vendredi 16 octobre 2015

Some thoughts about WSC 2015 in Sofia


In case of some reader wonder about what these 3 letters means: W means World, S means Sudoku, C means Championship.

I just came back of a very emotional and intense week, taking part in WSC and having the chance to be invited to compete in the sudoku GP final, which was an incredible experience for me that I will probably report in another post.


Good news are that we had a decent WSC, with lot of very high quality sudoku, it was very fun to solve them. Secondly, on a personal note, it is the first year I had the strong feeling I played at a very good level during the 2 days, not playing perfectly, but fast and efficiently - This was my goal this year as I was a bit disappointed last 2 years, not about my ranking but more concerning the way I played. This led me to reach the official 13th rank, which is by far my better result in a WSC.
Kota Morinishi is World Champion second year in a row, other medals goes to Tiit Vunk and Jakub Ondrousek.

Now about bad news...
A lot of weaknesses appeared this year. Without revealing hearsay, I think that it would have been a catastrophic event without 2 things:
- The fact that organizers had WSC 2014 as reference. In 2014, the competition director Tom Collyer proposed a very coherent championship and even if I've the feeling that organizers of this year's WSC didn't even know in which way it was coherent and a very successful event in 2014, the fact that they basically copied all rules and did the same model of competition made that we didn't lose all the consistency.
- Even more important: A single guy, devoted and very professional, external to the organization, saved the WSC. Without his work before and during the event, I don't know if the WSC would have occurred ! He transformed alone what would have been the probably worst ever WSC (I wasn't there each year, especially in Zilina in 2009 where reports showed serious deficiencies in several aspect of the event) to a decent level competition.

The only thing we can hope right now is that it will not repeat in the next few years, in which case I fear that our discipline will be in a serious crisis. Let's hope that the organizers of next years competitions will be more involved and have more professional spirit.


Now about the competition itself.
It seems to be the tendency this year to allow puzzles that are not sudoku in sudoku competitions. Of course, this affirmation is very personal as it's quite impossible to have a strict definition of what is a sudoku which includes all its variants appearing in competitions and being commonly accepted as sudoku. This year I already had personal concerns on a sudoku in the 2th GP round, which was discussed here. The same applied in the 4th round concerning the puzzle named "sukaku", but I didn't protest as I already explained my opinion on the forum for the 2th round.

This WSC contained a few puzzles that I would not have considered as sudoku:
  • In round 2: 1234 "sudoku": The common classic sudoku rule to have a set of distinct digit was stretch at the limit. a digit repeated 4 times, another 3 times and a third one 2 times in the set. I think you have to be careful if you stretch a basic rule of the sudoku. Little stretches can be considered as remaining in the sudoku category, for example if for some purpose you want to use a set where a single digit is repeated twice. Here it went too far and I personally think it was not an appropriate puzzle for WSC. I heard a few players, including a top player that agreed with me. The further problem, if someone think the fact that some puzzle is not considered as sudoku is not enough to remove it, or if someone consider it is appropriate because he has the feeling it is a sudoku (I'm not a sudoku authority, so I think someone could argue in another way of my personal opinion). So I said the further problem was that it was a hard puzzle (95 points). I think organizers should be inspired, in case of including variant that could potentially be controversial, to propose an easy puzzle, so that the ranking is not much affected by it. It was the case for the second and third one I want to speak about. Here I didn't care that much when the instruction booklet was published without the point distribution, and then, when the final instruction booklet appeared (2 days before the competition) I was afraid of the 95 points it represented. Fortunately I had the time to create a training puzzle and guess what: I solved it during the competition.
  • In the 6th round: Overlapping "sudoku" and free block "sudoku": Overlapping sudoku was in fact overlapping latin squares. Latin squares are not sudoku, and I don't see in which way that can be changed by overlapping them. But it was worth only 20 points so who cares?
    The same applies for Free Block sudoku, which worth 35 points and appears in a round where no players was able to solve all puzzles. I just skipped them.
  • The 4th round: straight "sudoku": Here it was the worse choice an organizer can do with a puzzle that can be controversial to be a sudoku: a)The whole round consisted of this puzzle, b) it was worth 200 points, which means that it was designed to be a very hard one, c) a single mistake and the player would have 0 points for the entire round. I was very angry of all these bad decisions. It was a painfull round for me, I had no fun at all and if I wouldn't care a bit about my ranking, I would have left the competition hall after just a look at what the competition puzzle looked like.
The conclusion of it is that I believe the 4th round was far to test the players ability to solve sudoku, hence it will perhaps sound egoistic, but if we consider that this round has nothing to do in a WSC, we'll have another ranking before the play-offs where I would appear in the 6th rank.
I accept the rules of competition, so I'll always say that my official ranking was 13th, but I firmly believe that my performance showed that I was the 6th best sudoku solver during the 2 days of WSC.


