World Championship

Everything about Sudoku that doesn't fit in one of the other sections

World Championship

Postby Bigtone53 » Mon Mar 13, 2006 10:47 am

Code: Select all
. . . | 7 5 . | . . .
. 3 . | . 4 8 | . 2 .
1 . . | . . . | . . 6
----------------------
. 4 . | . . . | . . 8
7 9 . | . . . | . 3 1
2 . . | . . . | . 7 .
----------------------
5 . . | . . . | . . 7
. 8 . | 3 2 . | . 4 .
. . . | . 6 9 | . . .


According to The Times today, this was the final puzzle in the World Championship at Lucca. The winner apparently thought that it was a shame that the puzzle had to involve guesswork.

I don't think that I guessed in solving this, although I was nowhere near the winner's 15 minutes.
Bigtone53
 
Posts: 413
Joined: 19 September 2005

Postby vidarino » Mon Mar 13, 2006 10:52 am

Quite correct, this puzzle doesn't require guesswork at all.

At the "stop point", a bit of colouring and an X-Wing.
vidarino
 
Posts: 295
Joined: 02 January 2006

Postby Hud » Mon Mar 13, 2006 5:31 pm

A couple x-wings and locked candidates. I'd compare it to the average Pappocom V Hard. 17 minutes.
Hud
 
Posts: 570
Joined: 29 October 2005

Postby motris » Mon Mar 13, 2006 5:43 pm

Having finally gotten back to the states and read the Times story, I would like to clarify that neither my US teammate Wei-Hwa Huang or myself thought it needed guessing. We thought that, to win the competition by being the first to solve it, it would need guessing.

I agree that the puzzle is logically solvable; however, in a time competition, trying to do just logical solving is not the best way to win and is why I ended up not winning after leading throughout the competition (closely trailed by Wei-Hwa who also impressed this past weekend). I spent too much time on this classic sudoku trying to do it by logic, even after seeing that it would involve the X-wing forced on the grid when you get to the point where I first got hung up.

The reason using logic I believe fails on a classic sudoku puzzle is that competitive sudoku solvers get fast enough at propagating numbers through a grid (we can all do mediums in 3-4 minutes I'd say, sometimes much faster) that once you get to a set point in a filled-in grid, spending as many as 2-15 minutes to see the hidden logic is likely to lose to spending the 1-2 minutes it takes to propagate both branches of a quickly deterministic logic forward to find a contradiction in one of the two branches.

As a counterpoint to the "final" puzzle of the playoff round, I would like to accent the second to last puzzle of the playoffs that will not get much play in the press. It was a toroidal but one of incredible difficulty to get started. In my mind, it fits everyone's adjectives of what is needed in a championship puzzle. It was very very difficult. "Super ultra mega wicked fiendish" or whatever some might call it. The key part was it was a puzzle that was not at all guessable and it basically needed you to use a new logic rule to solve. In the end, I think the fact Sudoku often forces you to use new logic to solve a puzzle is what keeps it interesting.

Anyway, after 15 minutes on this toroidal puzzle, 3 of the 4 competitors, including the world champion, could only fill 1 square! That's correct, 1 square! I solved the puzzle in 14 minutes. While I did not win the overall championship as judged just by the last puzzle (not even playoff round results propagated forward, it was always just a survivor-style last place person is eliminated), my personal opinion is that by doing the toroidal no one else could and finishing consistently well in all the rounds leading up to the final puzzle, I put my stake in the sand in being a Sudoku master. At some point, I will try to post this toroidal, if no one else has, because I find it infinitely more satisfying both as a high-challenge sudoku and as a championship style puzzle, than the one that actually decided the whole competition. I think the Sudoku community, even the purists who prefer square grids, will find this the most interesting puzzle to come from the world championship playoffs.

- Thomas Snyder
motris
 
Posts: 71
Joined: 13 March 2006

Postby CathyW » Tue Mar 14, 2006 10:03 am

Interesting to read Tom's views on this. It took me 22 minutes on paper without guessing but I can understand that in a competition where speed is of the essence that T&E could be quicker. Different tactics required for speed competitions!

I'd be interested to see some of the other puzzles from the competition - haven't tackled a toroidal before.
CathyW
 
Posts: 316
Joined: 20 June 2005

Postby motris » Tue Mar 14, 2006 8:24 pm

The toroidal puzzle mentioned above is now posted in the variants section of this forum here.
motris
 
Posts: 71
Joined: 13 March 2006

Postby tso » Wed Mar 22, 2006 11:15 pm

I would disqualify all those who were unable to solve the torridal and give the title to motris. This is the difficulty of *all* competition. If you bowl a strike -- but your toe is 1/1000 of an inch over the line, you get nothing. It's arbitrary. The rule could just as easily be that you are penalized a single pin.

