TNoodle - WCA scrambler replacement

[Stefan]

But with my idea, the length of a scramble would always be between 9 and 13 moves (and 13 moves would be really rare), and a 9 moves long scramble has no reason to be particularly easy.

Then why not just set the minimum number of moves to 9?

But I'd like the fixed 11 moves better, because saving two moves doesn't save much and setting a minimum like 9 is arbitrary (why not 8 or 10?).

 

[jfly]

Did you make it look for 3x3 scramble during 200ms?

I did not, sorry. Just did: https://github.com/jfly/tnoodle/commit/421ba4c53ed7efd01d01fcf071946ed7a9924177. I've also released 0.6.8 with that change.

Then why not just set the minimum number of moves to 9?

But I'd like the fixed 11 moves better, because saving two moves doesn't save much and setting a minimum like 9 is arbitrary (why not 8 or 10?).

Yeah, fixed seems fine to me. The number of moves saved is so inconsequential. Not that I see this happening, but lets not let this turn into a debate about scramble filtering. TNoodle does not filter any scrambles. To do is a decision the WCA would have to make, not TNoodle.

 

[Vincents]

Yeah, fixed seems fine to me. The number of moves saved is so inconsequential. Not that I see this happening, but lets not let this turn into a debate about scramble filtering. TNoodle does not filter any scrambles. To do is a decision the WCA would have to make, not TNoodle.

It does mask scrambles, however. So people shouldn't know whether each scramble is actually easy. If you're going through the trouble of trying everything to see if it is, then you shouldn't be competing.

 

[TMOY]

It does mask scrambles, however. So people shouldn't know whether each scramble is actually easy. If you're going through the trouble of trying everything to see if it is, then you shouldn't be competing.

That's precisely my point. It's possible to realize that a 4-move scramble is coming without really paying attention, just by noticing that the scramblers seem to go faster than usual. But if you're trying to tell 9-move scrambles for 10-move ones then you're definitely cheating.

 

[cityzach]

Why Tnoodle of all names? haha

 

[jfly]

Why Tnoodle of all names? haha

Quoth Ryan (the guy I made CCT with):

Ryan: ccT scorcHed And Its remaiNs OODLEd.
me: LOL
i like it
Ryan: Let's call it tnoodle

There's a restaurant in Berkeley called Thai Noodle where the cubers often go to eat.

 

[qqwref]

I would like to say that TNoodle is too silly and undignified of a name for the official scrambler used by the international speedcubing authority. You guys can do whatever you like - I don't even go to competitions anymore - but I think it's a really poor decision to call it that.

That's precisely my point. It's possible to realize that a 4-move scramble is coming without really paying attention, just by noticing that the scramblers seem to go faster than usual. But if you're trying to tell 9-move scrambles for 10-move ones then you're definitely cheating.

The difference between a minimum of 9 moves and a minimum of 11 is negligible - knowing the scramble length in either case provides no useful information. The reason for the specific two starting moves is that the solver does a depth-first search through all 11-move solutions and prints the first one that fits, so if the first set of the first two moves it tries gives a solution, it won't even try the other starts.

For the record, the default (suboptimal) qqTimer scramblers use a minimum 2x2 length of 9 and a minimum pyraminx length of 8, and this doesn't generally have the issue of commonly repeated moves. Those two lengths are the lengths with the most positions in a God's Algorithm table.

 

[CarlBrannen]

In that case, wouldn't it be better to add U R at the end of every 2^2 scramble and L' U' at the beginning of every pyraminx scramble and do the appropriate cancellations ? The state would be just as random and the scrambles would be shorter on average.

I agree completely. That the scrambles always look like that is an indication that the scramble code is not very optimal. Basic Shannon information theory.

 

[qqwref]

I agree completely. That the scrambles always look like that is an indication that the scramble code is not very optimal. Basic Shannon information theory.

It's not *supposed* to be optimal. That's the point.

 

[cuBerBruce]

The maximum number of moves for FMC is 80 (including rotations), but you give 99 spaces to enter a solution. Why not offer just 8 rows of 10? It's easier to count that way, anyway, and alerts a competitor that their solution is too long if it doesn't fit.

Because people like to write out their solutions spaced out within the form, so it easier to make corrections and change a solution to use an insertion. The form should give you space for at least 200 moves.

 

[Bob]

Because people like to write out their solutions spaced out within the form, so it easier to make corrections and change a solution to use an insertion. The form should give you space for at least 200 moves.

No, I really think that's what the scrap paper should be for. A solution can easily be transferred in the last couple minutes. The judges should only have to read a complete, clear and legible solution.

 

[Vincents]

Because people like to write out their solutions spaced out within the form, so it easier to make corrections and change a solution to use an insertion. The form should give you space for at least 200 moves.

You are allowed scratch paper.

 

[Sebastien]

Because people like to write out their solutions spaced out within the form, so it easier to make corrections and change a solution to use an insertion. The form should give you space for at least 200 moves.

I don't like the design to much as well, but in fact, there is no need to actually use that paper for the solution.

 

[clement]

Especially when trying to reach a certain length (like 11 moves for 2x2 and pyraminx), because the solver can often use its favorite first few moves and still solve from there in the remaining number of moves.

