Sunday, November 29, 2020

PuzzleDuel Contest- dOubLESHOW

The results for the Doubleshow contest on Puzzleduel were recently finalized, so I figured it'd be interesting for me to talk about my thoughts on the contest and my participation. The format being spread out over several weeks with live updating results was a fascinating change of pace, even if a bit demoralizing at times. But that's getting a bit ahead of myself. The contest page can be found here, but this post will read better if you don't know the results beforehand. There were prizes for 2nd and 22nd place, and I went in hoping for 2nd but resolved to do the best I could throughout.

Puzzle 1Tapa
1Freddie Hand7.1
2Tiralmo6.1
3Para5.8
4EKBM5.6
5Prasanna Seshadri5.5
6umaya5.5
7wandersong5.1
8jaku1115
9Dmitry Grischenko5
10Muhorka4.9
Fastest: Freddie Hand (0:24, 7.1) | 10th: (0:42, 4.9) | Me: (0:52, 1 error, 2.2)
First up was a Tapa that revealed the theme of the contest- every single clue was a 2, in the shape of a 2. I correctly deduced the connection around the top of the grid but had made an error in the bottom half which broke the puzzle when I got back around. Instead of noticing and correcting my mistake, I reversed the top then corrected the mistake and hit check- only to immediately see I had a 2-2 instead. A quick fix later, I submitted the correct solution but ended up in 64th place after the first puzzle- a disappointing start costing me 3 points on the overall standings. I thought this would knock me completely out of the running, but the worst was yet to come...

Puzzle 2Maxi Snake
1EKBM17
2MaM13.4
3GD12.8
4xiao01wei12
5Uhu11.6
6rob11.3
7Tiralmo11
8mstang10.4
9Freddie Hand10.3
10Hyunmo Kang10.3
Fastest: GD (0:19, 11.5) | 10th: mstang (0:43, 6.8) | Me: (0:34, 7.9 points)
Ken Endo launched himself into a lead he would never relinquish, turning the battle for second into a best of the rest melee- I made up some ground on 2nd here, going from 3.9 to only 3.3 behind but that would be a mountain to climb. As for the Maxi Snake, I went in expecting many 2 clues and took a moment to think about the implications and was rewarded with a puzzle I knew only had two reasonable options. I guessed wrong, but still posted a very fast solve time.

Puzzle 3Suguru
1EKBM23.3
2MaM19.2
3xiao01wei16.9
4GD16.6
5Uhu16.5
6Tiralmo15.6
7IHNN15.5
8Para15.1
9Freddie Hand15.1
10rob15
Fastest: EKBM (1:21, 6.3) | 10th: (1:46, 5.3) | Me: (1:45, 5.4 points)
Suguru isn't one of my stronger genres, but it didn't seem to play to any of the top contenders strengths either. A jam of consistent times meant nobody really gained or lost ground here- I was now 3.7 behind 2nd, despite having made my way into the top 10. I don't remember how this solve went, so it probably went smoothly enough.

Puzzle 4Sky. Sudoku
1EKBM30
2MaM26.3
3GD23.4
4xiao01wei22
5Uhu21.8
6rob21.2
7strozo21.2
8IHNN21
9Freddie Hand20.9
10Tiralmo20.1
Fastest: Azade77 (1:20, 7.8) | 10th: Freddie Hand (2:06, 5.8) | Me: (2:16, 5.5)
Skyscrapers Sudoku is a combination of two types I tend to be slower than par at, and it showed here. I'm especially weak when it comes to this sort of 2 themed logic, and needed a lucky guess at the end to even stay competitive here. Still, I dropped to 5.3 points behind second place- a seemingly insurmountable deficit without some incredibly strong solves.

Puzzle 5Queens
1EKBM36
2GD36
3Para34.6
4WMathie31.1
5MaM29.8
6aras29.6
7Azade7729
8xiao01wei28.8
9Andrey Bogdanov28.4
10Tiralmo27.9
Fastest: WMathie (0:19, 17.9) | 10th: Invalid_D (1:00, 8.5) | Me: (3:05, 4.1)
I hate this puzzle type, to the point where I usually just skip them. But as it was part of the contest, I went for it in the hopes of a quick lucky solve. That didn't happen, and my first three "solutions" turned out to break one of the 2s as I found when I checked my work. Eventually I stumbled into the solution, but dropped to 12th with 25.1 points: 10.9 off pace for second place. Looking back, at this point I was fairly unsatisfied with the contest as a whole as most of the puzzles had been quite easy so far, leaving this high variance Queens as what I saw as the sole decider of the top of the leaderboard. I still don't think it's a good contest puzzle (or a good genre...) but I definitely took it way too hard for something casual like this. And besides, I had already resigned myself to a middling spot and was trying to maintain top 10, and that was still well within reach.