I think this thread brings up an important point about solving for time. In *my* opinion, this is the worst way way to judge the best solver. Imagine if ALL solvers ALWAYS solved ALL Sudokus for the best time above all else. Would any complex tactics every be discovered? We all agree that the best way to pick the best chess players is NOT 5 minute speed chess. I suppose there must be some sort of time limits -- but there should be no bonus for finishing more quickly. If 10 competitors all solve all the puzzles within the time given -- that's a tie. This is the traditional way to judge tests of mental ability -- two students with identical scores on the Scholastic Aptitude Test are NOT differenciated by who finished the test in less time. If one must break the tie, a very complex variant is presented, one which few will be able to solve -- maybe NONE will solve -- within the time alloted. Perhaps if no one solves it, whoever is closest to the solution wins.

Like it or not, if time IS an issue, an important skill is the judgement to know *when* (and where) to make your first conjecture.

If a puzzle is simple, one who guesses too soon (or at all) may lose to the skilled hard-and-fast logician.

If a puzzle is complex, especially if it is a variant the competitors are unfamiliar with, one who guesses too late OR too soon may be lost. It's likely that what might be considered the most efficient sequence of solving steps would be different if the goal is NOT to solve the puzzle, but to move the puzzle along to a point where the solver has the greatest chance of finishing it off quickly with a single well placed guess.

It would seem to me that in order to reduce to the minimum the chance factor in a solving competition in which solving time is considered -- the puzzles should come from one of two groups -- those that are very simple in which guessing will be a handicap -- and those that are very complex and/or variations in which most or all known humanly implemental tactics fall short and creativity and ingenuity will rule.
tso
 
Posts: 798
Joined: 22 June 2005

Postby motris » Thu Mar 23, 2006 5:58 am

While I would not use as strong a language as tso - I wouldn't "disqualify" the others for not getting the toroidal as it was very hard and probably unfair for a 15 minute time limit - I certainly agree with his point that just "time" on a very hard sudoku is not a good criterion for a competition's final puzzle.

As an organizer at the WSC, I would have kept the 7-puzzle playoff round mostly as it was (while there were many variants of varying difficulties, all the puzzles until the end greatly encouraged logic and discouraged guessing) but done something like use a cumulative score to rate performance during the playoff. The person who was eliminated in the toroidal round, Tetsuya Nishio, had actually had the best playoffs of anyone competing if you just count completed puzzles. He actually completed 4 of the 6 puzzles he saw, earning full marks more times than anyone else. If you used cumulative score from the playoffs, he would have survived the toroidal round and another competitor who had not done as well throughout the playoff would have been eliminated - in fact, the whole elimination order would have been much different.

A playoff system that does not reward past success may work well in the realm of sport (who doesn't like an upset by their underrated football team over a more favored eleven), but seems unsettling in a mental competition.

I also doubt that a standard sudoku can be made very hard without it becoming a better or at least quite reasonable strategy to "guess" than to use logical steps. At the very least, the puzzle must be strongly playtested to ensure there are no good guesspoints but I have not encountered any sudoku writer who really bothers to consider this possibility.

The impossible to implement way to judge the top solver is to make the puzzle solution be presented more as a logical proof than as just a filled grid. If you had to write out/describe your steps and your reasonings behind each placement, it would be harder to win by guessing as no judge would accept "well, it had to be a 1 or a 6, and I tried the 6, and that worked" as a logical step. As this seems too unwieldy, the only way I can see to make a playoff work well is for it to be over several puzzles, and particularly over puzzles where either guessing is not possible, or where using logic is not a penalty based on a time constraint.

My teammate Wei-Hwa's point, which seems in line with tso's, is that something like a couple (say three) moderate difficulty puzzles is a better playoff than a super hard sudoku that allows guessing. A great sudoku solver can run through a medium sudoku faster than someone who guesses. As you make the sudoku harder and harder, there more often than not arise "guessing" routes that reward luck and not skill. Imagine any great logical sudoku solver against 5-10 other people who all may resort to guessing after making the "easy" initial placements. Do you think, with enough other people, guessing won't win on just a single hard sudoku? Would you rate the best poker player based on how a single hand played out where the luck of cards could be far more important than any skill at reading people?

My ideal playoff in this kind of format would be a sudoku relay. Imagine 5-7 medium sudoku all set up so that solving the first one reveals the given numbers for the second and so on forward. The audience would still have the fun of seeing who is getting farther fastest, but the winner would be the person who could solve lots of puzzles well - maybe not the fastest on an individual puzzle, but the fastest over a fairly large number. Overall skill, not luck, would become more favored. Consistently being the best solver of each puzzle would be the only way to win.


