You are viewing motris

 
 
21 October 2007 @ 01:35 am
Sudoku update, part deux  
WPC report still to come, but I figured I'd clear out a piece of sudoku-related business first. Something about a championship with a big prize.



This weekend was the first US Sudoku Championship, held in Philadelphia by the Inquirer with Will Shortz as host, Nick Baxter in charge of organizing the puzzles as for the US Puzzle Championship, and some very familiar faces amongst the competitors that might do well. What was a big unknown was who would show up - would the groundswell of about 900 competitors bring new strong solvers into the public view? Would other US Sudoku team members past and present compete? Would the intermediate division be a better choice for all those who read Wednesday's inquirer piece and called the organizers asking to move down from advanced having read some of my times?

The competition would have three qualifying sprints - each set consisting of an easy, medium, and hard sudoku with the advanced solvers needing to get through all 3 within thirty minutes. The first to finish a round would advance to the playoffs with the exception that a person could not qualify multiple times so after the first win of a round, the results won't matter going forwards.

Having laid out my folder as a writing surface as Stamford vets know to do, I began the first round with the hopes of finishing real fast and getting a sense of the field. The first two puzzles went splendidly although as I forgot to put on my watch before leaving the hotel in the morning, I'll say maybe with 3-3.5 minutes. Then, I hit a hard puzzle. A real hard one. There are not many sudoku that make me sweat anymore, at least not any that should appear in a competition, and this seemed a real interesting curveball. After making about 7-8 placements, you can get very little communication in the grid. You aren't far enough in for many good places to guess. After what seemed about 2 minutes of not writing anything, even going through the 1-9 series to make sure I wasn't missing anything, I busted into the "power through a puzzle" mode of using good intuition and fast skills to get on the right path. I sensed something about a particular digit (which was an either/or choice), and indeed if I'd glanced at the or for a couple seconds, I would have seen it failed immediately. Having started writing again, I struggled my way through the rest of it, unhappy to be in a division with hard puzzles, but still finished the whole round in 9:59. Now, I sat in the very front row of the room so the spotters would quickly follow my times and I did see a nice "1" placard after my turn-in so I knew I had finished first but I hadn't checked the round. What I still failed to realize, given my struggles on the puzzle, was how bad it was for everyone else. I was waiting (well, solving the Shortz "Simple Sudoku" book we had been given, mostly) for the next declarer which wasn't until 8 minutes later. Crazy!

Somewhere below the toroidal sudoku at WSC1 and the spiderwebs (WPC Eger) and the feynman hexagon puzzle (SPIES 2006 Mystery Hunt) may fall that hard "Q" puzzle in this round as a solve I did in "miraculous" time with good intuition and a flexible approach and obviously a bit of luck, which does favor the prepared mind. As both Will during the event and Nick after the event would comment, it was clearly the hardest puzzle there. While it was even solvable if you removed the 2 and 6 in the upper right corner, throwing this toughy out early would serve to reset the field if indeed I won that round as the other two rounds would be completed by more people. Only 6 other solvers out of 150 in the advanced division finished that round in 30 minutes.

Rounds 2 and 3 went much easier for me although I'd already qualified - the hard puzzles were nowhere near the "Q" level of difficulty (on a scale of A to F), and I took the time to check all puzzles on each of them, fixing a blank square in one of the rounds. I believe my times were 8:45 and 7:40 on the clock for turn-in for both, and again that was first for both rounds, but only by about 3 minutes for each of those two. So a great start - not only was I solving impeccably well and across a range of difficulties, but I was feeling on top of my game, wearing my competition red.

I should comment that my US Puzzle teammate onigame (Wei-Hwa Huang) constructed these puzzles. I did not, during the event, even notice that they were all titled, which was a nice touch, or that some of the givens had Philadelphia-related shapes or other themes. I wondered why some digits in one of the puzzles were bolder than others, but wasn't bothering, in my race to just focus on the digits, to figure out if it was theme-related. Anyway, they were good puzzles over a range of difficulties and Wei-Hwa got the notice (and potential hate mail for the first hard puzzle) that he deserves.

