Prechampionship/Travel to Prague:

While I spent a lot of time before the championships in Lucca last year practicing sudoku of all kinds, this year I spent very little time solving puzzles. I did take the time to do all of the national championships I could find (the Japanese championship a real gem, with the Russian and French and Dutch and Czech being of good quality too), but I was mainly finishing work in lab as well as finishing the writing of my "Battleship Sudoku" book with Sterling in the week before the championships. This, I suppose, at least kept me from getting overly stressed about the championships. I could say the following things before I arrived in Prague: 1.) I was about 15-20% faster than I was last year when I had a great chance of winning if I could have handled the final puzzle format better, and 2.) I expected the field to be much much stronger in Prague with more time this year to arrange for proper national championships and not just talking world puzzle championship competitors. Needless to say, I was hopeful I could erase the playoff demons from last year and win the event, but my surest hope was winning the team competition, particularly if Wei-Hwa and I could compete as well as we did last year.

My schedule the day I was leaving for Prague was a mess and I found myself doing laundry up to the last minute or so before I intended to be leaving for the airport. With a broken washer, I was left with a load still in water that took forever to dry and so I actually took a taxi to Logan as the T would have taken too long and even then I caught my flight closer than I wanted too. My connection - just an hour - in Charles de Gaulle was a mess with over a half hour spent just getting from the plane to the terminal with a long bus ride from a very distant part of the tarmac. I easily missed the connection, but fortunately got a seat on the next flight and after two hours reading the book Brainiac, I was on my way to my final destination, the Top Hotel in Prague.

Unlike other sites of World Puzzle/Sudoku Championships, the Top Hotel was rather busy, with various groups of Italian schoolchildren, a Ford convention, and other events, ongoing as we were there. As accurately described on another blog, the hotel was a lot like a Vegas hotel, just off the strip, with multiple restuarants and bars, various health facilities including the well-advertised Thai Massage, a casino, etc. Our room was a rather spacious suite in comparison to past rooms. When Wei-Hwa and I entered, we were surprised to see no beds. Of course, we were just in the entry room and after turning a corner would see the entirely separate bedroom with a second bathroom. It was definitely the largest set-up I've had while at a puzzling competition.

We took the afternoon after we arrived to quickly travel around downtown, stopped at some puzzle stores, crossed the Charles Bridge, and basically fought to fight off jet lag and stay awake. Wei-Hwa had brought Race for the Galaxy, the prototype game by Thomas Lehman that we've been playing for the last few championships and we had a few good hands of that that evening with Jonathan joining in to make it three players. Wei-Hwa had also brought a nasty new variation of sudoku called "Udoku" which had three difficulty levels. They basically include extra boxes in one of many geometries, so while row/column constraints still hold for 1 to N, you do not necessarily know what digits are in each box which alters the solving style. While I asked why it wasn't easy to hard, instead of the hard to extremely hard he called them, it quickly became clear why as I had to put away the pen and pull out the pencil when I went to solving them. These were tough nuts to crack, and the extremely hard examples still remain unsolved even though I've spent awhile looking at them. They are really excellent, if at times confusing and confounding, puzzles to solve and I hope to see some more in the future.


US Team walking around Prague castle


General Competition:
Thursday brought the team photos in the morning followed quickly by the instruction section. While a lot of our questions had already been addressed prior to the championship, we finally learned the distinction between an arithmetic sequence and a "special sequence" - the description "touching digits are consecutive" would have been just fine - and were ready, after lunch to get started. We also learned, following a question from our team captain, that when we finished rounds early, we should yell "Baxter" or "Czech Republic" (depending on whether it was an individual or team round). I regret not having the fun to take this instruction literally later on and I only ever said "finished", albeit in strong voice, so people would know I was done.



US Team photo (from left to right: Jason, Grayson, Jonathan, Nick, Jim, Thomas, and Wei-Hwa)

The format this year for the individual competition was 6 rounds over two days, each with 9-10 puzzles, and always 1 full hour to solve, except for the second round, a "sprint round" of sorts, that still had 9 puzzles but only 30 minutes to finish. As last year I most separated myself from the field during the long one hour round, I figured the format would play into my hands if I could pull off some solid solving sessions. Of course, one added wrinkle was that there would be "bonus" points for individuals who finished a round early, 3 points per minute, which was rather sizeable. At least as the competition started, I thought that I'd declare as soon as I had finished as I had not been making any errors in either blank cells or transpositions in my practice. I then proceeded to hand in papers, of course, with errors in blank cells and in transpositions. But let me not get ahead of myself.

Round 1 started well with a normal sudoku and after I made some progress on the isosudoku, I skipped ahead to the parquet and worked through it and the little killer and other puzzles in the round rather well. I came back to finish the isosudoku and had the round complete with three and a third minutes left. With so little time, I figured I should collect as much bonus as possible, so I immediately turned in my paper, expecting to be amongst the early leaders as only two others finished the round, including Michael Ley from Germany who I expected to be strong competition.

