2012 Times sudoku championship write-up

Long story short:  a permuted 2 and 8 in R4C1 and R5C1 in puzzle #6 cost me a place in a grand final that, given my subsequent solve on the sidelines in 14 minutes and 25 seconds, I'd have probably won.  Still, I think the long story is worth it, so you'll have to allow me to indulge my writing whims dearest reader, and we'll walk down this cathartic road together.

This story begins at a wedding I was in at the beginning of June, where in passing conversation Andy Dolan mentioned whether I'd seen the fact that in the previous week the Times has been running the qualifying puzzles for their sudoku championship.  This took me aback a little because as far as I could remember, this always happened slightly later in June.  This was all a little academic for my own motives, because having finished 3rd last year I had a complimentary position already, but I had promised to keep a lookout in the papers for Rishi Puri, who has been studying in London this year, and as arguably India's best solver was quite interested in a shot at the title.
Later that week, I pass on the news and after a bit of a fluster he managed to photocopy one of the puzzles, and sent it off.  Fast forwarding the story to where he got his letter of confirmation, we find ourselves back earlier this week.  After mentioning that I was keen to avoid the fuss of last year, where I managed to arrive at Murdoch Towers 5 minutes before the competition was due to start, Rishi invited me to come and stay with him in London on the Friday night and enjoy a few beers and some curry.

Waking up on the Saturday I was perhaps an hour or two's worth of sleep short of what I'd have liked, but on the other hand I was solving the puzzles in the magazine quite quickly, and we were up and out the door at a reasonable time on a beautiful morning.  In fact we were massively ahead of our schedule, and so we even had a little time for Rishi to go sightseeing on Tower Bridge after we'd got off the train at Fenchurch Street.  Along the way we managed to bump into none other than Mike Colloby, UKPA treasurer.  After a few pictures, we strolled through St Katharine's dock and arrived at Murdoch Towers comfortably on time.

Andy was there too, and pointed out the free papers on the side.  Interestingly, this year they hadn't purposefully removed the puzzle pages and so Rishi was able to get in a bit of pencil and paper practice in before the competition began.  As detunedtom and purifire on the fed sudoku website, there isn't generally too much to pick between our online times, so it seemed inevitable that when Rishi started solving the samurai, I'd try and race him.  He started a minute or two before me, and I finished a minute or two after him.  Once again, too close to call.

Before the first of the preliminary rounds started, I had a chance to say hello to a few more familiar faces, including UKPA forum regulars Roderick Grafton and David Collison, 2011 champion George Danker and his parents, and serial pretender to the crown Mark Goodliffe.  Apparently Mark was on ominous form from an event in the mind sports olympiad a few weeks ago.

So, on to the competition.  I'm not quite sure what happened in the first round, but I'm normally quick enough to be able to recover from mistakes and still finish at the top end of the pack.  For example, last year in the first round I managed to cock up one of the first four puzzles, start again and still finished 2nd in the room.  This year I started solving quite nervously, writing in the odd digit here and there which was inexplicably wrong.  I got the first and second puzzles out in solid if unspectacular fashion, but the trouble lay with the third puzzle.  It was pretty unremarkable, but did require spotting a couple of naked singles to get you going.  (And yes, I am definitely laughing to myself having written that sentence).  I ploughed on through it pretty slowly, but towards the end I realised the remaining cells to be filled in weren't going to resolve uniquely.  Cue the ringing of meta alarm bells.  I decided to ignore this and bash out the fourth puzzle, again without particularly striking speed, before returning to confront the third puzzle.

The sensible thing to do in this situation with a niggling doubt lingering about whether there's a mistake or not is not to try and fix the thing, but to simply go nuclear and start again.  Unfortunately this was the hardest puzzle on the set, and I always seem to go slower the second time on a broken puzzle.  I suppose this is simply a psychological impulse to be more cautious.  This took what I thought was an unspectacular time of I'd guess 16-18 minutes for the set to a time of 22 minutes.  At the end of the round it was announced at least things were all correct, but I'd placed only 12th in the room.  12th!  Mark's room leading time was later announced as being 18 minutes, which is perfectly respectable, but doesn't seem lightning quick to me considering this implies an average of 4 and a half minutes per puzzle.

I spent the break rationalising that doing what I did was the right thing.  One side of my character is fairly secure, knowing that the things I choose to do with my time are things that I take seriously and do to the best of my ability.  This normally complements another side of my character which is massively competitive - I can't possibly be good at everything so when I'm not I don't have to take it to heart.  This however was different.  I decided to arbitrarily blame 12th on using pencil, announced to everyone I was going to blitz the second round, and clicked my pen open in much the same way I'd remove the safety catch from a gun.

