Simple version:

WSC4 happened. Flying home in 14 hours (after a train to Bratislava and then Vienna).

Longer version:

So the Championship is over. The "Championship" might continue tonight, but I'm out on a broken non-tiebreaker, at least that is what it seems to look like. Regardless of what happens, I am retired from WSCs for the foreseeable future, as the clear best sudoku solver on the planet.

Scores after individual parts (possibly readjusted with regrades but this is what I have a photo of):

1 - Thomas Snyder (USA) - 265

2 - Vincent Bertrand (BEL) - 236 * Vincent got the 3:06 time in the world record "puzzle" for 12 points

3 - Jan Novotny (CZE) - 232

4 - Jan Mrozowski (POL) - 231

5 - Gaurav Korde (IND) - 219

6 - Hideaki Jo (JAP) - 219

7 - Jakub Ondrousek (CZE) - 218

8 - Peter Hudak (SVK) - 212

9 - Mehmet Murat Sevim (TUR) - 209

10 - Pavel Jaselsky (SVK) - 209

26 others in the playoffs too, down to a score of 149. I think the signal to noise is pretty good here to make a judgment call.

Anyway, this morning we got a team round with 10 puzzles we solved one at a time. 8 were fairly straightforward, particularly with the 3 of us looking at them (WH and I writing most of the numbers, Jason directing others) - pretty cool I bet to watch as we really broke through on these. 1 was a new design that was both a very tweaky/guessy "puzzle", and lacked a good logical way in. We powered 3-4 possibilities to force contradictions before getting on a "at least two numbers in pen" kind of track. After that was done, we were left with a domino "sudoku" and 1h15m of the 2h15m of the round. My guess is the construction of this "puzzle" was done this way: lay out 21 dominoes (0-6) on a grid like a sudoku. Fill in the untouched squares with the remaining 0's-6's and nine 7's and 8's to make an answer. Then give solvers most of the 8's and 7's, and one 0-6. Give teams 3 grids because there might be multiple solutions, and let them at it. So, its not a puzzle - at least it does not have a set of logical steps to follow to reach a solution so much as you stumble into one if you can tweak an almost solution well enough. I got one, but two singleton cells were touching so it was invalid according to an additional constraint. There are many "puzzles" that are easy to construct and not meant to be solved. This one, with the 3 of us working on it for 1h15m, was never solved. It raised my anger a fair bit, but after other things at this tournament, was not a horrible surprise.

Then was a chat with the Times (of London) reporter - not features editor Michael Harvey this time - in which I handed off my "not sudoku" handout and also showed him my general style of writing artistic sudoku as a means of giving back to the community of solvers.

Semifinals of the "championship" then started. Being as far in front of 2nd as 2nd was in front of 11th earned me the first desk in a 4x9 rectangle. It actually was a broken table, with a leg that really really wobbled, and I eventually convinced the organizers to swap the table for me to something I could actually write on. 10 minutes of clarifying rules followed - we'd already done this 2 nights ago in a session that went past midnight - but 36 people will be anxious when most are newcomers.

Being 1st also meant I randomly drew the 8th, ..., competitors and that set included Peter Hudak (3rd in WSC 2), Yuhei Kusei (2nd in WSC 2 and 3), Ko Okamoto (who was 4th in qualifying in Prague but had to fly back to Japan and missed playoffs), and Michael Ley of Germany who finished in the playoffs each of the last 2 years. A great group of 9. One would for sure advance, four open spots also available.

There were 4 puzzles and 45 minutes and we all started at the same time. Yes, even with the enlarged playoffs, there was no reward for solving puzzles well in the first day. The first 3 puzzles were tough but interesting. The third - a sort of "anti-sum of 10" puzzle - was maybe my favorite of the championship - and I got through the logically solvable ones in 30 minutes or so. I spent 15 minutes trying to solve another version of that tweaky/guessy new "sudoku" and could not manage a solution. While some did, not many did, and there were not 8 complete papers. In fact, there were no complete papers. Still about 10 solvers had said finished as if they were done. Unknown to me - another "advantange" in that front left seat as I could not see the proctors walking around the room - the proctors were signaling to the audience with their fingers how many puzzles were done by a solver as they walked by. Contestants in the back could have seen this, but anyway the audience knew I was one away for a long time. If this affected their game planning - when I was still stuck without 4 solved - I won't ever know. All I know is during the round I, and the audience, heard 8-10 "finished" and everyone was a liar or a quitter (well, someone "gaming" the playoff system in a smart way, but that's what I view either action as).

