Below I’ve collected some thoughts on the first half of the WPC 2015 in Sofia, mostly about the puzzles and how they went for me. The second half is still to come (here), but this is getting and taking so long I think I’d better post now. I’ve collected some information from the full result table at http://www.wscwpc2015.org/wpc2015_results.xlsx, particularly the 10th best score of each round, which seems like a good point of comparison. Especially for my personal analysis, since I was hoping to take a shot at the top 10 this year.

Since this doesn’t come with any summary yet to put the inevitable criticism into perspective, I’d like to start by thanking everyone involved, but particularly the puzzle authors. While there were some issues with puzzle formatting and point allocation, and general organizational problems, the puzzles themselves were of consistently high quality; I’m not aware of a single broken puzzle.

Edit: Here’s a link to the instruction booklet.

Round 0 – Puzzle GP finals

Competitions proper started for me on Wednesday night with the GP finals. This was 10 person play-off among the filled-up over all top 10 of this year’s GP series.

I didn’t notice too much of what happened around me, but the table at http://gp.worldpuzzle.org/content/final-results-1 gives all the details. You can also find the play-off puzzles there; they were overall of very high quality.

The Scrabble puzzle at the start clearly caused trouble for the two Japanese players. My solve was strange in that I forgot about the longest word until the very end, and quite happy I still squeeze it in. Was anyone else wondering how GOOGLEPLAY made it into the puzzle.

The Slitherlink went smoothly, then I was … very lucky when fixing a careless mistake in the Tapa. I really struggled with the Neighbors puzzle, bifurcating a couple of times, and not just one level either. Not at all confident when handing in, I had no idea what to do should it have been wrong. But, lucky. I’ve solved it from scratch afterwards, and made my way through deterministically, but feel like I must have missed some argument there. By the time I had worked my way half-way through the Pills puzzle (same weird rendering as in the GP), the top three had finished. The final Find the Pair by Ko Okamoto puzzle is quite worth a look by the way, it actually has a very elegant “solving path”.

Congratulations to Ulrich Voigt, James McGowan (from 11th!) and Ken Endo for finishing on the podium. I made my way from 7th to something like 5th, which was quite reassuring since I went into the WPC a bit unrelaxed and unclear on my puzzling form.

Round 1 – Welcome

This was a nice surprise, after some hints in the Q&A session: A variety of rather easy standard puzzles on Möbius strips. Beautifully done, even if the puzzles didn’t really make logical use of the Möbius strip. Unfortunately there was a printing problem, with the 111 Tapa clues coming out as 11.

I thought this went pretty well, finishing with 7 minutes remaining. It turns out the rush got the better of me and I should have done a little more checking: I missed one mine in the Minesweeper, one black cell in the Tapa, on tent in the Tents, and wrote one number across two fields in the Different Neighbors. That’s 140 instead of the expected 470 (minus extra checking); 10th place was 400. (140/400)

Round 2 – Classics

Classics are usually not my strong suit, but this round went quite well. No regrets here, except maybe that my broken solution to the harder Triangular Minesweeper would have been really easy to fix. (520/495)

Round 3 – Latin Squares

Latin square puzzles are even less my thing than classics, but again this one went decently well. I lost 20 points to a careless mistake on an Ayda puzzle, and a proper mistake on the Product Skyscrapers. I’m not sure the 2D Magic was worth the time I spent on it, even though I got through it quite cleanly with knowledge of some “extra regions”. (510/645)

Round 4 – Pathfinder

Generally I’m good at path and loop puzzles, so this one was meant to be good, even with some uncomfortable types. I didn’t really do as well as I should have, putting quite a bit of time into puzzles I ended up not solving. Notably, I couldn’t make the easy Worms puzzle work, and I repeatedly and convincingly broke the Dotted Snake. I had to guess on the Double puzzle, a type which feels a lot like Geradeweg, so I think that one still probably went well. The Word Connection was fun, the Crossing Lines puzzles seemed overvalued, but perhaps that’s just because I’m familiar with Laser from croco-puzzle. (475/415)

Round 5 – Hybrids

A round of very tough hybrid puzzles that went spectacularly badly for me. I may have started with one of the Easy as Tapa puzzles, which were anything but easy. I moved on quite quickly to the Masyu Pentamino. I found the elegant break-in quickly, solved to about half-way, contradiction. I did this twice more, and with half the round over, I was at 0 points. At this point I thought to cut my losses and collect some easy points, except there were barely any. I solved the easier Japanese Railways and the Coral Magnets quite smoothly, but everything else I touched I got stuck on. Drew in some rooms on the Easy as Tripod, say, but couldn’t tweak it to work. Some 15 minutes from the end I started on the Easy as ABCD123 and made steady progress there, but too late to finish. I can’t even blame the broken Snake in the Naval Forest (the non-standard fleet was missing).