Round 2, the sprint, required a lot of very fast writing and I did my best. For the whole competition I sat next to David McNeill of the UK and I can say he is an awesome guy and a solid competitor. We'd chat before and after each round and while I think (at least entering the second day) I may have served to unnerve him a little at times, it was a comfort to decompress after each round with him. Anyway, Round 2 was the most solid performance David put together and while I expected to sprint rather fast, he finished as I was working on my last puzzle with some 9 minutes of bonus. I ended with 21:01 on the clock (which counted up, and not down), so I figured I had enough time to do a mini-check of sorts and still turn in for eight minutes of bonus. I went through each puzzle row by row (and only row by row) to make sure I hadn't repeated a number, then turned in at 21:54 to get full marks and 24 bonus points, or so I thought.

Round 3 was easily the hardest round of the first day, and while I was a little unnerved that the 10 point killer took me easily 10 minutes (over twice that of the 20 point killer and definitely more than every other puzzle in the round), I was glad to see that I was done with ten minutes on the clock. The round had some real gems, and some moments of "AAAAAAAAAAHHHH!!!!" discovery in them as well. For the latter, the Magic Square puzzle was a good example where I had started it, and gotten somewhat stuck, and thought there must be an easier way to do this. I reread the instructions and found out the puzzle had a diagonal constraint (meaning the diagonals contained just 1-9 once, just like the rows and columns and boxes) but this was not marked by a diagonal on the grid itself. Note to future typesetters - a diagonal constraint MUST be drawn on the grid. There is no excuse to not put it on the grid. I am glad I've solved enough puzzles that I know when something is too hard to be reasonable, and so here my experience led me to the instructions to resolve my problems. The Transfer puzzle in this round I found a very nice challenge, harder than the examples, but still tractable. During its solution though I violated my rule of never crossing out a clue number before I enter it in the grid and so as I got near the end of the puzzle, I needed a 1 to transfer from the row or column surrounding the lower-left box, and there was none to be seen. My heart raced, until I realized I'd prematurely greyed the necessary clue. Anyway, with 10 minutes left on the clock, I made a quick check that all puzzles were filled, checked that every cell in the cross (including the irrelevant ones) was filled in as this was a puzzle I expected I might have left an empty cell on, and then turned in for full marks and 27 bonus points.

The first day was over and then came the slow trickle of results. Certainly, having finished every round and with time to spare, I had nothing to complain about. But then I got back round 1 and instead of full marks and bonus, I had lost credit on the Parquet. The reason? A box in the topmost row was completely blank. Anyone, on inspection, could have gotten the right number in about 2 seconds. But I got no partial credit, just a slow start in third place. But that was ok. If that was the worst mistake I'd make that day, it was for my smallest bonus and on the lowest possible point value for a puzzle (10). But then round two came back and I'd made a simple transposal (28 in place of 82) in a column. So, while I'd checked row-by-row for that round, my error was in the columns and I hadn't caught it. Amazingly, it is in the exact same box as the error I would make, but be able to find and correct, in the playoffs, so maybe there is something about Box 4 that I should be extra careful with.



Two silly mistakes = One slow start.

I was suddenly sitting in seventh, over 45 points behind the leader, with two rounds in. I'd left some 53 points on the table and while I would have been just slightly the leader, it was clear the competition was very very close this year. I figured I'd be playing catch-up for the rest of the rounds. But the third round, as I described, was the hardest of the first day, and it sorted the scores the most. I was the only to finish clean, and with the 27 bonus points, I caught all the way back from over 40 down to take a slim lead into the second day. While people would congratulate me for having a good day, and being in first that evening, I'd have to deal with feeling personally that I'd really underperformed and let down my team by leaving a lot of points stupidly on the table with my careless errors in rounds 1 and 2.



Day two started with a hiccup as a printing error in some of the books for Part 4 forced that round to be delayed a minute into it until after Part 5. The quick jump to new puzzles I hadn't read the instructions thoroughly for (how I spend my last minutes before a round) was a bit unsettling but this whole day seemed tough and I had no time to let anxiety get to me. I got through the Creasing and the Ratio and the No Touch in standard order. I skipped the Many Times Times (somehow my multiplication is much weaker than my addition, although at my starting point that would seem quite subtle to most people I meet) and did the 12x12 more gladly even though it was a bit of a 6-7 minute slog. Finishing the Dual Doku, one of my least favorite puzzles, as it involved essentially copying one correct answer into a second grid (with some digit to digit transitions like 1 becomes 4) left just the Many Times Times with 39 minutes showing on the clock. I proceeded to attack this, albeit slowly, and finished with 51 minutes elapsed. Instead of turning in, though, I did a very thorough check this time, of every puzzle, and of every row and column. I have the nervous tic as people would learn later of running my finger left to right and then top to bottom as I do this along with my eyes and I'm sure the proctors, waiting to pick up my paper, anxiously watched my three minutes of checking. At 54:36 I turned in the round sure it was clean (and it would be).

Part 4 then reappeared, in a "special sequence", and this fell pretty well too. The Sudoku 2007 puzzle at the end - four linked puzzles - was not so hard to do and was enjoyable as it got very close to completion. I finished this round and then, again, spent 3-4 minutes checking but turned it in clean for a large bonus again. At some point after this round, I spent a fair bit of time just looking at the tablecloth in front of me after turning it in. I became fascinated in the almost sudoku-like grid of 7x7 squares (with some outer boarder almost making it 9x9) and thought it looked like it would make a nice puzzle. I then spent the time writing such a puzzle. This would end up being an exploration with Wei-Hwa later that evening but more on that to come including two puzzles to solve.


