Sunday, 18 September 2011

2011 Times sudoku championship

Time for a “quick” and rather sleep deprived write-up of the championship – if I don’t do it now then it’ll end up like the Beijing tournament where I never quite get around to doing it.

First off, I suppose I should offer a little explanation as to why I now refer to this competition as the Times championship rather than the UK championship. This is because the UKPA now has had two competitions calling themselves UK championships, and as WPF representatives have a more authoritative claim on being able to award a national title. Even if the Times event has more participants, glamour and shiny trophies!

Anyhow, semantics and pedantry aside, onto the report. Coming into the event I was feeling really quite confident in my classic sudoku solving times, and believed that if I could go through the championships without making a mistake, then there’d be at least a 50% chance Times title would be mine for a 3rd time. Obviously there’d be some strong competitors, and with a grand final format when anyone from 8 can take the title inevitably a little chance comes into things, but I thought I had an edge and as such could see little point worrying about things being a lottery.

The day itself started with a minor panic. Registration for the event was between 10.00 and 10.45am, with the 1st preliminary round kicking off at 11.00am. I’d got on the 8.50 train from Coventry to London Euston, due to get in about an hour later, thinking that another hour would be plenty of time to make it over to the News International building near Tower Hill. I had not counted on the signalling issues which had left the train still in the middle of countryside at 10.00am, delayed by about 20 minutes, so that when I arrived at Euston I was a bit of a fluster. I briefly headed down to the underground station, was greeted by a lack of funds on my oyster card together with a monstrous queue for tickets, before biting the bullet and getting a taxi to take me to Thomas More Square, near Tower Hill (as opposed to the one in Hackney – more on this later!)

I passed through the security and went up the lift at about 10.50, with little more time than to appreciate the spectacular views looking down onto the Thames and beyond, say a quick few hellos to some familiar faces (Mike Colloby; George Danker and his parents Martin and Max) as well as some less familiar faces whose names I recognise from the UKPA forums – David Collison and Rodderick Grafton.

At this stage I will have to make my apologies. Lots of lovely people came over and mentioned that they read my blog, which was really quite flattering. Unfortunately I don’t think I’m able to list everyone individually here. Suffice to say I really appreciated all the kind comments people gave me, and I hope you enjoyed the competition!

Anyhow, onto the first preliminary session. I think I was still a bit shaky as I went to tackle the first puzzle, because whilst apparently cruising through it I’d manage to make some sort of screw-up towards the end of the solve. I tried briefly to fix my error, before remembering the lessons of last year’s competition. Rubber out, start again. As is often the way when you restart a puzzle, progress often seems slower, but I plodded my way through and was on my way. The remaining 3 puzzles solved very smoothly though, and when I was done I managed to look round and see George Danker and Nina Pell still working, so I thought I’d afford myself the luxury of a couple of minutes checking. Everything still seemed good, and I turned in. As it happened, even with 5+ minutes wasted from the first puzzle I had still turned in 2nd in the room, behind someone I hadn’t come across before, Gareth Fuller. The usual suspects of George and Nina were also shortly done.

I should at this stage describe the agony of turning in your solutions very early, when you have as prolific a track record as mine of making Dickhead Errors. When I solve for pleasure, I’m always certain I have the right answer because I don’t guess and I have full faith in the processes of reason and logic. I don’t know my exact times, but I’d guess it was well over 40 minutes and it’s not a fun thing to sit there in silence, glancing round at the room and generally praying that you haven’t made a mistake.

Thankfully it turned out that I hadn’t, but I was still a little shaky heading into the short break – a good proportion of my cup of tea ended up in the saucer as apparently my hands weren’t in the most steady of states. Again there were lots of people who I was talking to but I couldn’t really focus on any one conversation. I suddenly realised that there was no sign of my good friend Rishi Puri. I called him up and it transpired that Rishi had made it all the way to the other Thomas More square, but he was hopefully of making it in time to watch the finals.