There was a bonus round, a single puzzle, that would determine age group winners of a hundred dollars before the playoffs. I did well, finishing in 1:24, but given the chaos of the room, at least for 20 year olds, it was not clear if I was first or not. One proctor misidentified my group, or didn't alert the right spotter, so I feel they were overcareful in correcting my turn-in time, but I'm fairly certain Roger Barkan, at 1:15, nipped me by 9 seconds. As a gesture of sportsmanship, when the 20-year old prize was announced as mine, I turned it over to the second place person, who happened to be Roger, who likely had won the round anyway. The organizers need to fix their prize structure a bit, and certainly not run a speed round for all the money where most top solvers will be seconds apart but maybe many desks apart.

Anyway, the playoff contenders in the three divisions were announced and we all went on stage. They ran the beginner division first and I was sitting to the side with the other higher level finalists and got to meet my two competitors in advanced, Tammy (who had introduced herself to me the night before at the welcome/museum event as a reader of my blog and, not unexpected for a talented puzzle solver, a google employee and mit grad complete with typical brass rat) and Sarah who had traveled the shortest distance of the three of us (tammy had also flown in from california) but had survived the tough qualifying hurdles as well. We got out first glimpse of the large whiteboards - far too large to solve comfortably, with a repeated problem from the Sudoku Smackdown of not having thick enough tape to demark the bolded boxes so you had extra problems getting lost with boundaries when looking at the middle.

After Lori DesRuisseaux won the beginner round, we collectively watched on our side as Ron Osher attempted the intermediate championship puzzle. Ron was on our WSC1 team, and did well in the sprint rounds but lost ground on the "marathons" in Lucca and so some might view him as a squatter by being in the intermediate division. Having seen him pull out his "bag of candidates" method while solving on stage, I think the choice of intermediate was right for where he is at with good speed on "reasonable" puzzles. It also was financially a brilliant move as he won the round to get the second largest prize. However, the solvers on stage all seemed thrown by the big board. There were three to four work-ins I kept trying to urge Ron to see (of course he was wearing a headset and could hear nothing) but the three solvers took much longer than audience members on paper would have for that puzzle. During the solving, I was chatting with Tammy and Sarah and trying my best to calm everyone's nerves, joking a bit, etc. I know its a hard thing to be solving in a round against someone with my reputation and speed. Knowing that the horrible prize structure means finishing second equals a grab bag can't make it any easier. Still, I volunteered to take the center board to have the more focused point of attention for the challenge to come. I put on my headset and started listening to unintelligent babelling. Maybe next year, to increase the difficulty, my headset will spout out numbers "one ... four ... nine ... eight ... eight ... EIGHT!!!" to confuse me. As white noise, it was just fine.

After a while of standing in my spot but not being able to hear anything, so not knowing when the round would actually start, the proctor finally removed the cover page from the whiteboard. I had thought to be cute I'd have a marker in both hands to be ready to solve in a new and interesting way (writing with both hands - maybe next year), but the puzzle was a doozy and so I stuck to normal (read: sane) approaches. People will often comment, when they get the chance to watch me solve, that I often am missing fairly obvious placements that a 1 to 9 search might find. I will admit this is true. However, I'm often trying to attack a puzzle from its weak spots, which are never the singles you get at the start but harder things that arise later. I'm also storing a lot of information both on the puzzle and in my head even if I'm not instantly going through the digits in order to find simple placements. This puzzle definitely felt the same way to me. I was making some progress, then stalling, then making more progress. With the large headsets over my ears, I could actually hear my heartbeat as I feel they were a bit tight to my head. While a couple years ago this might have unsettled me, I've been in enough playoff puzzle rounds now - albeit not with 10k on the line - that I am better at focusing out those things. Still, the large size of the board, the lack of good bolding on boxes, and the fact my normal note style is easily interrupted by having to write over a piece of tape that displaces my marker was not easy. That, and the puzzle was also nasty. I hate to do it, but you need all your tools in a competition. My ultimate and important crack came from using uniqueness. I described this at the championship for the five people in the audience who might have understood me, but after a while of searching for something, I got a magical lucky seven that let the rest of the puzzle fall out. I checked the grid my patented way, and declared in 7:07 to win the US championship. Tammy finished in a super time of 10 minutes to be second. It is a shame that Tammy, who I feel would have creamed the intermediate solvers when she could do a much harder puzzle in only 2 more minutes of time to Ron's 7:57 immediately before, left with a gift bag. Let's hope that in future years the prizes are set up so competing with me is not a bad financial choice. The ACPT does it right by having third in the top division always pay out more than first place in a lower division. The best solvers should not be avoiding "Dr. Sudoku" just to have a better shot at money.