Isn't that pattern almost a sudoku?


Part 6 would be the first (and only) time in the championship I could not finish a puzzle. Maybe it was fatigue from earlier rounds. Certainly, during the championship, I was fighting some lung problems with a hacking cough entering the championship. Unlike in America, where in many parts of the country one goes outside to be able to smoke, in Europe I find that I have to go outside to be able to breathe. Right outside the testing room was the large lobby in which a majority of the air would be filled by smoke and on several occassions I'd break into a strong cough and almost throw up in reaction to the smoke. I'm normally not that sensitive to smoke when healthy, but with the cough I brought through customs into Prague, I was in real bad shape throughout the championship. During rounds, the adrenaline and focus would often keep me from coughing, but as soon as I finished I'd really need water or something as I couldn't readily breathe. Anyway, part 6 was again a hard one, and I proceeded during the round to hideously break three of the puzzles. When the round started, I began at the back to avoid the scary looking "Neighbouring Sudoku" with that superfluous u in the first word. I got the transparent sudoku (a new type, and very unknown how hard it would be), rather fast and then moved to the crossnumber. The crossnumber is a type where, as you are essentially first doing a "fill-in", there can be a lot of writing and then erasing. I was trying all the possible placements, and making lots of errors, and then I realized that instead of there being a camera person in my mind right over my shoulder, there actually was a camera person right over my shoulder. The German press, who were covering Michael Ley at the championships, had caught onto my success as well and wanted footage of me. For all of day two and day three, I'd be the subject of "shots" they wanted to take. At the start of part 4, for example, they wanted a shot of me walking to my desk at the start of the day. As someone walked into the shot on the first try, I was told to go back and do it again for a "second take". It would end up being a theme for the whole championship. They wanted to film me ending a round by saying "FINISHED!", and then they wanted to film that a second time as well. Anyway, the time I noticed my grid was being filmed was on the one puzzle I really hated to have it being filmed, when I'm erasing much more than writing, and I might have been slightly thrown but I could not break into this puzzle. I went back and did two others and then got to one of my stronger types, the irregular, and then proceeded to break that too by writing what I believe was a 9 twice in an odd shape when I shouldn't have. A full erasure and re-solve of that later, and a full erasure and re-solve of the crossnumber later, I had just that "Neighbouring" to do but time was running short. There were maybe ten minutes to go. It would be a sprint to finish. Others started to declare, but I wasn't making progress. I'd never done the example of this type - not enough time beforehand - stupid, Thomas, stupid! - so I didn't know the circles on the outsides worked in a fairly easier way. Still, when I was about 60 digits in and feeling on the right track, I discovered that this puzzle was again horribly broken and not fixable from where I was. I could not recover and finish it before time was up and so I did not finish the round.

Again, scores would slowly trickle in as the day wore on and my solid round 5 and 4 performances gave me an ~ 42 point lead that I hoped would hold up after my round 6 mistake. The team rounds began and they were basically challenges in team depth, with lots and lots of puzzles for the group to solve. Our team (which ended up being Jason Zuffranieri, who was a great wildcard for the team and ended up tenth, Wei-Hwa Huang, and Jonathan Rivet) assigned some puzzles to specific people. Wei-Hwa took most of the WPC-style ones in the middle but I was given increase distance to start on which I finished in about two minutes, and then Battleship Sudoku which, having written 200 of them, I figured was a specialty of mine. Indeed, it used a good constraint involving dead numbers in the grid (the givens are not allowed as ship segments, and many 9's were already given which was crucial for the cruiser placements) and in 4-5 minutes I had that done. I then went on to looking at the Roll Sudoku which was a very unique type and one whose difficulty we were unsure of. I discovered quickly how they'd written it to make it a reasonable solve, and found first divisions by box, and then divisions by three-box groups, to solve the placements of the rolls. I wrote the givens below and passed the rest off for someone else to do (Jonathan - who caught my one error in copying the gives thankfully). I jumped to the 12x12 as I figured I could do that faster than anyone else, and it was much harder than the individual one from part 4. I eventually did my "power through" technique, and used a forced chain-type method to get the whole puzzle to fall after reaching the sticking point. I did my cleanup on the double (another puzzle that desperately needed a dashed line indicating it had not one by two diagonal constraints in adjoining grids - the givens spelled SUDOKU which explained the two grids touching). I got the whole left on my own and Jason was a very valuable assist on the right side as we finished that up. We'd been rather efficient, and been checking answers throughout, so when the last digit in the double fell and was checked and I saw the clock at 51:59, I yelled finished so we'd get a full eight minutes of bonus. The organizers decided our solutions, not yet rebound in the test book, could not be accepted yet, so we got 52:30 on the clock when we finally finished the reassembly. We watched the Japanese to see if they would finish and indeed they did not so we'd caught up about 120 some points in the first of the team rounds. Of course, the Japanese had been scoring serious points in the individual rounds so we still had a lot to catch-up.



