New type of 2x2 scrambler

[TMOY]

14 Erik Johnson 1.72 USA US Nationals 2009
20 Phil Thomas 1.86 USA Wisconsin Open 2008

maybe more.

IIRC Doudou's 1.66 solve at Toulouse Summer Open 2008 was 8 moves.

 

[Kenneth]

Look at this: sub-1.5 single, 7 second average - is that fair, and is the single solve representative of that person's skill? No, it is not.

If you think that's the worst example then look at this:

Kristianstad Open 2009 First 18 3.33 13.49 17.03 12.13 15.55 3.33 12.80

That was my brother. In the same round Simon Westlund broke the NR with 3.00 but that's more fair, he is one of our best 2x2 cubers.

 

[qqwref]

TMOY: Interesting. A 1.66 by Edouard would make a good single WR as the time is still beatable by a fast solver with a good (but non-ridiculous) solve, and as he is fast at the event and thus deserving of a record. (I don't mean offense to Erik, but rather to people who are slower than me at 2x2 but have had their lucky sub-2 single.)

Kenneth: Actually that IS the worst example I could find... your brother has a single/avg ratio of 0.296 and a single/avg rank ratio of 0.126, whereas Antoine has a single/avg ratio of 0.199 and a single/avg rank ratio of 0.006 (!!!). So I think Antoine's case is definitely more unfair to other competitors such as me.


I have heard that in Worlds there was one group of 2x2 competitors who had the extremely easy scramble, and another group which did not. This is an extremely unfair situation and I'd like to remind anyone who reads this topic that the whole purpose of removing extremely easy scrambles is to remove these unfair situations where people feel disadvantaged by not being in a particular competition or solving group. The single ranking for any event, even 2x2, should be determined more by the person's speed and luck than by the scrambles they are given.

I am sure it seems like I am making a big deal about nothing, but if the WCA's goal is to have "more competitions in more countries with more people and more fun, under fair conditions" then I think if an unfairness is found it needs to be corrected quickly, so we do not risk people feeling that we are as a whole amateurish and unprofessional.

About the 8 moves limit: Of course this can be changed at will (by modifying a single number in the code, it is not hard) but the question is then "to what?". On average a 2x2 takes 8.755 moves to solve so we do not want to pick something too far from this, but on the other hand we don't want to eliminate too many scrambles. I think discarding everything under 6 moves might be a good compromise, but this is after all still a bit subjective.

 

[DavidWoner]

TomZ's idea of discarding scrambles with 5 or fewer moves is a good one although it is still somewhat arbitrary.

Uhh, it is neither TomZ's idea nor arbitrary... Did you not read what I wrote?

As I recall, Javier's 1.63, Patrick's NAR, Eric's current NAR, and this most recent scramble from WC were all FOUR moves. Not five or six or seven, but FOUR. So wouldn't discarding all positions <5 moves from solved make a lot more sense?

<5 is not arbitrary because all of these slow people are getting their fast singles on 4 move scrambles.

 

[AvGalen]

On the WR-scramble Erik got the WR because he is really good, not because it was a lucky scramble. Many other people got the same scramble but either don't possess the skill to see the solution or don't possess the speed to beat his time. Even on a 4 move scramble nobody has beaten that 7 mover (Erik and I are still both convinced we actually performed 8 moves although we have redone it many times and always used 7)

If a scramble would show up that has 1 layer done only some people would know how to solve it. Your scrambler would actually increase the chances of that happening compared to the current scrambler.

From a theoretical point of view I can see 2 problems:
- Not all positions might be reachable in exactly 11 moves. (I believe they can, but that's not proof)
- Deciding that "<8 moves scrambles" are easier and will result in faster times hasn't been proven for me.

Changing the scrambles in this way would make 2x2x2 a different event and would invalidate old results. I don't think that is needed. The solution is worse than the problem

 

[hr.mohr]

I don't mean offense to Erik, but rather to people who are slower than me at 2x2 but have had their lucky sub-2 single.

So what you are saying is that it's OK to be lucky if you are faster than you?

I like that we all get a random state scramble and the chance to get lucky but should practice because the winner are found by average and not by a single lucky solve. Even if 5 lucky scrambles were to be found in the same round the winner would still be the one with the fastest fingers.