Throughout the day, and certainly after the championship, I was talking to a lot of reporters. There are some good articles such as from the inquirer itself - which also has photos and video (although the reporter decided filming additional things during the finals, but not the finals themselves, was the best use of his cameraman's time). Fortunately he captured the end with me declaring - but only then to immediately pan to the audience clapping instead of seeing one of those rare times I have a true and genuine smile.

However, with any wealth of interviews, and with the pace of my speech, my quotes often get horribly cut so this is that time in my blog when my loyal readers will recognize me clarifying things that got into press. The main culprit this time is the Reuters piece which claims I think puzzles are not meaningful. I will clarify, although it should need no such clarification, that I love puzzles, I love writing puzzles, I enjoy competition, but I someday hope to be more known for the products of my work in science than for the speed with which I can solve a sudoku. I hope, by making a career of science, and not of just setting more sudoku speed-solving records, to benefit others. This point got shortened by the Reuters reporter into me saying puzzles are not "meaningful" and that's a horrible one word way to tear at my heart. Puzzles are very meaningful, and are excellent tools to exercise the brain, teach rules of logic and deduction, explore mathematics, interest kids in learning, and so on. I just don't plan to make a career from solving sudoku faster than everyone else when I can use my talents to do other things as well.

So anyway, I've gone on long enough. I can finally head for my early morning flight back to SF. I promise to have some more puzzles up here soon with some cool designs, and I still promise many more exciting things to come.
 
 
( 21 comments — Leave a comment )
EpicureanAngelepicureanangel on October 21st, 2007 04:17 pm (UTC)
Congratulations on your win It was well deserved!

I was very glad to have gotten the chance to see you in action, and while at first I thought had I known I was up against you I might have dropped down to intermediate, I was glad I gave it my best shot.

Great writeup, by the way.

- Tammy
motrismotris on October 21st, 2007 09:43 pm (UTC)
Thanks, and congratulations on your strong showing as well. It was great to meet you on Friday before the championship and a nice surprise to see you on stage as well. I hope to see you again in the future.
Maelstrommlstrm on October 21st, 2007 07:18 pm (UTC)
Hey, congratulations!
jdyer on October 22nd, 2007 01:33 pm (UTC)
Congrats!

Do you know if the puzzles from the championship are going to be made available?

(Also, I really wouldn't have minded if you posted the worldpuzzle update yourself. Someone who was there would have more to say about it than me, who can only read bad Reuters quotes from afar.)
motrismotris on October 22nd, 2007 02:41 pm (UTC)
I tend to forget to post there, but in general it would just be a link to my report. The Inquirer has promised to publish the sudoku used in the championship in the paper, so I'm hoping they end up online for people to see and compare times with.
(Anonymous) on October 22nd, 2007 01:57 pm (UTC)
Well done, Snyder. It was a real pleasure to watch you apply your system. I was glad to hear you discuss the exclusionary 'trouble' with that 9th column 7, wish we could have gotten more of that (gah, acoustics). I took the final puzzle home and completed it (on my whiteboard) in 10:47 - but this was with the double advantage of having observed the three of you solve it AND being in the comfort of my own home. I thought it was a nasty puzzle, but not as bad as that one from the first round.