Yesterday a WPF guidebook was published on the WPF forum. The purpose is to standardize the rules for the WSC/WPC hosts. It's a good sign that the WPF want that some things change and that the federation sees that some problems need to be solved. I'll take this opportunity to share my views there and see where it goes. The section named "puzzles" needs to be clarify much more in my opinion. I'll probably do lot of suggestions during the few next weeks and see if my point of view is the same as the commonly accepted one. In any direction that it goes, it can be an improvement. However if it goes at the extreme opposite of my views, I may consider extreme solutions: retiring myself from sudoku competitions or at least WSC, asking to be ranked unofficially during WSC and allowing myself to skip some parts of the competition, etc... It would be the death in the soul, but it shows at which point I'm affected when I feel that some people's goal is to make sudoku players guilty of not liking other puzzles.

Feel free to comment this post, even if your opinions are not the same as mine.

6 commentaires:

  1. Hi Fred,

    I agree to most of the things you said about Sudokus. Especially about Round 4, even if I haven't ever seen the competition puzzle.

    The type is clearly not a sudoku.
    Even as a puzzle, it feels mediocre and boring, especially in the given size. (But thats personal taste.)
    Round concepts consisting of a single big hard puzzle are just plain bad.

    But there are also some points I want to remember:
    First, there was partial credit. The total number of score 0 is 44 out of 186 (official and inofficial) contestants. Last years Double Jigsaw round was worse. The number of score 0 out of the Top50 contestants is 5. From the Top50, all but 12 people managed to completely solve the puzzle and got bonus points. You got 86 points on the puzzle according to the final results, so you lost maybe 180 points to your contestants for the Top10 spots, but I think every one of them jumps a lot if they scratch their worst performance. Realizing that one has to be careful while solving this puzzle and adapting solving speed is in my opinion a skill worth testing in a WSC.
    And last, looking at the final results, it seems to me, that starting from place 6, the point differences are much smaller than one would expect. Place 6 only has a 5% edge over place 16 or 150 points. Thats 3-4 medium difficulty puzzles over a solving time of more than 6 hours. I think thats very unusual and might be a reason for the potential big ranking jump.

    I know it feels bad that this one puzzle might have cost you the playoffs. Hope you are not so frustrated, that you stop solving, there are enough people that don't want to compete anymore because of such things.


    1. Hi Christoph,
      Thanks for the comment.
      Theoretically I agree with you that sprint round containing only hard puzzle are better when a lot of players are able to solve it, as I agree that having partial points is better than all or nothing.
      The problem with these rounds where a lot of players complete the round is that it's very difficult to concentrate after the first few players having submitted. Then you just hear "finished" each 5-10 seconds from each corner of the hall with different voices. I'm probably able to solve a sudoku while being half concentrated, but it's way harder when it's not a puzzle type I master. I managed to finish the round 9, which was the other "one puzzle sprint round", but I had big frights because I did a lot of mistakes due lack of concentration.
      Also you're totally right concerning the very tight ranking between place 6 to place 16. Thus I know everyone in this gap could feel he would have deserved a place in play-offs. Everyone have bad rounds during the 2 days, or did a few mistakes, etc... so that each player could be a bit frustrated being so close of qualifying to the play-offs.
      I had also a bad round and some critic moments with small mistakes to fix, etc... My bad round was round 5, where I solved 2 times entirely the consecutive clone (very nice sudoku) having a mistake at the end both time. But here it's not the question of having a bad round, it's the question of inappropriate puzzles.
      I'll reassure you that I'm not that frustrated, finally the ranking isn't the most important, and I respect each player, having better ranking than mine or not.
      I'm more tired about the fact that finding inappropriate puzzles (in a personal view of the word "inappropriate") is more and more frequent, and my nervous tension before and while discovering some instruction booklets is sometimes too high.
      If I decide to retire from sudoku competitions, it'll certainly not be only because of this puzzle. But if the turn of having more and more non-sudoku puzzles that contaminates sudoku tournament, I'll end up being disgusted...

  2. I might be a bit biased being a puzzle solver. I agree that 1234 Sudoku stretched it too much, and I also agree that Overlapping is not a Sudoku. Repeating digits/ no regions, are a definite problem.

    I disagree on both Free Blocks and Straight though. My arguments are pretty simple.

    Free Blocks - It is basically a dynamic Deficit Sudoku. I think if Deficit is accepted, this should be.
    Straight - Again, Deficit being accepted, and Renban Groups being accepted, and 12x12 Sudokus being accepted, I think this was a hybrid of those, with letters to reach a theme (card pack).

    I will however say that I wasn't a fan of the Straight round either, but that is for different reasons. I think the rule of getting a 0 for one mistake could have been replaced by just having negative points for incorrect digits. Other than that I was fine with the puzzle but I generally don't expect much (which says a lot considering I am now very unhappy with the WPC).

    Lastly, I would argue about the 6th rank bit, because the sequence of rounds affects every player in its own way. Some player may have had a bad round 5 after not doing well in Round 4, or become complacent after doing very well. In the end you cannot predict these things. Its sad you missed out on the playoffs, and I do think the round had a negative impact on you and other solvers because of the strict restrictions. I just feel the problem was more about awarding a 0 for even one incorrect digit, And the focus of this being a single standalone round, than it being a non-Sudoku puzzle.