US during team round, team UK in background


Team round 2 began, as all the other rounds that day, with a lot of camera action around my table and at me, but I was determined to not be affected by it again and except for that moment in Part 6, it did not bother me in Prague. I started on Team Part 2 on the Big Big, which reminds me a lot of some of Ruud's clueless puzzles on sudocue.com, and indeed my solving the blue squares off-grid on graph paper was not so hard, I got three of the individual squares done too but with the 12x12 irregular untouched, I handed off the 6 remaining classics to another teammate as I worked through the 12x12 irregular in smooth fashion. I really enjoyed the puzzle. I then just picked up puzzles left that were in front of me which included the Diaeven and, I think, the Diagonal, although I can't remember exactly. I eventually found that the Mirror was untouched and although Jason and Jonathan had looked at it, neither had divined the rule at play. I saw the rule immediately, but dreaded the prospect of having to write the correct answer 4 times with all of the reflections. Thankfully, Jason tagged in and copied the last 2.5 grids after I'd solved one and begun copying the second, and I moved to other puzzles. We ended up with just the even left in the round, which was a nasty puzzle considering it had the lowest point value in the round. It ended up being a three-way solve with Jonathan and Wei-Hwa having the most success, and we finished the team round with less than a minute left. Team Japan finished with a little over a minute left, and got 5 points back on us, but I'm proud that we fully completed both of the team rounds and with cleaner performances from some of my teammates and myself in the individual rounds, we could have been closer or even passed the Japanese. Still, with the 2nd, 4th and 9th place finishers on their team, the Japanese were deserving team champions.

Playoffs:

The individual results were finally announced and I met my one goal of again qualifying first overall. I would have hated to again been sour grapes after the playoffs and finished second or worse, but in the long "regular season", I was again the best just as in Lucca. I'd be the odds-on favorite for the playoffs.

We met onstage to get the instructions for the final round and I had a good time joking with Vita, the organizer. He had made a point during the competition to say that this round or that round would be "for the kids." The Sprint round is when he started that, and then in Part 3 he said, "this is for the kids, but older" which was worth a good chuckle. Here, I asked if these puzzles would be "for the kids, and if so, how old" and after a laugh he said that five of the puzzles would be nice, for the kids, and one might be harder. We got instructions for 7 puzzles, 6 would be used, and instructions for how the system would work. The playoffs would take the top eight competitors (Ko Okamoto in 4th had to fly home early so Yuhei Kusui, also of Japan, in 9th filled in) and run them in two 4 person quarterfinals with 1,2,7,8 and 3,4,5,6 competing with each other. There would be two puzzles, each run simultaneously for the four competitors for ten minutes, with the most important criterion being number of correctly completed cells. Thankfully, there would be no subtracted points for incorrect numbers. If there was a tie in correct cells, then time matters. If there is a tie in that too, then finally the seed matters. In other words, my finishing first meant little except determining the identity of my first competitors. Given that two of those competitors would be Japanese, I was worried they would eliminate the one individual American left given how they had already beaten our team over the last two days. Fortunately, my other strong expected competitor, Michael Ley of Germany, would not be in my quarter.

As I mentioned, we got instructions for seven puzzle types (jigsaw sudoku, slant numbers, multiplication table, little killer, irregular, diagonal, even) but no knowledge of which would appear, or when, so every round that revelation would be a key moment. New this year as well, instead of placards, was solving on desks with cameras above. This is absolutely awesome and should be done for all future years here or even at the WPC for many reasons. First, by solving on desks, everyone is in their more natural element. Second, by projecting the grids at the same time, people could track where solvers were. I've been in the audience in a "placard" based WPC from the second row and it is not very possible to watch the solvers. I have eagle eye, and it is still difficult to read when they are on stage. Here, the solving pace was very evident - at least once everyone knew they were meant to use the red pens, as the projection was sharp and large. Third, is Youtube. Competitive puzzling will never seem like a sport until people see the best solving these puzzles against each other. I can't wait to get my hands on my solving video of these rounds and even to write up an "anatomy" of a solution in my mind but needless to say, a digital recording of a sudoku championship will be a good legacy of how strong this field truly was.


As I had a whole evening and night to kill before the playoffs would begin, we again played more Race. I spent awhile joking about fun ways to make use of the camera to relax myself. I began with the idea to have a "This Space for Rent" sign after I turned in my puzzle 1 which I'd place under the camera to be read. This would then be replaced by an ad for a company after puzzle two. I asked Wei-Hwa how much I could get if I put up Google after #2, but never got a concrete number. I at least had the Ruffles ad on the back of the Turkish magazine if nothing else. I also had my own battleship sudoku to pull out and solve - yes, a puzzle to solve between puzzles - if, at the very least, to get cheap notice of my excellent construction work for my forthcoming book. By the next morning, these camera tricks would turn into the idea of a "metapuzzle" in my solutions in which I'd try to hide a number in my puzzle. I hypothesized I could "dramatically" have an error in a puzzle until the very end and then fix it, with the error being a hidden number. I was aiming for 314 or 316, depending on whether I'd go with the gospel of pi or John, but at the very least, I was trying to think of anything but the playoffs and the pressure itself. I spent the night further developing the "tablecloth" puzzle with Wei-Hwa, as he modified his solver to allow us to write puzzles of this new type. We talked a lot about the possibility of starting a US venture similar to Nikoli to publish logic puzzles from a range of authors for what we felt was a very untapped community as most puzzle books are rather vanilla and simple, and seek to avoid anything too new and original. Wei-Hwa's Udoku experience certainly an example - although I'm thankful Sterling let me go forward with my Battleship Sudoku idea - we will see what other ideas are pushed into a book form. Anyway, I'll write more about this idea at a later time, but for now read/comment about it on onigame.livejournal.com. So, as I couldn't fall asleep, I thought about this venture a bit and in particular about what symbology would be best for the difficulties. Nikoli's smiles to frowns is pretty famous. I thought, to borrow the sloppy naming of some of Will Shortz' books, we'd start with a "sunny afternoon" and move onto a windy day, then a rainy day, and then thundersnow. Not ideal, but the off-puzzle thinking calmed me enough to sleep for several hours. Breakfast was spent with more "playful" playoff ideas, and solving the UP-themed version of the tablecloth sudoku we'd written the night before (Up being the name of a sudoku in the main event that apparently just referred to the givens looking like a bird. UP here meaning the givens spelled UP).