As for the prize, I wonder how many serious puzzlers would prefer a decent shot at $5000 over a chance, however remote, of beating you. I know I wouldn't, and I'll be there again next year with the same intent :)

Nathan
motrismotris on October 22nd, 2007 02:44 pm (UTC)
The first round puzzle was both nasty and had a very very cool and unusual logical problem in that early sticking point. I don't know if it has a set name, but it reminded me of the visualization needed for me to crack into the toroidal at WSC1. Since it was the same digit that was critical in both puzzles, maybe there's something about me seeing something with that number. (I'm trying to be spoiler-free, but you probably know the relevant digit).
(Anonymous) on October 22nd, 2007 06:15 pm (UTC)
Sudoku and Contributions to science
From my point of view, determining fast methods of solving sudoku
could lead to scientific contributions in mathematics and computer
science that would rival current contributions being made today in
your chosen field. Don't dismiss it quickly, especially if it gives
you a fresh perspective in your postdoctoral work.

Sudoku is a good candidate for something called a "constraint
satisfaction problem". People are studying these problems to find
efficient ways to solve them. I recently found a group that is
attempting to tackle the theory of these problems using methods
from my area of graduate study (check out www.aimath.org workshops
for some gory detail if this interests you). If some of your methods
could be refined and formalized, and placed in this context, that
might generate a significant advance in this research. I suspect
that it would at least help in developing heuristics applicable to
a much wider class of problems being studied in computational
complexity.

I join others in congratulating you. Also, good luck in your postdoc.

Gerhard Paseman
motrismotris on October 23rd, 2007 04:38 pm (UTC)
Re: Sudoku and Contributions to science
Thanks for this very interesting perspective. I've thought about it for the last couple days and I agree with you that there "may be a there there" when it comes to using human strategies to puzzles to address computational approaches to constraint satisfaction problems. I still wonder if things I sometimes call "intuition" - sensing where a puzzle is likely to be cracked - can be formalized, but I will continue to follow the mathematical development around sudoku and see if I can contribute.
(Anonymous) on October 26th, 2007 07:13 pm (UTC)
Re: Sudoku and Contributions to science
I found your blog via Wei-Hwa's - I was a contemporary of his at Caltech where he introduced me to many kinds of puzzles and games. Anyhow, I wanted to lend a supportive voice to your desire to pursue a career in scientific research. I, too, hope to make a lasting contribution in the biological sciences, though I have other (arguably better developed) skills. I think that the information-rich nature of contemporary biology provides countless opportunities to apply puzzle-solving logic. Though of course those problems won't present themselves in a 9x9 grid.

It will be interesting to see how your career develops and I wish you success.

All the best,

Jon
knightwizard on October 22nd, 2007 11:20 pm (UTC)
Hello!
It was a pleasure to have met you and watch you in action! I agree with what you said about the prize structure...actually, I feel Tammy should get an invitation on the U.S. team, she was just that good. Congrats again, and see you next year :-)
motrismotris on October 22nd, 2007 11:24 pm (UTC)
Re: Hello!
I agree with you that Tammy (or Sarah) should be kept in mind when the formal - likely online - qualifying process for the rest of the US team is decided.

Unfortunately, the general setup of the Philly championship did not give enough of an opportunity to rate all the best people and Nick Baxter will want to choose the best US team that can solve all the kinds of sudoku that may be at the WSC as it certainly is not a classics only event. In past years it was a bunch of people sending in their times on classics and other kinds of sudoku and I expect the same process this year once we know how many can go to India.
(Anonymous) on October 22nd, 2007 11:22 pm (UTC)
Congratulations on another win!

The Times could learn a thing or two about hosting a championship over here in the UK! I don't think they're even sure about sending anyone out to Goa yet - and I'm sure I heard that there was no UK presence in Rio for the WPC. The prize money doesn't quite compare either hehe.

I'd be interested in seeing this beast from round 1, especially re our discussion about hard championship puzzles and their levels of "guessability". It is funny that sometimes there's almost a 6th sense when it comes to these puzzles...i guess that comes from doing far too many of them!