The second round was up there with some of my best ever speed solving.  Unfortunately this means it wasn't like my 3 nikoli time trial wins, ahead of a world class field, it was probably more like my first round in Beijing, 2011.  In that instance I solved 8 puzzles in 18 minutes to finish the round in 3rd place behind undisputed speed kings Jakub Ondrousek and Jan Mrozowski.  Unfortunately I had 4 incorrectly placed digits across 3 puzzles to piss right all over my parade.  Here, I must have had the first two puzzles, nominally rated "Fiendish" by the Times, out in less than 4 minutes.  This is a good time to finish just one of these puzzles!  The third puzzle was a bit stickier, but I must have had that out in 3-4 minutes.  And with the fourth puzzle I simply missed the trick about two thirds of the way through the puzzle - a box reduction in the top centre - but feeling my momentum slipping I took a guess and got it out, probably in the order of 5 minutes.  After a quick check for empty cells, I declared in about 13-14 minutes.

This was 4 minutes ahead of the room.  I had achieved what I'd set out to do - to utterly destroy the room.  I was a little concerned about my guess and that I probably had a mistake, but I didn't care too much.  Live by the sword and die by the sword and all that.  I suppose it probably sounds a bit funny and disingenuous when I say the trophy and the prize money are secondary to the pride I take in being best.

At the end of the round, it was announced by just how much I'd smashed the room, but when the list of 8 finalists was announced, David Levy's caveat at the end was that I'd made a mistake.  I'd spent the rest of the round mentally preparing for what is beginning to seem like an inevitability so I took this quite well; but all the same reflecting back on things now, the fact I've had to use the word "inevitability" is something that is beginning to gnaw away at me.

Later, I'd check to see exactly where my mistake was.  I initially thought it would be on the 3rd or 4th puzzle, but observed with a certain hint of dry humour that I'd probably cocked up something silly on one of the first two puzzles.  As with Beijing, this would turn out to be exactly the case.  Two indescribably frustratingly permuted digits.  I could only let out a weak cry that "it's a dickhead error."  Again.

Rishi had made it through to the final, and I had a few words with Steven Gerrard, with Mark and with George who had all also made the final.  I commented that the field was quite open and that the top 8 were all quality.  Privately, I knew this was Rishi's to lose.  The UK standard is certainly getting better, but Rishi is an outside bet for a world championship play-off some year, and is certainly a league up from the usual Times championship crowd.. I should add here, this is definitely not a denigration of the usual Times crowd,who as David Levy also pointed out have been getting better year on year. Anyhow, Rishi had not qualified spectacularly well, but had an awful lot in reserve as he'd been spending minutes checking before handing in.

An interesting point that I'm sure people will think as they are reading this is whether I should have done something similar.  My answer to that is probably not.  I am an appalling checker of my own work.  I have hammered done the submit button in online interfaces assuring me I've made a mistake but where I've gone and scanned through everything and verified in my own mind that I am right.  It always turns out that there is a permuted pair of digits or something similar - but the point here is that in checks of boxes and of columns there is no mistake, it will only be in two of the rows.  Very easy to pass over.  I'd be confident of actually spotting something like this 50% of the time.  My strategy instead has always been to ensure that I slow down as I enter the last few digits, as this is where the errors always slip in.  I'd argue that I've gotten better at doing this over the years, but I can't explain why I keep doing things like I did here.

As always with the grand-final, there was one minor organisational cock up.  It turned out that (thanks George for correcting me! - see comments below) that Ned Walker had been missed off the list of qualifiers.  Morally this should have disqualified George from the final as the 9th best finisher from the preliminaries.  He was still allowed to compete.  I was to be solving on the sidelines again, just like the pain of 2010.  As I was solving, finishing puzzles 1 and 4 in less than 5 minutes, I commented to Andy that these were easy.  The second puzzle was blitzed in less than 3 minutes; and finally, perhaps inevitably, the third I cocked up at the end.  I still had time to redraw a grid from scratch, cane through it and finish in a time of 14 minutes and 25 seconds.  Rishi had declared slightly before this, but as with the samurai earlier in the day, he had started a minute and a half before me.  I estimate that our times would have been close, but I take the opinion that with the option of asking for a fresh grid, rather than having to redraw my own (effectively a 30 second penalty), this was another title that would have been mine, just.  If I'd made it to the final.  Too bad.

As it was, I'm glad Rishi won given that I didn't it.  It adds serious credibility to the championship to have a solver as good as him take the title, and whenever we've met at various competitions around the world we've got on really well.  Saturday was his birthday as well, and it was great fun to party the night away.  Long live the champ, for this year.  It's a run of 3 years now without the title for me, so you'd better believe I'll be coming for it again next year!

(I'll probably post up the puzzles from the grand final soon - can't be arsed to do the formatting now...)