I did. It's not like if I came across a trivial scramble I would suddenly take 3 seconds to solve it... [quote="Pedro_S":cb79rnru]What if it were [b:cb79rnru]you[/b:cb79rnru] who got a lucky scramble (and a WR, maybe)? Would you go and ask the delegate/main judge to disqualify your solve because you got lucky?[/quote:cb79rnru] Why? You don't understand at all what I'm trying to say. I'm saying that the existence of easy scrambles, overall, is a bad thing, not that some individual one is a problem. But if I was the first person to get a non-scramble, and set a WR... then YES! Of course I would ask the main judge to disqualify it! I'm an honest person, unlike some others in the WCA community. It is the same situation as Grzegorz Prusak - I'm sure you know about the Polish Open 2008, where he got a scramble that was far too easy and far easier than anyone else had ever got, and later agreed to have his world record revoked. Square-1 and 2x2 are similar events in that it is relatively common to get scrambles that are obviously very very easy.

Right, so what's exactly your idea? Clean up all 2x2 times ever and start again, with no stupidly-easy-scrambles? Also, you really care that much about sum of single rankings? I thought you'd know that single rankings aren't really a good representation of your skill...that's why we have averages (of 5 at competition, of 12 or even 100 or even more at home). Also, [quote="qqwref":388mewuw] I have competed before and have only gotten "harder" scrambles. Isn't it unfair to me that I didn't get one? Either everyone should get such a ridiculous scramble or nobody should. [/quote:388mewuw] You do realise that that's a "problem" that we cannot solve, right? Different people at different competitions will get different scrambles. Should we give everyone Erik's 7.08 scramble, so everybody has a chance of getting the WR? That's just not going to happen. That's the reason we have averages and several competitions. You're not stuck to a "bad" scramble. It's the same as in Athletics. Does everybody get to run in the same track, same day, same conditions? No. But they're all ranked in the same ranking, as long as certain standards are met.

[quote="qqwref":2fq1zl08]It is the same situation as Grzegorz Prusak - I'm sure you know about the Polish Open 2008, where he got a scramble that was far too easy and far easier than anyone else had ever got, and later agreed to have his world record revoked.[/quote:2fq1zl08] It was a mis-scramble. If he had simply received an easy scramble, his record should have stayed. [quote="Pedro_S":2fq1zl08]It's the same as in Athletics. Does everybody get to run in the same track, same day, same conditions? No. But they're all ranked in the same ranking, as long as certain standards are met.[/quote:2fq1zl08] But if the conditions turn out a little too unfair, certain results are disqualified from world record consideration. (Consider wind assistance in running..)

[quote="Lucas" [quote="Pedro_S":3fc4s3m1]It's the same as in Athletics. Does everybody get to run in the same track, same day, same conditions? No. But they're all ranked in the same ranking, as long as certain standards are met.[/quote:3fc4s3m1] But if the conditions turn out a little too unfair, certain results are disqualified from world record consideration. (Consider wind assistance in running..)[/quote] Yes, that's exactly what I meant. What I fail to see is how to "solve" this situation without making it even more unfair to the newcomers. Keep everything we have and start filtering now? Erase everything and start again?

How about this for now: remove all 2x2 single solves (from the single solve ranking only) which are less than 1/4 of that person's personal best average. If a solve was not really so lucky, the person should have another decent solve to replace it, so it won't matter a huge amount. I know that this isn't perfect, but I think it is good enough at nulling any absurdly unfair times, while still making it possible to get a PB single with a new scrambler. You will notice that with this idea all current continental and world records do not change. So no problems would arise there. Only the most ridiculous single solves would be affected.

[quote="qqwref":2bei6ru3]I know that this isn't perfect, but I think it is good enough at nulling any absurdly unfair times[/quote:2bei6ru3] You have yet to come up with a definition of unfair times. If there is really such a big problem with luck with this puzzle then you really should be arguing for it's removal as an official event. Crippling the event and removing "unfair" times based on your arbitrary definition are really pointless and actually a bit annoying.

[quote="qqwref":1mosrw3r]How about this for now: remove all 2x2 single solves (from the single solve ranking only) which are less than 1/4 of that person's personal best average.[/quote:1mosrw3r] Sounds a little arbitrary. I'm a little biased because my 1.06 single is less than my (bad) 4.42 average, but I don't think we should do this. Times should never be removed except in individual cases. If you really want to this, I would suggest starting over with a new 2x2x2 event. This is also messy, but we should avoid breaking the precedent of keeping previous results valid.

Maybe Mads is right - we should just remove the 2x2 single ranking entirely. I guess we could keep the ability to look at the list on the WCA site (for fun), but there's no need to have it in people's profiles and to keep track of world/national records. It is an index of luck, not skill.

[quote="qqwref":1684fhi9]Maybe Mads is right - we should just remove the 2x2 single ranking entirely. I guess we could keep the ability to look at the list on the WCA site (for fun), but there's no need to have it in people's profiles and to keep track of world/national records. [b:1684fhi9]It is an index of luck, not skill[/b:1684fhi9].[/quote:1684fhi9] Well, not exactly...give the same scramble (let's say a 4-move FL and LL skip) to Rowe and to me...he's of course going to have a faster time, even if I get the skip (I suck at looking ahead on 2x2)

[quote="qqwref":ljzs288x]Maybe Mads is right - we should just remove the 2x2 single ranking entirely. I guess we could keep the ability to look at the list on the WCA site (for fun), but there's no need to have it in people's profiles and to keep track of world/national records. It is an index of luck, not skill.[/quote:ljzs288x] No, no, no. Stop putting words in my mouth. I said that YOU should be arguing for removing the event ENTIRELY. This idea of removing the single ranking is just plain ridiculous. It's just a list of solves nothing more and nothing less. I don't think for a second that it's luck, not skill when I look at the top rankings. Erik, Rowe, Lucas. Yeah, no skill there...