Now, lets talk a little bit about playoff structure. Ties were broken in this order: first, total points, then time to turn in, then existing score order. So I had the third one. Except, if you turned in an incomplete set of solutions to "post" a time by giving up on the last "sudoku", then you would beat me if I didn't finish. I refuse to not try to solve a puzzle, even a "sudoku" which I've learned to "solve" fast well since championships call on the best sudoku solver to solve "sudoku" too. So when the scores came out, and 8 people were tied at my score of 31 including the person who got credit for "winning" my group, the fact I had not handed in my paper a minute or more early left me (and others) on the outside looking in. Another broken structure in another broken round in a horribly broken "championship".

For scoring specifics, there were three 36's (with 1 unsolved puzzle), one 32 (with 1 unsolved puzzle), and nine 31's including me that tied without the one without a solving path. The points were hardly representative of the difficulties anyway, so you can consider people at 3/4 or people at 36/32/31 for our purposes. So, 13 people needing to be cut down to 8 including 9 ties. Well, for not stopping, solvers like me, Hideako Jo, and Jan Novotny were eliminated by not quitting soon enough. Also Cen Chen of China and Jocelyn Bogreau of France. I would be unhappy both eliminating or being eliminated on a "who quit sooner" criterion. Sudoku are simple to partially grade and a tiebreaker based on how far solvers were into the fourth sudoku in terms of correct digits would be incredibly more fair than what happened. Of course, if the sudoku has no sure digits as the 4th one did, then even that tiebreaker is more chance than anything, but at least it gets solvers to keep solving puzzles because some part of an answer is worth something.

So, the terminology we will now use on this blog is that there are sudoku and "sudoku". The first can have you tell me how it is meant to be solved and the latter not. The Guinness record puzzle is certainly a "sudoku". "Sudoku" are possibly interesting computational exercises, but not for human solvers. "Sudoku" tend to occur at a "Championship" like this, where there might be a Champion - who solves the most puzzles consistently well and gets a good number of the "puzzles" too - but there is also a "Champion" - who "solves" enough things to survive an individual round with 35 others and then again a round with 7 others. That guy may be a good solver, but he/she also got lucky at the right times several times.

As I said, I'm retired from WSCs, the question for me is if there will be another WSC without the C in quotation marks. I'm not necessarily retired from those. And while I know the US organizers will make a good one without me, I feel there are enough innovations to return sudoku-solving to solving and not bifurcation (even in classic puzzles only) that I really need to be a man behind the curtain next year.

WSC4 happened. Flying home in 14 hours (after a train to Bratislava and then Vienna).

Longer version:

So the Championship is over. The "Championship" might continue tonight, but I'm out on a broken non-tiebreaker, at least that is what it seems to look like. Regardless of what happens, I am retired from WSCs for the foreseeable future, as the clear best sudoku solver on the planet.

Scores after individual parts (possibly readjusted with regrades but this is what I have a photo of):

1 - Thomas Snyder (USA) - 265

2 - Vincent Bertrand (BEL) - 236 * Vincent got the 3:06 time in the world record "puzzle" for 12 points

3 - Jan Novotny (CZE) - 232

4 - Jan Mrozowski (POL) - 231

5 - Gaurav Korde (IND) - 219

6 - Hideaki Jo (JAP) - 219

7 - Jakub Ondrousek (CZE) - 218

8 - Peter Hudak (SVK) - 212

9 - Mehmet Murat Sevim (TUR) - 209

10 - Pavel Jaselsky (SVK) - 209

26 others in the playoffs too, down to a score of 149. I think the signal to noise is pretty good here to make a judgment call.

Anyway, this morning we got a team round with 10 puzzles we solved one at a time. 8 were fairly straightforward, particularly with the 3 of us looking at them (WH and I writing most of the numbers, Jason directing others) - pretty cool I bet to watch as we really broke through on these. 1 was a new design that was both a very tweaky/guessy "puzzle", and lacked a good logical way in. We powered 3-4 possibilities to force contradictions before getting on a "at least two numbers in pen" kind of track. After that was done, we were left with a domino "sudoku" and 1h15m of the 2h15m of the round. My guess is the construction of this "puzzle" was done this way: lay out 21 dominoes (0-6) on a grid like a sudoku. Fill in the untouched squares with the remaining 0's-6's and nine 7's and 8's to make an answer. Then give solvers most of the 8's and 7's, and one 0-6. Give teams 3 grids because there might be multiple solutions, and let them at it. So, its not a puzzle - at least it does not have a set of logical steps to follow to reach a solution so much as you stumble into one if you can tweak an almost solution well enough. I got one, but two singleton cells were touching so it was invalid according to an additional constraint. There are many "puzzles" that are easy to construct and not meant to be solved. This one, with the 3 of us working on it for 1h15m, was never solved. It raised my anger a fair bit, but after other things at this tournament, was not a horrible surprise.