Onto the second preliminaries. I think it’s safe to say I nailed this, clean solves all the way through, and even after another extended checking period it was still a few more minutes until Nina was second in the room to hand in. Again more agonising in silence for what seemed like an absolute age, but at the end, the results were announced, and I was the best qualifier for the grand finals, having finished the two preliminaries perfectly with the 2nd and 1st quickest times. If I hadn’t already had enough confidence in my abilities, then I couldn’t really argue with these facts.

There was an hours break for lunch, and again I tried to get hold of Rishi – he’d made it as far as Tower Hill but couldn’t find anyone to direct him to Thomas More square. I put him over on the phone to Martin, who is a London cabbie, and sure enough with his calm directions on board, Rishi was there at reception just as I was heading out to pick up a sandwich from the nearby Waitrose.

Quick aside – the Waitrose BLT is really good!

And so onto the grand final. As the best qualifier, I had my pick of the 2x formation of seats. I’ve had this discussion with Thomas Snyder who argues there is an advantage to choosing a seat at the back because it gives you an idea of what your competition is doing – in particular if you have time to do a quick check. My own take on the matter is to simply bury myself into my own world – for this reason I like to take the front seats. If I really believed I was quickest, then anything anyone else was doing would only serve as a distraction. The other qualifies in order were: Nina, George, Mark Goodliffe (the serial Times crossword champion), Gareth Fuller, Matt Cannon, Kit Collingwood and Mary O’Connor.

Onto the puzzles themselves then. Puzzle #1 solved pretty smoothly, and judging by the lack of paper turning noises I was the quickest to get this done. Next up (due to the quirks of the folds in the booklet) was puzzle #4. I was making fairly solid progress through the puzzle, spotted some fairly subtle things which had given me some numbers in what appeared to be sticking points, and generally was under the impression that the puzzle was about to fall. And then I hit a sticking point. I put in some extra pencil-marks in the hope that I’d just missed a pair or something, but nothing doing. OK, fine, onto puzzles #2 and #3. These solved fairly quickly and without hiccup, so back to number 4.

Still nothing. So I’m half a puzzle from victory, so why not take a guess!? (Because you are terrible at guessing?). I try something and it doesn’t get me anywhere. A little erasing and start again. And then Mark Goodliffe declares! Panic! I try another guess and again nothing doing. Another declaration is made. Shiiiiiiiiiit! Shitshitshit! I decide to do the sensible thing and go nuclear with the rubber again. I get to the same stage I was at. Another declaration. ARRGHH! Still stuck. And then I bloody well see it. With a digits in R9C5 and R8C6 and a fixed pair in R9C4 and R9C6, the 8’s in the 4th column and the 5th column trivially forced the 8 to go in R7C6. How the bloody hell had I managed to miss something so bleedingly obvious!? I literally took my palm and applied it to my face. And then filled out the remaining half of the puzzle in about half a minute, as my self-directed rage propelled a flurry of solving. I was resigned to not winning as I saw George was one of those who had declared before me. The other was Matt Cannon.

It seemed obvious to me that Mark Goodliffe had again applied the same all-or-nothing approach to guessing as he had last year – and had as such hopefully made a mistake. I know Mark will probably be reading this, so I hope he doesn’t take too much offence when I say that the puzzles in the final in my opinion weren’t at the difficulty needed to needed for guessing and it would be highly unsatisfactory to me if a title was won on the basis of serial guesswork. Of course I fully understand his position – Mark is the first to admit that whilst he is certainly a pretty good sudoku solver, he isn’t quite at the fastest level, and as such his strategy gives him the best chance of winning. From his point of view it is a no-brainer and I fully respect that! It is certainly nothing personal to Mark when I say that I hope it doesn’t come off for him whilst the final puzzles remain at the difficulty that they do.

As it happened, serial guesswork is probably an exaggeration as he said he guessed only on 2 of the puzzles (as opposed to all 4 last year). Nevertheless a point of interest was raised whilst we waited for the final results to be confirmed. George, sitting in front of Mark, had apparently inferred by the fact Mark had turned in first that guessing was required for the last puzzle, and had taken his own guess to ensure quick progress to the end. This was exactly the position I was in, so it’d be highly hypocritical of me to pass any judgement here. If I’m right in remembering (and please correct me here) I think Matt had also said he’d guessed and that he’d probably made a mistake.