Anyway, I've rambled on enough. These are points to think about and discuss as competitive Sudoku events continue. On an aside, I finished writing up my WSC trip report and for those who have enjoyed my writing so far on the event, you can find it on my blog here. While not all of it is relevant to sudoku and some is intended more for my friends, it does gives a sense of my experiences as a competitor in this event and may prove interesting for some of you to read and hopefully interest you in trying to qualify for your country's WSC team next year.

Thomas Snyder
motris
 
Posts: 71
Joined: 13 March 2006

Postby Moschopulus » Thu Mar 23, 2006 7:22 pm

motris wrote:At the very least, the puzzle must be strongly playtested to ensure there are no good guesspoints but I have not encountered any sudoku writer who really bothers to consider this possibility.

One poster to this forum (gsf) considers how many "backdoors" a puzzle has.

Roughly speaking, a "backdoor" is a cell where, if you guess the right digit, the rest of the puzzle becomes easy. Unless there is another backdoor.

As far as I remember, most puzzles have 0 or 1 backdoors. Very few have 2 backdoors, and none are known with 3.

Of course, I haven't defined what I mean by easy. The number of backdoors will depend on this definition.
Moschopulus
 
Posts: 256
Joined: 16 July 2005

Postby tso » Thu Mar 23, 2006 10:50 pm

tso wrote:I would disqualify all those who were unable to solve the torridal


I retract this comment. What I should have said was simply solvers should be penalized for each puzzle failed and/or rewarded for each one solved. Two competitors who get the correct solutions to the same puzzles should have the same score regardless of time taken to solve.
tso
 
Posts: 798
Joined: 22 June 2005

Postby PaulIQ164 » Fri Mar 24, 2006 12:32 am

Yeah, it makes sense to me that ability to solve the puzzle should be the first criterion, and time only to break ties. But I guess it doesn't make for a very tense final round if one person has already won by virtue of having solved more puzzles correctly already.
PaulIQ164
 
Posts: 533
Joined: 16 July 2005

Postby tso » Fri Mar 24, 2006 5:10 pm

Are the puzzles from the WSC available somewhere now, or must we wait a two or three years for Random House to publish them? (That's the typical lag time from WPC competition until an annoyingly incomplete subset of the puzzles are put out by Random House in it series of books that are annoyingly impossible to tell apart by there covers or titles.)
tso
 
Posts: 798
Joined: 22 June 2005

Postby motris » Mon Mar 27, 2006 6:11 pm

As per WPF rules, the puzzle publishing rights go to individual member countries as a way of offsetting the costs of these events so the puzzles cannot be shared online.

I know that Will Shortz's sudoku book publisher, St. Martin's Press, will eventually be putting out a book of puzzles from the WSC or on other nation's qualifying tests but the focus will be on mainly the "vanilla" or classic sudoku as the US audience does not like variations. While there were not as many classic puzzles at the WSC as I expected, the ones that were there except for the final puzzle that you've already seen in the Times/in this forum were pretty standard and the interesting ones IMHO were mainly the variations I had not seen before. A couple of the puzzles - at the very least the mechanical sudoku that was on 9 3x3 transparency slides that had to be reassembled into a valid sudoku grid with rotations and flipping allowed as the numbers were in a fairly symmetrically shaped font - cannot transition into a printed format.

I doubt, though, that the US Sudoku book around the WSC will present the puzzles in their round format with points/times to let people try their hand at the puzzles as seen in the competition. I've never liked, or bothered to buy, the recent WPC books for this reason. The feeling from the editors/publisher is that the general audience that would buy the book does not want or need this information. Not being in that general audience, and being someone who wanted to compete at the WPC (and finally did this last fall in Eger, Hungary), I always found this disappointing.
Given the publishing cycle, I doubt this book will be put together and published before the end of the year if even by then.

However, I expect that the Italian organizer nonzero spl may put together its own collection of these puzzles in their sudokumix magazine in a format more to your liking. Importing might be hard and I certainly don't read Italian so I cannot tell how you can get the magazine outside of italy - its not like some foreign sites, like Nikoli, that have some infrastructure for foreign customers - but there is a chance that a foreign puzzle magazine will have the puzzles both sooner and in a better format for sudoku enthusiasts to do.

If I hear of any further news on where and when the puzzles are appearing, I will post it on this forum for this community to see.

Thomas Snyder
motris
 
Posts: 71
Joined: 13 March 2006

Postby tso » Mon Mar 27, 2006 7:01 pm

motris wrote:... but the focus will be on mainly the "vanilla" or classic sudoku as the US audience does not like variations.


If this is true [that the focus will be on mainly classic sudoku], it would be incredibly shortsighted and dense. The US audience did not consume logic puzzles of *any* type to any measurable degree until Sudoku took off in the UK. Puzzle magazines in the US cater to word puzzle solvers with only a obligitory nod to the logic puzzle solvers. If one want's logic puzzles, one must import magazines from the UK, Japan, Turkey, Germany, Korea, the Netherlands, etc. Now of course, Sudoku is ubiquitious.