For these "tablecloth sudoku" - notice the pattern, minus the extra line for assymmetry compares nicely to the picture earlier - fill in the numbers so that in every row and every column, the numbers 1 through 9 appear just once. The large squares on the outside belong to 2 or 3 different rows and columns. Along the edges, there are just 4 numbers, but these four must be unique. The UP example is of medium difficulty, and of necessary training before you can do the second example. Wei-Hwa and I worked on a "Spinal Tap" version of the tablecloth, which would have a standard 9x9 sudoku in the middle and then a border as here, taking 11 digits - hence the Spinal Tap - but these are still unfinished and maybe not even possible to write/solve.



Finally playoff time arrived. Well, first there was the pomp and circumstance of the Czech President arriving and the finalists being announced, etc. I'm not sure what any of them said. I was in the back of the room pacing and singing "Creep" and not thinking about anything but solving the sudoku to come. Last year I let the magnitude of the coverage affect my solving both in the beginning and at the end of the long playoffs, and I knew, if I could maintain my focus, I should be almost untouchable. This would be my best chance (in a desk format) to win a world championship. The first puzzle was announced to be an even, and we were given the red markers to start on it. As I began, I had to learn to adjust the size of my notes to fit the grid and be readable. I also had to adjust to the slightly larger grid-size (although still smaller than the placards), but I went right to work.

I think there are many ways to approach a sudoku or a variant. Many choose to attack based on a given number. Look at the 1's, then see if you can place other 1's. Go to 2's, continue. Others choose to attack based on the state of completion of a row/box/column. If a row has three empty cells left, to take a 2,3,4 say, then you can look in the surrounding boxes and columns to find any naked or hidden singles. Finally, I think you can always attack it based on the constraint. This is not always very relevant, and on the even puzzle here it was not much more than identifying the 2/4/6/8 locations in the center square, but one thing I think is very different for me from other solvers is that I transition between all of these three methods rather fast. I've not gone from 1 to 9 in order in a real long time. Typically, if I place a digit, I search by digit and by row and by column and by box immediately so I am jumping from strategy to strategy all the time. This leads to the surprising result that I am sometimes much slower on a very easy puzzle where 1 to 9 searching works well than on a medium level puzzle, where flexibility is more key. An awesome Nikoli.com sudoku the week before the championship could be solved completely in singles going in a 1-9 order. Looking at 1's gave you all nine 1's immediately. Then looking at 2's gave you all nine 2's immediately. Needless to say, I did not catch this gimmick until later on and was slow relative to most of my nikoli.com times on that puzzle.

One other comment on my solving style, which is now well shown on those solving videos if they ever come up again, I only ever write notes in one situation - if a number can go in one of just two cells in a box, and only in one of just two cells in a box, then I write that number in small size in the corners of those boxes. If those cells adjoin, then I save fractions of a second by only writing the number once, over the edge connecting those two cells. The reason for this is to store what I call cascading information in the puzzle. When I can finally place a digit, I immediately search the row and column it is in to find that same number in any note. If I see it, then I can immediately write the same number in another cell. This continues to cascade and I can write a lot of numbers in very fast order when a puzzle starts to fall. Hidden pairs fall out rather fast with these notes as well, and with enough practice you can get used to seeing hidden triples (which are usually at the site of 2 touching notes with a third digit that can only be in the three remaining cells). The result of this note system is what the Times reporter described as an almost automated or "robotic" solving style where my eyes are already scanning the puzzle for other digits before my pen reaches the paper for the digit I found. The reason this doesn't mess me up is my writing instruction from my brain says "TWO!" at the same time my eyes are looking for twos, and so the coherence of TWO! and two works. It is only when I'm cascading without notes that I'll do transposals from going too fast.