 

[Cride5]

I have heard that in Worlds there was one group of 2x2 competitors who had the extremely easy scramble, and another group which did not. This is an extremely unfair situation and I'd like to remind anyone who reads this topic that the whole purpose of removing extremely easy scrambles is to remove these unfair situations where people feel disadvantaged by not being in a particular competition or solving group. The single ranking for any event, even 2x2, should be determined more by the person's speed and luck than by the scrambles they are given.

I have to agree with qq on this one. If its in our power to make the system a fairer one then why not do it? While removing < 8-move scrambles won't guarantee against lucky solves, it will certainly vastly reduce their frequency.

Additional filters could be built into the scramble generator to remove known easy scrambles, for example: scrambles which only affect one layer, or result in solved orientation etc. However, the problem with going down this route, is that the design of the filter is likely to result in bias towards particular methods. For example: removing 1-layer scrambles will result in a penalty for CLL users, while Guimond users will be largely unaffected.

 

[Pedro]

the single 2x2 ranking is indeed pointless, and so are almost all the other single solve rankings (except for the blds)

luck is part of the game and I don't think we should try to remove it.

It's not fair to just erase a lot of times. What happens with the averages?

the winner is decided by average, so even if someone gets a stupid scramble, they won't win if they're not good enough

I have heard that in Worlds there was one group of 2x2 competitors who had the extremely easy scramble, and another group which did not. This is an extremely unfair situation and I'd like to remind anyone who reads this topic that the whole purpose of removing extremely easy scrambles is to remove these unfair situations where people feel disadvantaged by not being in a particular competition or solving group.

If I remember correctly, there were like 4 groups for 2x2 at Worlds.

Not everybody is going to be at all competitions, so some people will always get an "advantage". It just happens like that. If everybody got the same easy scramble at Worlds, would it be more "fair" with all the thousands of people which weren't there?

 

[Kenneth]

Kenneth: Actually that IS the worst example I could find... your brother has a single/avg ratio of 0.296 and a single/avg rank ratio of 0.126, whereas Antoine has a single/avg ratio of 0.199 and a single/avg rank ratio of 0.006 (!!!). So I think Antoine's case is definitely more unfair to other competitors such as me.

Yes, maybe so, but the span between single and average for my brother is over 10 seconds and that is a bit... lolable

I'm not making jokes on his behalf here, he nows he sucks at 2x2, it is the scrambler that is funny, not he.

 

[qqwref]

I don't mean offense to Erik, but rather to people who are slower than me at 2x2 but have had their lucky sub-2 single.

So what you are saying is that it's OK to be lucky if you are faster than you?

Not quite, but I am saying that people who are slower than me (and I am not even fast) do not really deserve such a fast single time. If someone is much faster than me, maybe a sub-2 single can be gotten on a normal solve.

the single 2x2 ranking is indeed pointless, and so are almost all the other single solve rankings (except for the blds)

luck is part of the game and I don't think we should try to remove it.

It's not fair to just erase a lot of times. What happens with the averages?

the winner is decided by average, so even if someone gets a stupid scramble, they won't win if they're not good enough

But that is the thing, I don't think the 2x2 single ranking HAS TO be pointless. Anyway, I don't want to 'erase' the times, just mark them somehow so that they don't show up on the single ranking. A lucky single does not affect an avg5 much if at all.

 

[tim]

A lucky single does not affect an avg5 much if at all.

Why does this thread exist then?

 

[qqwref]

A lucky single does not affect an avg5 much if at all.

Why does this thread exist then?

I think this could restore some legitimacy to the 2x2 single solve event.

 

[esquimalt1]

I'm still with qqwref on this. I think it's much more fair, and it's much nicer to not have an event (such as 2x2 single) that has to be about luck.

 

[hr.mohr]

You guys do know that there is no single solve event?

There's only an average event and a wall of fame for people that are lucky and skilled. The same scramble at Geneva where handed to several people and yet only one a few were able to break the current wr at the time. Luck is a big part of records but not the largest part.