The results came back, and as I suspected, Mark had made a mistake – but the rest of the 7 had solved perfectly, which handed George his 2nd consecutive Times championship. I was left to rue missing something so blatantly obvious that had cost me in the order of 5 minutes. I could apparently afford to do that in the first preliminary round, but as I said to George, with solvers as good as he is around you can’t afford to throw away 5 minutes and expect to win. My congratulations go out to him!

The rest of the competition had me sat in my same chair in a strange state of frustration/annoyance/acceptance, as we continued the post mortem. The two solvers closest to me were Matt Cannon and Kit Collingwood – both it turns out relatively new to the competition and put everything into refreshing perspective. If you are reading, it was good to meet you!

The presentation was handled seamlessly by David Levy, and the trophy was handed to George by Andrew Stuart, author and owner of the excellent Scanraid/sudoku wiki website. I think he was pleasantly surprised when chatting afterwards that I (and numerous other top international sudoku solvers) regard scanraid as THE objective sudoku grading resource. He is more focussed on the solving strategies required to solve the sort of fiendishly difficult classics that I (and numerous other top international solvers) believe have no place in competition, yet still occasionally crop up. He very kindly gave me a copy of his book, detailing some of these strategies. I think it says it all that the techniques I draw the line at before guessing myself – x-wings/swordfish and y-wings – were detailed very early on in the book!

And with that, we said our good-byes, and I headed to the pub with Rishi and Mike. I think I’ve still got a bit of an inner-conflict going with my feelings about the championship and my own personal performance, but looking at things more objectively it was great to see some old friends, even better to meet some new ones, and a pleasure to indulge for a day with a large group of enthusiasts who share the same passion as I do. See you all next year – I will be hungrier than ever!

7 comments:

  1. There’s long been discussion on how to deal with preliminary scores/times when it comes to playoff seeding, and I think I’ve got a good idea; everyone in the final should know exactly how much progress has been made by every solver who finished lower in the preliminary rounds.

    What’s the edge for being first if all it means is you’re herded in with the rest of them? Time advantages are acceptable but are both rarely used, perhaps because they’re such a huge edge. But the understanding of how far a solver has come—that really doesn’t change the dynamic too much (first time wins!) but it rewards those who’ve done well while not giving them such an edge that they can still take longer on the puzzle(s) and come out on top.

    Of course it’s also a logistical nightmare – at least for paper solving (which, let’s face it, is just about the only solving you ever see at any real competition). But it can’t be impossible to display that information (let’s say on easy-to-read placards) so that a column of solvers could see how everyone in front was doing – if not down to some estimate of completion on a puzzle (55% complete) then at least the number of complete solves.

    What say ye?

    ReplyDelete
  2. First off JZ – I’m not sure I can be too critical of the format. After all, I came from being the 7th best qualifier in 2009 after being decidedly shaky in the prelims to take the title in the final, albeit in a fairly dominant fashion.

    When it comes to all the meta information that could or could not be made available to everyone, I think that’d be more helpful to some than others. Like I say, I prefer being in my own bubble, and as Thomas says the information only really becomes useful after you’ve finished solving. I certainly made good use of a quick peak around the room during the preliminaries in order to see that I could spend a few minutes checking. Conversely, knowing what other people were doing in the final whilst I was struggling with my mental block only seemed to put me off even more!

    So yeah, I think the format of the Times championship is generally very good, and certainly finds a worthy champion each year. They used to do 1 puzzle play-offs with the last 8, quarters, semis and a final and so-on, but I don;t know whether that’s any more or less of a lottery scenario. The one thing I might mention is that the I’m not such a fan of the smaller sized grids – you’ll perhaps wryly note that general grids on this blog are 32×32 pixel squares, whereas I think my sudoku format is more like 48×48 pixel squares – and maybe this helped in me overlooking something as easy as I did…on the other hand I had no problems at all with 10 of the other 11 puzzles, and even that 11th was probably a case of making a nervous “typo”.

    Thanks for the ever-thoughtful comments guys!