Further -- the "US audience" does not purchase past collections of World Puzzle Competition puzzles. The "logic puzzle solving niche market" does. In order to cater to this small but rabid niche -- they've got to include everything. By putting out a collection of just the classic sudoku -- they will cater to NEITHER market! The small logic puzzle niche won't bother with it and the (currently, but not for long) large Sudoku consuming market won't care a bit that the puzzles were in a competition.


motris wrote:I doubt, though, that the US Sudoku book around the WSC will present the puzzles in their round format with points/times to let people try their hand at the puzzles as seen in the competition. ... The feeling from the editors/publisher is that the general audience that would buy the book does not want or need this information.


Another foolish mistake on their part. They cannot hope to market to the "general market". If the chose instead to be specific, complete -- they wouldn't lose a single sale. I think the idea is that there's no reason to cater to the diehards -- the core audience -- instead assuming that they'll have no choice but to purchase them no matter what. Obvously, they are wrong.

The story of Hagen Daz ice cream (which may very well be exagerated and mythologized to some degree) -- they started selling three flavors -- Carob, Honey Vanilla and Rum Raison -- catering to a specific, small niche. The three flavors were very popular, the company grew very quickly and added more and more flavors to cater to the general market. Though the sales of the first three flavors never declined, the popularity of the new flavors far outstripped them. Caring not at all for these loyal customers who made the success of Hagen Daz possible, all three flavors were discontinued.
tso
 
Posts: 798
Joined: 22 June 2005

Postby motris » Mon Mar 27, 2006 9:35 pm

I completely agree with you tso. Unfortunately, the publishing world does not seem to care that there is an opportunity now to put out a special book that can offer more than just regular sudoku and bring attention and growth to logic puzzles in America. The people who buy a book of World Sudoku Championship puzzles are not going to be average consumers, they will be people already into sudoku that want a new challenge and want to see how they would stack up to the best. Publishers constantly underestimate their consumers in this way by dumbing down these books. I find the stickers placed on the recent WPC books saying "featuring sudoku" to be both crass and insulting when the book has just 1-2 sudoku (and even then its an irregular shape most likely) and dozens of other interesting puzzle types. However, this is the state of puzzling, at least in America, at the moment.

Good puzzle books, like Dave Tuller's and Michael Rios' Mensa Math and Logic Puzzles books, are few and far between, and publishers willing to print a book that doesn't look like four hundred others on the shelf is rare. The "me too" effort in publishing is to print sudoku books like they are money which - on the whole - ends up with lots of similar books that are of average quality at best, books I immediately pass up each and every time I see them no matter how prominently they are displayed in book store after book store. Its the same reason that 40 or more Dan Brown-clones already have book after book hitting the shelves, not because content is important, but because pulp fiction sells. Somehow, having the WSC book be a pulp sudoku book is going to be a huge disappointment to me (if this is how it does turn out). A book that features championship sudoku puzzles seems to me to be one of the few opportunities to put something special on the bookshelves, a book that has cachet just because it was from the world championships, and a book that can stand apart from the dozens of others books by offering different but beautiful and fun sudoku with little twists in them.

For me, I want a book I'll be getting to give as a gift not just because it will mention my 2nd place finish but because I think it has the best puzzles it could have in it and will offer a unique and rewarding experience for people who already do sudoku but pick up this book for a special challenge. I'd like to be able to see hard puzzles like the playoff toroidal in there and not just by adding 15 more digits into the puzzle to make it easier. Yes, there could be a page where there are hints to get started, but the unique puzzles at the WSC should not be dumbed down or removed for the casual audience. I think this underestimation of the audience is why Dell magazine had sudoku (a.k.a. "Number Place") for over 20 years without it being a memorable part of their magazine at all. With so many givens, I know I did the puzzles when I was young but never found them either hard or interesting - at least their cross sums were good enough to cultivate my interest. Only after England via Japan showed that challenging sudoku with much less than half of the 81 squares prefilled in would sell did Dell get the message and fix what was their puzzle.

There is something beautiful about a finely crafted puzzle. Something about how it stretches your mind, forcing you to be creative, and when you solve it, you feel rewarded for having gone through the effort of having solved it. In my mind, no publisher has ever understood this in America. It is why the top US solvers at the WPC and WSC all have gotten good at puzzles by doing foreign magazines and foreign books. I wish the sudoku craze could tap into the unrealized potential of the logic puzzle market in America. So far, I see little signs of it doing much. Instead, I expect to see more and more ludicrous books like "Pat Sajak's Super Sudoku featuring Code Numbers (TM)" and still, as before, nothing from an America publisher I really feel like buying.
motris
 
Posts: 71
Joined: 13 March 2006

Next

Return to General

cron