I’m still a bit unsure what to make of this round, and my performance. I’ve since solved all puzzles, and all of them turn out solvable, for quite a few there’s even very elegant solving paths. They’re also almost consistently very hard, a lot harder than the later round of “Tough Puzzles”. Luckily for me, others also struggled, the round scoring by far the lowest points per minute, so that limited losses somewhat. Others were hit harder, particularly Yanzhe Qui with 60 points, who made the play-offs anyway after a superb second day. World champion Ken Endo scored 175, less than half of Ulrich Voigt’s 395. (65/380)

Round 6 – Poker Battle

This one came after knowing I messed up the previous round (but before the results of round 1). I got through reasonably smoothly in a careful steady solve, with the only hiccup thinking the puzzle did not resolve uniquely until I remembered the final interpretation of the Joker rule and how it applied. The fact that someone carelessly gave a wrong interpretation of the rules on the forum instead of checking the puzzle or with the author is one of my main points of critique towards the organizers — as with all the other problems, it probably comes down to lack of manpower and time towards the end. But then better just admit to that and tell us to wait. (225/245)

Round 7 – Color Me In (Team)

The day ended with a team round of four color-themed puzzles by Tawan Sunathvanichkul of Somewhere over the rainbow. Each of the four team players chose one colored crayon, and each of the puzzles needed three players to contribute to the solution. On team GER-A, we approached this by taking one puzzle each and solving with pencil: Florian took the Tapa, Philipp the Fruit Basket, Ulrich the Pentomino and I the Akari. Philipp got through the Fruit Basket well, probably within the first half of the allotted 40 minutes. I was making steady progress on the Akari, and with Philipp’s help we finished that one with time to spare. I don’t remember precisely how long Florian took to solve the Tapa, but after he was finished, he and Ulrich put some more time in to the Pentomino puzzle, which appeared to have turned out quite a bit too hard, so at some point they aborted that. It might be that the Akari still wasn’t quite done by this point, I’m not sure. Anyhow, we wrapped up the remaining puzzles and colored them in, with some time left to check.

It turned out that no team finished all four puzzles. As far as I know nobody finished the Pentomino puzzle. So with 1500/2000 points this would have been a good result. Unfortunately, we found out on Friday evening that we made mistakes on both the Akari and the Tapa. The Akari mistake would have been hard to spot due to the color composition rules, but with two people checking the Tapa solution, it’s a bit disappointing we didn’t find that one. Though admittedly the puzzles were huge.

So, what to make of this round? I think the basic idea of the round is great. The color-specific rules on the Tapa and Akari were fun (I’ll post a practice Color Akari I made when I get around to rendering it). But, there were two factors which made the round suffer quite a bit:

For one, the difficulty and/or size of the puzzles was off. Certainly, the Pentomino puzzle was not solvable in the time, probably not even with a full team effort. I’ve gone through it since, and after some initial progress, I made my way through eventually after making heavy use of uniqueness deductions. I probably missed something, since in the end the solution didn’t quite turn out unique. If I remember correctly, I would have wanted the green pentomino set to have an extra X for a Z. The other puzzles were probably fine, but the sheer size made them hard, in particular hard to get completely right.

My bigger complaint is the way that the optimal approach made this mostly a non-team round: We solved individually, then just came together to fill in the colors. Here I would much have preferred if pencils had been disallowed and teams had to solve the puzzles one-by-one. Something similar was executed well last year in London, but of course that requires more than a handful of invigilators. I’m sure that Color Akari would have been a lot more fun with three solvers drawing in different colors at the same time, Green shouting “I need a red over here” when they see that’s needed. That should have made for far more interesting team work.

In summary, a great concept that was harmed by what appears to be lack of testing and lack of manpower.

---

That’s it for day 1. I was feeling pretty bad after those rounds 1 and 5, so I worked on that with a beer or two too many. Tune in soon to hear how that worked out.