Puzzle 6Akari
1EKBM45
2Para39.6
3GD39.3
4Azade7735.1
5Tiralmo35.1
6MaM34.6
7aras34.5
8WMathie34.2
9xiao01wei33
10IHNN32.5
Fastest: EKBM (0:29, 9) | 10th: Muhorka (0:58, 5.8) | Me: (0:39, 7.4)
Akari in the Puzzle Duel interface is a matter of either not needing to mark empty cells, or being smart about which ones to mark. I fall into the later group which is slower, but at the cost of consistency. I wasn't about to make yet another error, and didn't through what felt like a decently tricky puzzle, as shown by the relatively higher scores for it. Para managed to take over 2nd place here after a pair of exceptional solves, and would become the primary contender for the position.

Puzzle 7Ripple Effect
1EKBM54.9
2Para45.6
3GD44
4MaM43
5Azade7742.4
6Tiralmo42.4
7Freddie Hand39.7
8aras39.4
9IHNN39.2
10WMathie38.9
Fastest: EKBM (2:51, 9.9) | 10th: strozo (5:00, 6.9) | Me: (5:14, 6.7)
Thank you, Wen, for making such ridiculous Ripple Effects such that some semi-advanced logic is innate to me now. This was still a very tough puzzle and a satisfying solve, as mostly clueless regions with 6s usually prove to be. At this point I was 6.4 points behind second place with 4 puzzles to go, and not really gaining any ground. I was also too far ahead to effectively sandbag for 22nd, so I kept pushing to try for 5th- a target I was much closer to.

Puzzle 8Magic Snail
1EKBM62.5
2Para50.2
3MaM49.2
4Tiralmo46.9
5Azade7746.7
6IHNN46.1
7WMathie45.3
8GD44
9Andrey Bogdanov43.5
10xiao01wei43
Fastest: Eugene (1:21, 7.7) | 10th: mstang (1:56, 6.1) | Me: (1:36, 6.9)
Magic Snail is a type that I do not understand how to do logically, and so I just frantically put numbers into the grid and moved them around until all of the constraints were satisfied, guided by nothing but intuition and seeing how combinations of placements would break. This puzzle claimed Freddie Hand, who errored lost over 4 points, taking him out of contention. Somehow I reached 6th place on the overall here, my highest placement yet!

Puzzle 9Cave
1EKBM70.1
2MaM54.4
3Para53.3
4IHNN53.1
5Azade7751.1
6rob50.7
7Tiralmo50
8Andrey Bogdanov49.6
9Freddie Hand49.4
10WMathie49.3
Fastest: Muhroka (1:52, 8.6) | 10th: Prasanna Seshadri (3:22, 5.9) | Me: (2:37, 7)
Finally, another puzzle in a type I'm comfortable with, and with easy 2 (and 22!) logic. I took extra care when counting the 22s to make sure I didn't make a logical mistake, something that happened to both Tiralmo and Para, and MaM had a bit of a slow solve here. At this moment, I knew I was in contention for second if I solved like crazy, but it was going to come down to type.

Puzzle 10Nurikabe
1EKBM75.1
2IHNN61
3MaM58.9
4Para58.3
5rob55.6
6Azade7755.1
7Tiralmo54.8
8Freddie Hand54.6
9Andrey Bogdanov54.4
10xiao01wei53.5
Fastest: IHNN (0:44, 7.9) | 10th: EKBM (1:30, 5) | Me: (0:44, 7.9)
And did the type deliver! My Nurikabe rating on the site is currently 3114, which is the best single rating for any person for any puzzle type on the site to the best of my knowledge. This solve was no exception- I went in expecting a 22, identified the region no 2 could reach, spotted the path the 22 would have to take, and intuited huge swathes of the puzzle to end up with 21 cells in the 22. The fix was very simple- just move shaded cells up from the border, and I laughed when I saw that the 22 ended up in the shape of a 2. At this point not only was I in second place, but I was comfortably there by over 2 points- if the last puzzle was low variance and thus low scoring, any decent solve would be enough to keep second! I might actually pull this off! I honestly couldn't believe it, given the 10 point deficit halfway through.

Puzzle 11Minesweeper
1EKBM81
2IHNN64.9
3Para62.2
4MaM60.1
5rob59.7
6Azade7759.1
7Tiralmo58.8
8Andrey Bogdanov57.4
9xiao01wei57.3
10WMathie56.4
Fastest: EKBM (0:23, 5.9) | 10th: Tiralmo (0:42, 4) | Me: (0:43, 3.9)
My wish was granted- an easy, low variance puzzle in a friendly type. Unfortunately, the times were blistering- any lost second would cost points. From the results, I knew that a time of around 45 seconds should be enough to maintain second place, but I really wanted a 35 second time or better. I was a bit slow to spot that a 2 on the bottom was satisfied but turned in a solve time I knew was good enough: only Para stood a chance of catching up, and he would need to surpass Ken's ridiculous time, or else many people would need slower times to increase the scores at the top end. Para turned in a 43 second time as well, and the results didn't really shift as the battle for 22nd went on (snuke took it with a day to spare, and then jaku111 took it in the final hours, narrowly avoiding passing lovemathboy).