 

[hawkmp4]

http://mzrg.com/miniSites/scramblers/222 11moves.html

I'd like to propose for this to become WCA-official but this type of thing isn't under consideration right now so there is no point in posting it in the WCA forum yet. Anyway this is a modification of the current WCA-official scrambler. Here are the changes:
- The official scrambler does 500 moves to find the random position, but this one is better because it simply picks a random position directly.
- This scrambler will discard scrambles that are 7 or fewer moves from solved (this happens <20% of the time). This means that with this scrambler there are no incredibly easy solves! I think this could restore some legitimacy to the 2x2 single solve event.
- The scrambles here are always 11 moves (the maximum number of moves a 2x2 scramble could require) so there is no way to tell how easy a scramble is by looking at the moves.
- I removed the "minimum scramble length" thing because that was not necessary.
- I cleaned up the code a LOT, removing any unused functions and rewriting some things to make them smaller without changing what they did. (For comparison's sake, this is 10kb and the official one is 16kb.)

In order of your points-
-I think that this is the right way to go. No reason we shouldn't be doing this, especially if we're doing it for 3x3. Makes no sense not to for 2x2.
-While I agree that 2x2 SHOULD have something like this, implementing it...well. I think throwing out 20% of the scrambles is a bad idea. I'd rather see more like 5%, MAYBE 10%. A lot of these scrambles are short movecount wise but as many people have said, it may not be easy to see. I think everyone can agree though, a 3 move scramble isn't really representative of anyone's abilities. (More on this in the conclusion) Also, though unfortunate I think, there ARE already records and rankings, and without a reaaalllly compelling reason I find it hard to either take away records and such from people who have already solved, or take the opportunity to break records away from future solvers. For better or for worse, 2x2 single is ruled by luck. It's due to the nature of the puzzle, and if we mess around too much with removing scrambles from the pool...well, it just doesn't seem right to me.
-With all due respect, what's the point? If this is intended for competition, scramble length won't matter. Competitors will never see the scramble algorithm nor see their cubes being scrambled so I don't think this is really necessary unless I'm missing something. I think in general, less moves is better for scrambles. Less moves to screw up leads to better consistency.
-Not much to say on the last two points. Good work.


I think that the best and most fair we can do is rely on a truly random solver and the distribution of move counts for optimal solutions. Besides the practical implications of cutting out certain scrambles and the effects of that on records, we still have the issue of judging which scrambles are 'too easy' or not. And I think that's better left alone. That's just my thoughts. Good work on the scrambler though- I think I'll be using it for practice now.

 

[Tim Major]

You guys do know that there is no single solve event?

There's only an average event and a wall of fame for people that are lucky and skilled. The same scramble at Geneva where handed to several people and yet only one a few were able to break the current wr at the time. Luck is a big part of records but not the largest part.

True, which is why introducing this would mean introducing a whole new event, due to the almost impossibility of breaking Erik's record with proper scrambles.

 

[qqwref]

ZB: Erik's solve was 7 moves. If we keep 7-move scrambles, we will clearly not need a new event because anyone who got a faster solve would be able to break Erik's record. Besides, do you think we need a new 3x3 event because 7.08 is impossible to break? Clearly not, even though nobody else has done a sub-8 yet.


hawkmp4: I said 20% because the number of scrambles up to 7 moves is <20% while the number of scrambles up to 8 moves is >20%. The number of scrambles up to 7 moves is under 10%. The 20% number was just used because that is our definition of what constitutes lucky.

For the purposes of math:
if we discard every scramble up to 7 moves we discard 7.89% or about one in 12;
every scramble up to 6 moves, 1.70% or one in 60;
every scramble up to 5 moves, 0.33% or one in 300;
every scramble up to 4 moves, 0.06% or one in 1600.

 

[Dirk BerGuRK]

I would be interested in seeing a poll about this subject. I think this is certainly a topic worth discussing. I seem to remember a discussion on TwistyPuzzles a while back about removing lucky scrambles for a 3x3 that didn't receive much favor, although this is a different scenario since entire solutions can be seen during inspection.

I have seen a few one and two move solutions appearing on hi-games, but I have never heard about any appearing in a competition. If these appear are they thrown out?

 

[Tim Major]

ZB: Erik's solve was 7 moves.

Timer start, pick up cube, execute 7 moves, drop cube, stop timer. This was done in 0.96? WTF?!?! Did someone recreate the solution? I would be interested to see it. I can barely do 4 moves in that time, and that is 4 random moves. Insanity.

 

[Anthony]

I'm not really for or against using this scrambler, whatever happens happens. But, I can think of one situation where easy scrambles definitely influenced an average. Check this out.

Michal Robaczyk is ranked 17th in the world with a 3.97 average. Here are his times. 1.66, 5.48, (1.55), (6.63), 4.77.