Tom C
motrismotris on October 22nd, 2007 11:42 pm (UTC)
I'm hoping they'll be web accessible at some point. The Inquirer promised to reprint them all but they may be print-version only. If you search through enough photos of me you can reconstruct the "Hard" championship puzzle so that's likely fair for me to repost at some time.

The first round "Q" puzzle was the first classic puzzle that fit my criteria (described in Simon's post I think after the Times final) of solvers getting stuck early so guessing is not as easy or anywhere near as tempting. While I was a bit lucky to solve it as fast as I did, it has a very cool break-in point. I was saddened to not see any Brits in Rio and had read that the Times champion may no longer travel to the WSC; a real shame if that happens.
(Anonymous) on October 25th, 2007 08:28 am (UTC)
yeah a similar situation with the UK's when I wanted to show friends etc the puzzles, a few photos helped piece together a couple of them together. I'll certainly give recreating that one a go when I get a spare mo...

I'm glad that such a puzzle does exist, if someone has created it then the mechanics of such a puzzle are at least known to someone so there's a good chance more similar puzzles are there to be created! And I'd probably suggest that once you are at a certain level, you almost "make your own luck", the whole sudoku 6th sense thing.

Well I obviously hope they do pack me off to Goa, but even if they do, chances are we won't have the awesome sense of team spirit that we and the Irish had in Prague. They (I'm not sure it's necessarily The Times who organises championship teams) would be crazy if they didn't send David McNeil either way...and I think he'd have been a bit of a dark horse to well in Rio as well had he actually gone!
motrismotris on October 25th, 2007 02:19 pm (UTC)
I've actually now seen a second puzzle, again from Wei-Hwa, that meets the criteria. It was actually considered too hard to use in the championship. Scanraid completely fails to not use ridiculous steps so it is legitimately tough, but I found a way to crack into it that uses some of that intuition. If I had your email, I might accidentally make the mistake of sending you the relevant puzzle(s).
motrismotris on October 26th, 2007 02:09 am (UTC)
Looks like the puzzles you want are now somewhat available. The final puzzle from the championships is here and my time on white board is well known if you want to compare. The "Q" puzzle is almost available as Wei-Hwa posted a harder version on his blog here. While the full challenge there is fun, the competition puzzle adds in a 62 in R1C89.
(Anonymous) on October 27th, 2007 02:23 pm (UTC)
Thanks for posting those up...I've not yet had a crack at your championship final, but the Q puzzle (the harder one on Wei-Hwa's page) was very good! The 2 jumps of logic that needed to be made i cant claim to have come across very often - and thus it took me a good while (20+ish minutes) to finish it - but they were very nice touches. And especially good for championships like you say because the rest of the puzzle was straightforward to whizz right through.

I've never really thought about creating puzzles before, or the actual mechanics that goes into making a difficult puzzle so I have no idea as to whether something like those could be done...however it is refreshing to see that they do exist, and at a stretch a few one offs could be made for a championship!

If you wanted my email anyway its t.collyer (AT) warwick.ac.uk

Tom C
(Anonymous) on October 23rd, 2007 03:46 pm (UTC)
WPC BRAZIL PUZZLES
where can i download the wpc brazil 07 puzzles?
motrismotris on October 23rd, 2007 04:34 pm (UTC)
Re: WPC BRAZIL PUZZLES
The rights to the WPC puzzles (just as with the WSC puzzles) are given to the relevant organizing publishers in each country to use in their magazines/books and will not be available for free download. If you were a competitor, then you should get copies of the puzzles eventually but I can say that even I am still without clean copies of the WPC puzzles.
Teesside Snog Monster: swingsjiggery_pokery on November 2nd, 2007 10:53 pm (UTC)
As well as being an excellent solver and clearly a highly accomplished scientist, you're a fine writer! Thanks for all the detail; very enjoyable reading.
( 21 comments — Leave a comment )