I participated in the German Logic Masters on Saturday. Ulrich Voigt won the play-offs, ahead of Martin Merker, Philipp Weiß and Nils Miehe; I placed fifth. Puzzles and results should be published at the page linked above at some point. I’ll probably be posting some puzzles I made in preparation in the future; for now some thoughts on the tournament.

Round 1 (25 minutes, 70/120). The traditional welcome round, also traditionally not that welcoming. A couple of mostly standards (Slalom/Slant, Snake, Different Neighbours, Sterntaler, Tapa, Coral, Trinudo) with 3-2-1 themes. The Sterntaler was a bit of an odd one out and a bit too tricky for my taste as the least familiar type in a welcome round, but happily I skipped that one anyway. Less happily, I made a mistake half way through the Trinudo which rendered the end unsolvable in a vague enough way that I kept trying to fix that. But this was an ok result, I always do poorly on these first rounds, and Ulrich was the only one to finish, with one minute to spare.

Round 2 (75 minutes, 195/350). A round of JaTaHoKu puzzles. JaTaHoKu is a somewhat bizarre hybrid of Japanese Sums, Tapa, Skyscrapers and Sudoku. Three “standard” puzzles, one of which might qualify as straightforward, as well as knapp daneben, cylindrical, triagonal and cryptic variants. I solved the standards (but made a stupid mistake for 30 points wrapping up one of them), the knapp daneben, managed to get the big one (the cryptic) out, skipped the cylindrical after not seeing any break-in and guessed myself close to a solution on the triagonal before time ran out. Almost all of these were past the fun level of difficulty for me. The cryptic solve in particular could easily have gone wrong, and except for the first one none seemed like clean solves.

Round 3 (60 minutes, 90/300). A round of stacked puzzles (two puzzles of different types with equally shaped grids with some interaction between cells situated “above” each other). While I had scraped by on the previous round, this one got me. Failing to realise that maybe the authors had generally missed the mark in terms of difficulty, I again chose to hit the big one (Dotted Snake over Japanese Sums) after one that should come easy (Fences over Domino). I did manage to get through the fences, though not without more guessing and case distinctions than I’d like, but then I must have spent half an hour on the big one before abandoning it — I had something that appeared to be close to the correct solution that I just barely but reliably couldn’t get to work. After giving up, I got the one easy puzzle of the round out, and ran out of time on one other.

Round 4 (75 minutes, 320/505). The standard long round of assorted puzzles. This one seemed to have the best balance in terms of difficulty, along with round 1. Or maybe I’m just saying that because I wised up and skipped the big 70 point Summon, even though I should know that type better than most. I did get the three next largest (Pentowords, Loop Fillomino and ABC-Box), though those all involved more guessing than comfortable (not sure, the Loop Fillomino might have gone fine). I had a go at the Summon today, and I don’t think it’s a puzzle I’ll want to finish even without time pressure…

Round 5 (45 minutes, 80/200). Several non-unique standards where we were asked to find both solutions. I was hoping this would be an easy round — it didn’t turn out to be. I messed up here, by not really realizing the scoring mode: Full points for both solutions, 40% for just one, which makes a lot of sense. Quite annoying after an otherwise thorough preparation. Not that it would necessarily have made that much of a difference: I did think I found both solutions to Fences and Kropki, but made stupid mistakes on both of them (connecting the ends wrongly to form two loops, and inexplicably writing one single 5 instead of a 6). Finally, I ended up with one solution for each puzzle. This is one round I’m looking forward to solving again, I think it might well be a nice one by itself.

Round 6 (50 minutes, 125/240). A selection of puzzle types from the Rätselportal. While it started bumpy with a very cheap Geradeweg where I didn’t see the logic, this one went nicely through Neighbours, Skyscraper Blocks, Kropki-Cave and Achtelwenden (yes, I learned not to do the hard ones early). But in the end, it didn’t matter if I solved those quickly or not, because I could spend the rest of the round not getting the really rather hard Star Battle-Coral out. Maybe I should instead have spent it not getting the equally over-hard Abstandsgebiete out.

Round 7 (60 minutes, 245/320). A set of six broken puzzles (ABCtje, Coral, Fences, Fillomino, Japanese Sums, Sudoku), to be fixed by replacing some of the clues using some predefined sets of replacement clues. This one was also pretty hard, but it didn’t feel that far off the mark, compared to some of the comparable rounds in previous years. It was certainly a bit too large for the time, but the difficulty as such seemed fine. I got the complete mapping of correction clues to puzzle types, and solved all but ABCtje and Coral, having enough time left that I should have stood a chance at the Coral if I was solving better.

If I’m to judge the competition over all (trying to ignore my own performance for the moment), I’d say that while the puzzle quality was excellent, balance wasn’t.

The main issue is difficulty, which I feel was quite far off the mark. There were a few individual puzzles which were just too hard for this kind of contest (I’d name the crypto JaTaHoKu and maybe the cylindrical, the Dotted Snake/Japanese Sums, the Summon, the Coral/Star Battle). These are puzzles that would have had three to four stars of difficulty on the notoriously hard Rätselportal. (I think the working definition of “easy” (two stars) is: “Looking back at this puzzle I just solved, none of the individual steps seemed that hard.”) Then, some rounds should have been a bit easier on average (rounds 2 and 3). The JaTaHoKu round in particular would have been much better if half of the puzzles had been straightforward. Finally, there were no breathers in the whole competition. A single very hard round could be just fine, if you also have some medium and easy rounds around. This was over six hours of hard to very hard puzzles.

(My highly subjective definition of “too hard” here is something like: 1. An individual puzzle is too hard if I personally can’t expect to reliably work my way through it in half the round’s length. 2. Several puzzles are “too hard” if I regularly need to use case distinctions or guesses to make progress at good pace. You may prefer these alternate definitions: 1. I lose a competition because I can’t solve this puzzle. 2. I don’t enjoy solving these puzzles in a competition.)

Then, there were too many puzzles. Maybe we were just that much worse than the test solvers? To my knowledge, Ulrich finishing round 1 was the only early finish at all. The high difficulty might play a part here; I expect that we might have been faster at round 7 if we didn’t have the previous 6 hard rounds before.

Finally, I’d argue that there was a bit too much focus on Rätselportal-style hybrid puzzles, and maybe a bit too few classics. We had a full round of JaTaHoKu, a full round of stacked puzzles, as well as Sterntaler, Kropki-Cave and Coral-Star battle. Authors’ hand-writing, I guess, just like the high difficulty.

As to my own result, naturally I’m unhappy that I failed to qualify for the WPC. (I expect I’ll eventually be happy to take part on the B-team.) Regarding the reasons, I’m sure that the high difficulty is one factor. It’s certainly something I’ve struggled with before, and while I think I’ve improved there, not enough to deal with this level. But I’m also pretty sure that I’d have dealt with this better if I was in better (non-puzzling) shape; after round 3, I was worn out. Finally, expecting myself to do well really doesn’t help.