So, the even sudoku has started and I've written some notes and placed a couple digits, when I make my first real big breakthrough with about 10 digits in 4 seconds. I hear oohs from the crowd. I realize they are reacting to me. From that point onward in the playoffs I knew to use the audience to calm me. I could sense when I was in the lead, and I tended to not hear reactions at times I wasn't writing, except for what proved to be when competitors made errors. As the audience was reacting a lot to my writing, particularly when it was really fast, in this round, I knew I was probably in the early lead. After a second breakthrough, resolved by finding a naked single in box 1 (again - jumping between the solving styles lets me find things faster than most), another cascade happened which finally gave the even digits in the middle. Then, I was off notes, but still just propagating digits. I finished, and the audience didn't know what to expect. Most thought I'd yell "Finished!". Instead, an odd spectacle developed. I described above my nervous tic of running my finger over the rows as I'm checking them, then over the columns, then around the boxes. Well I did this twice. As it started, a rumble grew in the audience as they realized the crazy rate at which this checking was going on. My eyes just scanning wouldn't have been good drama, but this machine-like scan certainly was. After two full checks, I finally turned in the even with just four minutes and seven seconds elapsed [edit: I took that time from a press story and, as in the comments below, it feels wrong. I bet it was 4:07 left in the round, or about five minutes fifty-three seconds in - with the clock counting up the time may have been misread by someone expecting to do subtraction]. I pull out a battleship sudoku to solve (I needed to finish test-solving my book after all, and didn't want to get nervous during the wait given my early lead) and Yuhei in 9th finished about a minute later. I was hoping that no one else would finish as that would give me the advantage of no time pressure in the 2nd round. As the final minute counted up (yes the clock still was counting up), neither of the other competitors could finish. I should be safe provided I didn't screw up.

Puzzle 2 was a little killer, which always seemed to break in in exactly the same way (corners, then side boxes, then middle box), and which always had diagonal constraints although they were never marked on the grid. This lead, back in February, to me doing a lot of work with a Czech to English translation engine to make sure there was a good reason I couldn't solve the Czech example in the main round. The Little Killer was the final puzzle in the Czech playoffs and I needed to make sure I could solve it. Needless to say, another "hidden" diagonal constraint is at play in these puzzles. The playoff puzzle had much the same character as the others, and I made reasonable progress, although apparently I flipped a 2 and a 5 towards the end. While I had joked about hiding the sequence 314, and this error would indeed be how I hid the number "3" in the meta-playoff puzzle, this error was not intentional. If I'd gone immediately to the crazy check method by row and column, I would have found the error in five seconds. However, I wanted to check all the outside constraints anyway, so I started there. It was seven sums in I found the first problem - a sum needed to be 15 but was 12. Uhoh! I continued around the grid and eventually found three other errors that intersected in the two reversed cells and made the simple correction. Somehow, while my sudoku skills are not affected by the pressure, my ability to add in my head is impaired a bit. Still, I was clean on both puzzles as was Yuhei who actually won the two rounds on total time and we were both through. Hideaki Jo, who gave me a strong run during the qualification and was the 2nd seed, was among the two eliminated contestants.

The second quarterfinal (with the four other players having been sequestered) then began and while I caught maybe the first minute on the screens, I decided my time was best spent getting some air outside. I got intercepted by a reporter covering the championships for India who proceeded to take tons of pictures at various angles of me. I also talked with Wei-Hwa to Ferhat of the Turkish team about getting our Tablecloth sudoku (particularly in our "Spinal Tap" form) into Akil Oyunlari and he took a picture of Wei-Hwa and me that will work when we finish our submissions. I came back just in time to hear the second quarterfinal results, with the two higher seeds Michael and Nikola both being eliminated. Suddenly, the playoff favorites except for me were gone. It was 1-5-6-8, the highest cumulative sum I could have faced. I couldn't relax, as it was clear upsets might be the way of the day.

I calmed myself before the semifinal by paying an homage to Monk by tapping the light above the desk as Chaloub's character does all the time. I'd do this before each puzzle from then on. The semifinal began with an irregular sudoku. I always find this a type my "many angles" approach can solve really well. Building in LoL methods is important, but also is not forgetting that regular sudoku rules apply at times too. The ideal, of course, is the toroidal type I dominate, but irregulars are strong for me. I begin attacking this at the clear LoL's in the corners, and then focused on the fact, from the geometry, they might be hiding tricky naked/hidden singles on the sides. I find the one in the bottom right in row 8 that gets me started. The cross shape in the middle of the grid was then the focus of my attack with LoL again and then hidden singles around the grid let it slowly fall. As I was about 25 seconds from finishing, I heard Yuhei say he was finished. I worked through the end, only single checked this time, and declared finished as well. While it was ~6-7 minutes total time, the other two semifinalists could not complete the puzzle in time so again I could relax and just need to finish the second puzzle of the round to advance on points.

The second semifinal puzzle was revealed as the "Multiplication Table". This was the one of the seven types I hadn't brainstormed the night before, and given how I'd erred on the earlier mathematical sudoku, I was a bit worried. The geometry they used actually gave 3 uniquely solved boxes based on multiplication but I was slow to choose brute force attack by just multiplication in those regions. I broke into the lower left completely, then the upper right, and while I was stubborn about the middle, I could not do enough "sudoku steps" around it to help out. I got stuck on the riddle of how to have a product that was 2X. 28 = 4x7 struck my mind, but 7 was already in use and not possible. 21 never struck my mind, but it would fail for the same reason. I got stuck at 24=4x6 for a real real long time. This can't work as you can't have two 4's in one box. So I went back and made sure the 2 had to be the tens digit in the product and it clearly did. I then went back through the list, again eliminating 28 and again forgetting 21, and settled back on 24. Suddenly 3x8 came to mind. Not sure why it took so long, but with that "are you smarter than a fifth grader?" level of insight into products, the puzzle fell immediately. As the semifinals ended, I was the only one clean throughout and guaranteed to move on. David, who had been a great competitor and a great friend, was last having been defeated by the irregular, and Peter and Yuhei were tied in points at 156 but Yuhei, off the strong irregular round, advanced based on time.