Then was a chat with the Times (of London) reporter - not features editor Michael Harvey this time - in which I handed off my "not sudoku" handout and also showed him my general style of writing artistic sudoku as a means of giving back to the community of solvers.

Semifinals of the "championship" then started. Being as far in front of 2nd as 2nd was in front of 11th earned me the first desk in a 4x9 rectangle. It actually was a broken table, with a leg that really really wobbled, and I eventually convinced the organizers to swap the table for me to something I could actually write on. 10 minutes of clarifying rules followed - we'd already done this 2 nights ago in a session that went past midnight - but 36 people will be anxious when most are newcomers.

Being 1st also meant I randomly drew the 8th, ..., competitors and that set included Peter Hudak (3rd in WSC 2), Yuhei Kusei (2nd in WSC 2 and 3), Ko Okamoto (who was 4th in qualifying in Prague but had to fly back to Japan and missed playoffs), and Michael Ley of Germany who finished in the playoffs each of the last 2 years. A great group of 9. One would for sure advance, four open spots also available.

There were 4 puzzles and 45 minutes and we all started at the same time. Yes, even with the enlarged playoffs, there was no reward for solving puzzles well in the first day. The first 3 puzzles were tough but interesting. The third - a sort of "anti-sum of 10" puzzle - was maybe my favorite of the championship - and I got through the logically solvable ones in 30 minutes or so. I spent 15 minutes trying to solve another version of that tweaky/guessy new "sudoku" and could not manage a solution. While some did, not many did, and there were not 8 complete papers. In fact, there were no complete papers. Still about 10 solvers had said finished as if they were done. Unknown to me - another "advantange" in that front left seat as I could not see the proctors walking around the room - the proctors were signaling to the audience with their fingers how many puzzles were done by a solver as they walked by. Contestants in the back could have seen this, but anyway the audience knew I was one away for a long time. If this affected their game planning - when I was still stuck without 4 solved - I won't ever know. All I know is during the round I, and the audience, heard 8-10 "finished" and everyone was a liar or a quitter (well, someone "gaming" the playoff system in a smart way, but that's what I view either action as).

Now, lets talk a little bit about playoff structure. Ties were broken in this order: first, total points, then time to turn in, then existing score order. So I had the third one. Except, if you turned in an incomplete set of solutions to "post" a time by giving up on the last "sudoku", then you would beat me if I didn't finish. I refuse to not try to solve a puzzle, even a "sudoku" which I've learned to "solve" fast well since championships call on the best sudoku solver to solve "sudoku" too. So when the scores came out, and 8 people were tied at my score of 31 including the person who got credit for "winning" my group, the fact I had not handed in my paper a minute or more early left me (and others) on the outside looking in. Another broken structure in another broken round in a horribly broken "championship".

For scoring specifics, there were three 36's (with 1 unsolved puzzle), one 32 (with 1 unsolved puzzle), and nine 31's including me that tied without the one without a solving path. The points were hardly representative of the difficulties anyway, so you can consider people at 3/4 or people at 36/32/31 for our purposes. So, 13 people needing to be cut down to 8 including 9 ties. Well, for not stopping, solvers like me, Hideako Jo, and Jan Novotny were eliminated by not quitting soon enough. Also Cen Chen of China and Jocelyn Bogreau of France. I would be unhappy both eliminating or being eliminated on a "who quit sooner" criterion. Sudoku are simple to partially grade and a tiebreaker based on how far solvers were into the fourth sudoku in terms of correct digits would be incredibly more fair than what happened. Of course, if the sudoku has no sure digits as the 4th one did, then even that tiebreaker is more chance than anything, but at least it gets solvers to keep solving puzzles because some part of an answer is worth something.

So, the terminology we will now use on this blog is that there are sudoku and "sudoku". The first can have you tell me how it is meant to be solved and the latter not. The Guinness record puzzle is certainly a "sudoku". "Sudoku" are possibly interesting computational exercises, but not for human solvers. "Sudoku" tend to occur at a "Championship" like this, where there might be a Champion - who solves the most puzzles consistently well and gets a good number of the "puzzles" too - but there is also a "Champion" - who "solves" enough things to survive an individual round with 35 others and then again a round with 7 others. That guy may be a good solver, but he/she also got lucky at the right times several times.

As I said, I'm retired from WSCs, the question for me is if there will be another WSC without the C in quotation marks. I'm not necessarily retired from those. And while I know the US organizers will make a good one without me, I feel there are enough innovations to return sudoku-solving to solving and not bifurcation (even in classic puzzles only) that I really need to be a man behind the curtain next year.

9 comments | Leave a comment