    1. Hey Prasanna, thanks for your answer and thoughts.
      Before arguing, let me say big congrats to you with an impressive performance, including during the play-offs !

      I think we must be very careful with the reasoning like: "this one is accepted, so this one should also be accepted". This is typically what makes a definition of sudoku impossible. Think about a definition, then say "Aouch, this commonly accepted variant doesn't fit the definition, we have to change the definition" and then with your new definition, you have to accept a bunch of new variants that are further from sudoku that the one you wanted to bring into the rules.
      You say repeating digits is a problem. Then we should not accept star battle sudoku (round 1 and 4 of 2014 GP), or this one: . If we accept star battle, then does it means that we have to accept 1234?

      You compare free block with deficit sudoku. Your link with deficit sudoku is absolutely right. But: Is deficit sudoku fully accepted? I would say "yes, under conditions". Even Thomas Snyder expressed reserves about deficit sudoku: . [Blackout is also concerned here, and that remind me the missing digit sudoku of round 6 which stretch also the basic rules of sudoku in the same way that blackout sudoku. I didn't speak about it, because I think this was closer to sudoku than blackout sudoku, but striclty speaking, even if I accept it and have no complaint, I would not say it's a sudoku.]
      Commonly accepted deficit sudoku have normally region of size N-1 (I may have seen others somewhere, though), and only one cell has no local constraint. Would you accept deficit sudoku with region of size N-2? N-3? N-4? etc...
      That is exactly the problem of free block sudoku, adding that you can have free number of cells that have no local constraint. Basic sudoku rules applies local constraint for each cell in the grid. So if you stretch this rule, which number of cell do you accept without the local constraint? 1? 10? 50? etc..
      It's impossible to fix limits of what is accepted and what is not accepted. The only way is to be conscious of this problem, and to be smart enough a) from the player point of view : to accept little stretch in the basic rules of sudoku if the effect on the ranking is very small, b) from the organizer/author point of view : to limit the effect of these puzzles by containing their difficulty and the amount, and not stretching too much the rules of sudoku.
      Last year we had deficit and surplus sudoku in WSC. These were of size 7 and fairly easy so that no one would complaint. I think this is the right way to allow stretch rules of basic sudoku in a sudoku competition.

      About Straight sudoku: Yes, of course renban groups is fully accepted ... when there is an additional contraint. Here, it's not the case, it's replacing the local constraint of regions, and stretching the rule of deficit regions ! If you stretch rules of variants that are already stretches of basic sudoku rules and hybrid them, and so on, then after 3-4 step I fear that we'll have to accept fillomino as being an appropriate puzzle in sudoku competitions !

      About the ranking, I didn't want to imply that my bad round 5 was due to the round 4. My purpose was just to say that, like everybody I've good rounds and bad rounds, and that if I wasn't happy with round 4, it was not only because it affected my ranking.

  3. Thanks for the candor, Fred, and you clearly care a lot about the competition and where it goes from here. I appreciate you taking the time to write all this.

    I personally did not mind the non-sudoku stuff as much as you did, and I assure you I did as poorly as anybody on the Straight round. But speaking in the abstract -- Straight and Free Blocks are much further from a sudoku (in my educated opinion) than 1234. A sudoku to me is a puzzle where a collection of things (usually digits) are placed into a grid to satisfy row/column/region restrictions. 1234 had a set (not 123456789 but 1223334444) that met all these. Straight and Free Blocks did not. Additional rules (consecutive, non-consecutive, no 6's touch 6's, antiknight, et many many cetera) that do not alter the main set of rules are fine with me and do not change its category. Straight and Free Blocks do not have a full set in each region and so are further from the definition, at least to me. As are Deficit and Surplus, although at least they keep the spirit of the region restrictions by putting nearly all the cells in large regions.

    I realize sudoku is different things to different people, but to me it's always been about the three rules (thereby distinguishing them from conventional Latin squares) at minimum. In that regard I think 1234 belongs more than Straight and Free Blocks.


    1. Hi Jason,

      Thanks for the smart comment ! I also had the feeling when solving that 1234 was closer to sudoku than straight and free blocks. But it was perhaps just personal taste. I agree with your analyse though that I would replace "a collection of things" by "a collection of distinct things" to you basic rule. So 1223334444 doesn't fit anymore this category.

      And then, After just your comment and Prasanna's comment, we see that analyses are completely different, with from both sides some puzzles being ok, and other that aren't, but there are not the same for both of you.
      My preferred solution would to avoid these puzzles in sudoku competitions, but I fear it'll never be accepted by everyone. The other solution that could perhaps make consensus is to warn organizers that some players could not like to see puzzles that alter basic sudoku rules, and if they plan to propose some of them, they should use them with parsimony. I fear this rule would also be contested by some players...