Unlike in Lucca, we advanced effeciently to the finals (the President was in the audience after all, so no time to waste). The first puzzle would be a Diagonal, a type that can range over so wide a difficulty that I did not know if this would be the tough one. I've posted this puzzle below as it really determined the championship and is a much better "final" puzzle than the one that actually ended the event. I worked hard, but perhaps slowly, on this one, as it was rather tough to both place digits and to use the diagonal constraints, at least at the start. I kept forgetting to use the diagonals as much as I could, and every time I could make a semi-placement (one of two cells on the diagonal), I'd pump my fist in excitement. After a long time (maybe a minute) or just staring, and contemplating a guess and check route, I finally saw a hidden single 3 that defined the lower right corner and that solved the upper left then the upper right then the lower left then the lower middle and so on and I finished, exhausted, with full checking at 8:51. I knew I was doing well. The audience had only made strong reactions when I'd been writing, at least towards the end. I found out Yuhei had a lot to go and when time expired, I had a 25-30 point lead (it turned out to be 27). At no other point in the championship did emotion take over as strongly as in that two minutes between puzzles. I could sense I had won, if I could just not horribly botch up the next puzzle.


The first puzzle of the Finals, and the decisive one.

Vita put the Jigsaw Roundoku instructions on the projector, and I got even more worked up as this would be a
very very easy type, the simplest of irregulars to be honest. I had no need to hurry. Apparently, I was agonizingly slow as my teammates would tell me afterwards, to at least to do a simple thing like "look at the 4's". I was trying so hard to power other digits that I did not do simple scanning for about a minute. Of course, Yuhei got all those digits in the first instant. Still, I just needed to finish and four minutes later I had. A full minute of checking - for the first time without the frenetic finger tracing - and the World Championship was mine. Hard fought, with incredible competition this year, but my second in Lucca now an afterthought with a world title and wikipedia entry for myself now.



Aftermath (immediate):

What followed were a lot of interviews, an awards ceremony, and a fair amount of posing for pictures. A press conference was held and I did my best to speak up for variants and other logic puzzles and human creativity in puzzles - something the wealth of vanilla sudoku books out there completely lacks. There is a magic recipe of artistry and originality and interesting logic that makes for a good puzzle. The puzzles at the WSC fit this recipe a large majority of the time. Randomly sprayed digits (what I call Jackson Pollock sudoku) on a grid that happens to form a valid sudoku as most computer-generated puzzles appear on the other hand do not. You'll understand more of what I mean about this if you check the puzzles I've written on this blog, or buy my Battleship Sudoku book where I write a "sailboat" puzzle or a "baseball" puzzle or make other creative use of 3 allowed shapes to do more than just form a valid grid.

As champion, I tried my best to satisfy everyone's desires for photos and questions, even odd questions from a reporter for a site looking for sex advice from a sudoku player who'd already spent the morning questioning Wei-Hwa. Given that one probably belongs in the bedroom and one probably does not, I found this interview a bit trying, but I carried myself with as much dignity as possible, even as I felt this characteristic slowly being sapped from me as the day continued. The most taxing experience was certainly tied to the German press who had been filming Michael Ley and myself throughout the championship. I'd already done a couple takes of walking to my desk, and saying finished, and shaking Michael's hand to accept his congratulations. Well, the reporter (who apparently is putting together a 12 minute piece), felt the clinching shot would be a "confrontation approach" of the two of us on the Charles Bridge. It is worth commenting that unlike during the filming of, say, Mission Impossible, the bridge was not closed to foot traffic. It had hundreds of people passing each minute as Michael and I were asked to stand 10 paces apart, then walk towards a mark and stop and stare at each others eyes for several seconds. Needless to say, several takes were possibly lost as we cracked each other up at the ridiculous task of staring at each other from inches apart as if we were rivals when in reality we are friends. Still, the "boxing poster" profile shot was what was wanted. After a couple takes from the side, they needed shots from ground level of our feet approaching, shots from behind us as we approached each other, shots that circled us from hip height, shots that circled us from neck height, shots of us shaking hands and parting, ..., ..., .... (never before has ellipses of ellipses felt right) At some moments, passersby on the bridge caught on that there was filming and they tried some sabotage efforts by walking behind me like in a train of people, which obviously ruined the shot and made the process take even longer.

Throughout the weekend we'd been discussing Wei-Hwa's most recent puzzle gadget on Google which involved memory lists and naming all the items in a list in a specified amount of time like US States or stages of grief (Kubler-Ross model) or minerals on the Moh's Hardness scale or noble gases or spaces on the clue board or American Idol winners. Anyway, one of the ones I remembered was the standard character attributes in D&D, and I this filming continued I suggested adding a secondary attribute called "dignity". Between takes, I'd stop back with my team and report on how my dignity was falling. It eventually reached negative numbers and I think at -7 dignity I decided the filming had to be done as no special item could get me back to zero. I did a walk-through with a Portuguese reporter for the closing lines of his own report, which was just three takes (but thankfully just one and not twenty different shots). For a moment, I thought of the ironic headlines if I jumped off the Charles Bridge ("Sudoku Craze Dies, Along With Champion") but after only 75 minutes the filming exercise was finally done and I could be a person and not a prop again. Boy, that felt good.