Correct. If people don't like to see the same rotations at the beginning/end of scrambles, the order of moves during the search can be made random.

 

[cuBerBruce]

No, I really think that's what the scrap paper should be for.

When copying your solution from the scratch paper to the form, it's possible to accidentally leave out a move or make other mistakes. To fix the problem properly (in complete compliance with the rules stated on the form), you might in some cases have to pretty much rewrite the entire solution, perhaps even need a new form. It's always wise to spread out the solution so if you do make a mistake copying, you can fix it with a rather minimal amount of crossing out and rewriting. If the competitor is allowed some leniency in filling out the form (such as inserting a move in between two positions on the form because of a copying mistake), then I would like to see some sort of guidelines about that.

Also, let's say I come up with a solution with around 5 minutes left. I may want to write out the solution on the form right then to make sure I will have the solution ready when time expires. Then let's say I happen to find an insertion that improves my solution (even though I didn't expect I would in the time remaining). I may feel I don't have enough time to write out the new solution from scratch on a new form. Before these forms were used, I would simply have made sure to leave enough extra space around the moves of my solution so I could cross out some moves and write in some new moves in between two of the lines of my initial solution to create my new solution. But with the limited space you're allowed to use for the moves on these forms very greatly constrain the competitor's ability make this type of change. So these limited space forms actually create added constraints on the competitor that are not in any way specified or implied in the regulations. This is why I object so vehemently to these forms with such limited space/flexibility to write out the solution.

A solution can easily be transferred in the last couple minutes.

I note that the competitor is not guaranteed to be warned when there are two minutes left. A competitor also risks getting a DNF if for some reason he/she fails to copy the complete solution during the last two minutes. It can be very risky to put off writing out the solution.

The judges should only have to read a complete, clear and legible solution.

Spacing out a solution does not make the solution incomplete. I don't see that spacing out a solution makes a solution unclear or illegible. It seems quite clear you are only concerned about the judge here and not the competitor.

I don't like the design to much as well, but in fact, there is no need to actually use that paper for the solution.

I'm not clear what you mean. Are you saying if you have a clear legible solution written on a scratch paper, you can hand that in to the judge as your official solution instead of the official form? I'm not so sure Bob would agree with that if that's what you mean.

 

[Pedro]

I'm not clear what you mean. Are you saying if you have a clear legible solution written on a scratch paper, you can hand that in to the judge as your official solution instead of the official form? I'm not so sure Bob would agree with that if that's what you mean.

This is what the regs say:
E2c) At the end of the 60 minutes the competitor must have his solution written down clearly for the judge, notation according to Article 12. Penalty: disqualification of the solve.

It doesn't say you need to hand off the "official" sheet or whatever. You may write your solution on a napkin, paper bag or something else.
What we try to do with these sheets is make it easier to judge/count the solutions, but nothing stops (or at least nothing should stop) the competitor from writing the solution on the back of the sheet or another paper.

Also, I don't see the logic in counting rotations for the maximum lenght if we don't count them for the final result...

I agree that anything over 80 moves is just competitor having fun or whatever, but rotations shouldn't count here.

 

[Sebastien]

This is what the regs say:
E2c) At the end of the 60 minutes the competitor must have his solution written down clearly for the judge, notation according to Article 12. Penalty: disqualification of the solve.

It doesn't say you need to hand off the "official" sheet or whatever. You may write your solution on a napkin, paper bag or something else.
What we try to do with these sheets is make it easier to judge/count the solutions, but nothing stops (or at least nothing should stop) the competitor from writing the solution on the back of the sheet or another paper.

This. To use terms like "official Paper" or "scrap paper" is complete nonsense as such things don't exist for FMC.

Also, I don't see the logic in counting rotations for the maximum lenght if we don't count them for the final result...

I agree that anything over 80 moves is just competitor having fun or whatever, but rotations shouldn't count here.

It is to prevent competitors from having fun with lots of rotations then.

 

[qqwref]

1) We can hand in a solution on scrap paper? That would have saved me some real hassle last time :| I remember having to write really carefully to make sure the dark lines under each move on the official sheet didn't obstruct the moves and make it hard to tell apart stuff like B, R, F. This should be explicitly pointed out at the start of the round if it isn't already - in my experience people's experience with standardized testing makes them very wary of using non-official answer forms unless they are specifically told they can. Nobody wants to write down a valid solution and have it be DNF'd on a technicality.

2) Rotations ought to count as moves for FMC. Yes, I know we aren't used to thinking of them as moves; but then again we also aren't used to thinking of an M or M' as two moves in normal speedcubing. This would both ensure a more consistent movecount (between the move limit and the scoring system), and also help the judges: nobody can abuse the system. With the current rules someone could add a rotation or two in between every move of a 30-something-move solution and receive no penalty whatsoever. (I imagine someone handing in a solution written in <U,x,y,z> for the lulz.) And it would provide an incentive for people to eliminate rotations from their solution.

 

[ben1996123]

There are official FMC solution sheets? We didn't have any at rapid dash open, just handed in a scrap piece of paper with your name on and a box around your final solution.

 

[Pedro]

I don't think rotations should count, because they don't affect the cube state as the turns do.

If someone makes fun of the judge with lots of rotations, DNF the person...