    ReplyDelete
  3. Hi Tom,

    Really enjoyed seeing your perspective on the day, and thanks for the kind words. I had a great/lucky day and as you mentioned the final su doku I did guess at. I was not sure with the choice of seating advantage for the faster qualifiers (was it much of an advantage?), but think their definitely should be a reward for the faster finishers from the qualifying round.

    Regarding the su doku books that show you techniques, did you find them useful? not sure wether i should invest. When other competitors talk about bipurifecation?, x-wings and swordfish though I do not have a clue. I’m assuming I’m using those techniques but just don’t know the names.

    Cheers Matt Cannon

    ReplyDelete
  4. Regarding books and techniques – I’m not sure I’d bother to be honest. Everything that is in Andrew Stuart’s book is there in an at least more up-to-date version on his website, http://www.scanraid.com (actually he’s changed the name and domain but for some reason I remember scanraid and the link still seems to work, so whatever).

    The trouble with those books is that whilst it makes everything very clear for the examples highlighted, it doesn’t really prepare you to look in the right places in another puzzle. You sort of just develop a knack for it. For example, if a puzzle is requiring swordfish and x-wings, lots of people will just guess (aka bifurcate!), but because when solving I’m generally looking at rows and columns which are nearly full up anyway, you sometimes notice when the gaps align. That’s not to say that it’s always easy to spot them – you can sit and stare at a puzzle for quite some time before finding something, which is often true of the “Very Hard” puzzles as published in the Sunday Times.

    I’m still thinking choice of seating doesn’t matter too much – it’s not too much further effort to look over your shoulder behind you anyway! Actually it reminds me of 2009 – in the final I was again at the front (this time without much choice in the matter having qualified 7th) – and apparently as I finished one of the puzzles and went to take a swig of water, my friend watching saw those around me stop as if they’d thought I was about to declare!

    So yeah lots of mind games, and even the thing with George and Mark was interesting enough. I suppose my point of view is I know that if and when I bother to put up the final puzzles there will be plenty of international solvers (+ David McNeill I think!) who will claim a faster aggregate than George’s 19 minutes without needing to guess – the puzzles weren’t hard enough for the psychological side of things to matter too much.

    ReplyDelete
  5. Oh, there’s no doubt you are a contender these days Mark, you’ve certainly been more consistent in reaching the play-offs than I have in recent years! I suppose you pick up on the key point here:

    “that’s a ‘fault’ of sudoku, not its participants.”

    There’s certainly a fault with competitive sudoku when an obscenely hard puzzle comes up and the only sensible thing to do is to guess. This was certainly not the case here. Perhaps what I want to say is that the fault here is with the general standard of competition whereby the strategy comes into play to start off with. I’ve since resolved those final 4, as published in minuscule print in the paper on Monday in pretty rapid time (and still I felt I wasted time), and I have no doubt certain Czech and Polish solvers would have had those 4 out in a time even closer to 10 minutes.

    Or perhaps I’m just saying all this because my guesswork is generally shambolic, and I can’t quite imagine a situation where bifurcation becomes a realistic and tangible option to take. In that respect I can only take my hat off to you. Bifurcation, as Thomas S previously pointed out, requires a real feel for things – the sort of feeling you can’t get without already being a very good solver.

    Anyway, it’s all a very interesting point of discussion. I get the feeling I’m increasingly in the minority about the whole feeling uncomfortable about bifurcation during competitive solving. And yet there aren’t many (any?) top solvers who would think twice about doing so, if and when it seemed necessary.

    ReplyDelete
  6. Hi Tom – great write-up and thanks for the kind words Mark. I wish it could have finished after round one to be honest.

    I can only add to the majority in saying that I’m an unashamed guesser – I’d like to be able to avoid it but definitely find it’s a quicker way to complete the puzzle.

    ReplyDelete
  7. Thanks for the write-up, Tom. It was good to see you and the others. It’s fascinating to read your own posts and the responses of attendees and of others based locally and internationally. Saturday certainly reinforced the need for me to practice far more about how to undo mistakes under pressure alongside the obvious need for me to speed up my actual solving !

    ReplyDelete

Contact Form

Name

Email *

Message *