We walked around Prague to more puzzle and toy shops. At one of these shops, we saw a hilarious "dice game" that I'll include a photo of that had at the very least two serious problems with it. See if you can identify them.




When we finally returned to the hotel, we realized the beauty pageant the championship really was (the sign read MS SUDOKU 2007 afterall!) and headed to the awards dinner. I got a bunch of gifts from different teams as well as the team and individual medals, another SudokuBo peg-solving set that matches the one I got last year, and a set of Tescoma cocktail making tools (the BARBAR 6 pc set that won a 2006 excellent product of the year award) of which I could not identify many of the functions, teetotaler that I am.



That evening was spent with a lot more photo posing with different teams, lots of talks with different captains and competitors, and at the end, a long time chatting with the British and Irish teams. I was at my lowest point of needing to really crash on a bed when Simon from the UK, a mutual livejournal friend for several months that I finally got to meet in person in Prague, pulled the dreaded Big Brain Academy I expected to see. Of course, I was not anywhere near as sharp as I wanted to be as he tested my mind - a paltry 1478 g for a first try - but I felt I represented my exhausted state quite well with this addictive little game. David McNeill and I later raced on a Turkish puzzle with a time standard - he in 1'25", me in 1'11" although while carrying on a conversation, so maybe not as fast as could be. Both times crush my "Guinness World Record Time", but that was set on a much harder set of sudoku written by the late Michael Mepham. My best paper time ever, for those who care, is still 1'02" on a 33 given sudoku from Puzzler/Nikoli.

Then it was finally time to pack, sleep, and get back home. Amazingly, for the third straight transatlantic flight on an Air France plane, the audio/video at my seat was malfunctioning. The trip back from Lucca and the trip over from Boston this time had the individual tvs in the seats in front of me not working either time. Fortunately, with an empty middle coming from Boston I could lean and get movies from the center seat. Going back, though, with just cabin video from Paris, my audio plug had something stuffed into one of the holes preventing the jack from even entering the slot. Needless to say, I will consider other carriers in the future, as I really need something low intensity for the return flight after a world puzzle championship event.


Aftermath (long term):
While the sudoku championship has its share of gravitas, and I will be defending it in India, the main thing to come out of the championship for me is this idea strongly shared with Wei-Hwa that we might have an opportunity to create a puzzling community like Nikoli - for now named NICK-L-E in rebus form after our team captain and the two letters said aloud. Our goals are simply to offer an outlet for puzzle designers to get an audience for their creations, to gather constructive comments, and to help develop some logic puzzles into solid new types. Similarly, we hope to provide an outlet for solvers to see really interesting hand-crafted logic puzzles that are missing in most books and magazines in this country. We've thought of a web-comic like release, with a monthly pdf for purchase form if it becomes popular, but with a world puzzle champion and a world sudoku champion behind such a venture (and other designers sounding interested), this venture may very well get off the ground when I get out to the Bay Area.

Regarding Sudoku, who can say how the fad will last in the next years. As I commented in interviews, I think the first satisfaction of puzzling comes from the sense of completion of a new puzzle. Sudoku is a simple puzzle type that takes only a couple tries to start really improving at and can be addictive quickly as the sense of completion is immediate as is the rush of placing the last couple digits as one gets near a solution [much different than a crossword, where I'm often staring at a blank square or two and going A to Z in my head until I, without confidence, make a letter choice]. However, vanilla sudoku do get boring. The number of mediocre computer generators leads to a lot of mediocre and in the end uninteresting books that do nothing to stand out from the billions of other possible computer-generated puzzles. Variants (and ultimately other logic puzzles) are the true outlet, in my honest opinion, for the continued growth of the interest in logic puzzles that classical sudoku has begun. While the classic form will last in some places, like the daily newspaper, it will not be the sole puzzle being consumed by solvers in bookstores. My first book, "Battleship Sudoku", will certainly be a test of the market to see if quality puzzles, with both artistic fluorishes and compellingly hidden logical steps, can be a sale-able good. "World Sudoku Champion" on the cover might sell some more books, as might the Hasbro tie-in we expect to use, but ultimately at some point "Sudoku" must be used as more than a means for publishers to try to print money. At some point, the puzzles must be of quality. Crosswords at the best levels aren't written by computers and the most interesting and lasting sudoku will not be either. The puzzling community Wei-Hwa and I hope to build will try to address this and provide an outlet for quality, unique, hand-crafted puzzles of many different types.


The sudoku speed-solving world is certainly catching up. I did not walk easily through the early rounds as I did in Lucca. That I was the only repeat playoff performer speaks both to my talent but also to the amount of undiscovered talent that was found between the first and second WSC. I don't know if I have another 15-20% improvement to make between years and I don't know how much others will improve. Still, I will be in Gao and hope I can defend my title next spring. Our up and comers like Jason will hopefully help fill out the US team with Wei-Hwa returning to form (fingers crossed) and maybe we can overtake Japan for the team title. The World Puzzle/Sudoku Championships are a whole lot of fun and I'd like to thank all the competitors, the organizers, and all the puzzle constructors. This was a tremendous championship. Looking at the bottle of Sudoku '07 Chardonnay I was given at the start of the championships, I can say the WSC in Prague proved to be a really good vintage. Until Rio in the fall, cheers. - Thomas Snyder, World Sudoku Champion