[quote="MadsMohr":2rf5nohe][quote="qqwref":2rf5nohe]Maybe Mads is right - we should just remove the 2x2 single ranking entirely. I guess we could keep the ability to look at the list on the WCA site (for fun), but there's no need to have it in people's profiles and to keep track of world/national records. It is an index of luck, not skill.[/quote:2rf5nohe] No, no, no. Stop putting words in my mouth. I said that YOU should be arguing for removing the event ENTIRELY. This idea of removing the single ranking is just plain ridiculous. It's just a list of solves nothing more and nothing less. I don't think for a second that it's luck, not skill when I look at the top rankings. Erik, Rowe, Lucas. Yeah, no skill there...[/quote:2rf5nohe] If you don't want me to "put words in your mouth", maybe you shouldn't tell me what I ought to be arguing...? The problem is only with the single solve portion of the event. The average part is fine, but the single solve has far too much luck, and I don't really see the reason why it should be included on our profiles at all - the only real argument for looking at 2x2 single is that we do it for all other events. Basically I don't like the idea that profiles include a rank that is based more on luck than skill, because only a somewhat experienced cuber would look at the ranking and know that 2x2 single rank is worthless. I don't want a journalist or other non-cuber to look at a WCA profile and get a wildly inaccurate assessment of how good someone is. Anyway, about the ranking... look down a bit more. The next few people have average ranks of 248, 161, 654, 78, 5, 618, 148, 1063... The top few people may deserve their place, but it's pretty obvious to me that a good portion of the people in the top 2x2 single ranking got their achievement more through world-class luck than world-class skill. To me, the 2x2 single ranking is not by any means a list of the most skillful people, but just a list of the people who have gotten the most lucky, perhaps with a small amount of skill included in that rank. I don't know about you but I do not consider a 600 rank deserving of a top 10 result in the same event.

[quote="qqwref":p93u3wj8]I don't know about you but I do not consider a 600 rank deserving of a top 10 result in the same event.[/quote:p93u3wj8] And I don't consider this discussion deserving of our time. It's just a list of solves nothing more, nothing less. What you read from it is your own business.

There is no such thing as "single solve part of the event". Nobody does best of 1 rounds for 2x2x2. We do 5 times, take an average and rank people by that average. The single solve list is just that, a list. It's not a separated event. And, as you said, we have single solve lists for all other events, so it's just senseless to treat 2x2x2 differently. "Oh, it involves too much luck..." Yes, it does. That's why we use averages. Is Erik the best 3x3x3 solver in the world because he got a PLL skip and the 7.08? I don't consider so. He just happened to have a skip on a good solve. Media is stupid, anyway. But I think we'll get to a point where they will finally understand that the best cuber is measured by average and stop writing dumb stuff. Doesn't hur to dream, does it?

If the WR is 8 moves long, then we should change the regulatio now, or at least think about it when it's possible. Too short scrambles are uninteresting, and can possibly break results in a competition (for exemple with 2 groups).

The question was pretty simple, the discussion is going in different directions. Filter yes or no? For which puzzles? Which boundary per puzzle? Filter scrambles or filter results?

Another perfect example of the impossibility to make such a restriction on scrambles was a 2x2 scramble this weekend at Danish open (sadly I didn't get it since I was in a different group). The result of the scramble gave a solved 2x2 except for a 2 diagonal corners switched (ok you still had to do an R' move to get it, but that's the first move of my algorithm for it anyway). To switch diagonally 2 corners is 11 moves, which is one of the longest optimal cases there are (you can always solve a 2x2 in 11 moves), to make a long story short: this scramble was optimally solvable in 10 moves. It was obvious for a LOT of people if not all who got this scramble and did it in one look. Now tell me what is easier a 5 move scramble where no 2 pieces are together or this...

Nice to get the topic in a more strict direction ! Let's express in a cleaner way [quote="Ron":2ibzvzu3]Filter yes or no?[/quote:2ibzvzu3] I would say yes. [quote="Ron":2ibzvzu3]For which puzzles?[/quote:2ibzvzu3] At least for 2x2x2 and pyraminx. But I think it could be the same for any puzzle. Anyway, the probability to have a 5 moves or less solution for a 3x3x3 scramble is almost innexistant. So this isn't a big deal on puzzles others than 2x2x2 and pyraminx. [quote="Ron":2ibzvzu3]Which boundary per puzzle?[/quote:2ibzvzu3] 4 moves are easy to see. So I would say 5. [quote="Ron":2ibzvzu3]Filter scrambles or filter results?[/quote:2ibzvzu3] I don't really get that one. I think the computer could etablish a blacklisted position table. This isn't a big deal for modern computer to brute force any puzzle with a 5 moves depth. However, generator for 2x2x2 and/or pyraminx have a quite complete prunning table. So they already have all the needed data to handle filtering.

I think we should filter scrambles in the future, to prevent incredibly easy scrambles and unbeatable WRs, but I don't know any good objective limit. 2x2 and Pyraminx should be affected, and perhaps eventually Square-1; it's not necessary for other puzzles because of the low probability that such a scramble happens. (I say Square-1 because occasionally there are scrambles which do end up to be very very easy, and for instance I've achieved a few sub-10 times in practice even though my average is ~25.) For 2x2 and Pyraminx, it's reasonable to drop all scrambles with (at most) 4 or 5 moves.

This discussion is definitely not ready to be implemented in regulations. We will keep the regulations as is for 2010. But I would love to guide this discussion better for regulations 2011, because I think we do have a point here.