And somehow, against all odds, I managed to finish Doubleshow in second place. It was a lot of fun and a difficult experience- I look forward to the next round of contest puzzles here, especially as my strong finishing results have pushed me back into second place on the overall standard ratings too.

Saturday, October 31, 2020

Puzzles #249 & #250 - Ophiuchus (Word Nurikabe)

My favorite comment on the Puzzle Zodiac thread before it started was "No Ophiucus? (13th Zodiac sign)... :( That puzzle will be Snake puzzle..?" just because I knew that I had included a snakelike Word Nurikabe themed on Ophiuchus. It seemed like the splash image I used for the instructions booklet having a hidden symbol in the middle went unnoticed by most people. I poured a lot of time into making this final puzzle on the test interesting, difficult and fair with a stunning theme and it absolutely paid off. I think it's my favorite puzzle that I've made to date and I hope you give it a try.
Yep, I went with a vanity puzzle for the example. My only other requirement was that the grid wouldn't be rectangular, as I knew that the main puzzle grid wouldn't be.
Puzzle 25 on the test, undervalued at 100 points, and themed on A SIGN- Ophiuchus, specifically. As mentioned above this is probably my favorite puzzle that I've ever made because there's so much going on here despite only having 13 letters a grid with nearly 200 cells. Only 7 people solved this puzzle during the contest- and Ken Endo himself even missed it despite having over a half hour to work on it.

Some closing thoughts on the contest as a whole- I think it went mostly well. The puzzle grid typo in the puzzle booklet for the second puzzle weighs heavily on me as an error that shouldn't have happened, and I've learned my lesson that my personal testsolve needs to be on the final booklet, not the final puzzle images. It turned out a little harder than I expected (I was anticipating 5 finishers, not 1) but not by much- and I didn't sacrifice quality at any step of the process. Will I be writing any more contests on my own? Probably, I have a couple of ideas. It takes a lot of time and additional effort to get a contest together, though looking back I'm extremely proud of the set as a whole. Lots of mixed feelings here.

Keep an eye out for more from me- I've got a bunch of variant puzzles coming fairly soon!

Friday, October 30, 2020

Puzzles #246, #247 & #248 - Pisces (Anglers)

Pisces is literally a fish. Anglers is literally a type that has fish in it. What other choice could I reasonably make for the Puzzle Zodiac? All three of these were quite interesting to make, though I'm not convinced the type has all that much potential. Or maybe I just don't know how to tap it?
Oddly, the example puzzle turned out to be the hardest one to make. I'm not sure how much of that is because of my chosen layout, though.
Puzzle 23 on the test, worth 20 points, themed on the Pisces sign. Having 5 and 8 on two sides, and 6 and 9 on the other two sides just kind of happened as I was constructing this one. 12 fish on a 9x9 grid is surprisingly dense!
Puzzle 24 on the test, worth 30 points, and themed with pairs of each clue. The benefit of cluing this way is that I didn't have to worry about parity, but the downside was that it was tough to specify uniquely while maintaining the theme. Somehow 19 ended up working well!

Thursday, October 29, 2020

Puzzles #243, #244 & #245 - Aquarius (Aquarium)

Aquarius was the actual easiest pick for a Puzzle Zodiac type- it had to be Aquarium. I briefly toyed with Ripple Effect but it really did have to be Aquarium.

Like with the Double Chocolate, I themed the example on the fact that it is an example, and made sure to show that water level did not have to be consistent across a room, if there were no walls in between. It might or might not be the standard for the genre- I don't actually know, I just know that it's what I'm familiar with.
Puzzle 21 on the test, worth 20 points, and themed on the Aquarius sign. I attempted to make a 9x9 with the sign as two solid regions without much success in making an interesting puzzle, and so I dropped down to 7x7 with a bunch of 1x1 regions. I think that there's one key step required here that's a bit less obvious.

Puzzle 22 on the test, worth 30 points, and LMI themed. I... actually made this before I checked that LMI would be willing to host the contest, so it's a good thing that they were!

Wednesday, October 28, 2020

Puzzles #240, #241 & #242 - Capricorn (Icelom)

I really wanted to include a crossing genre in the Puzzle Zodiac test, and Icelom was the first one I hit on a good idea for as it doesn't require all ice cells to be visited. I matched it to Capricorn on looking up what sign is allegedly associated with ice, and that's what came up as the answer.
The example puzzle I used- again, playing with different symmetries. Most puzzles of this type don't put numbers on ice cells, but as with LITS and internal borders I had no such preconception and thought it could make for some nice puzzles.

Puzzle 19 on the test, worth 20 points, themed on the Capricorn sign. As with the other less abstract symbols I focused on finding the right aesthetic balance first, and then made a puzzle on top of it. I wasn't great at topological deductions when I made this puzzle so it's fairly simple to solve. It still has a smooth path and I knew having some easier puzzles on the test was a good thing.
Puzzle 20 on the test, worth 40 points, and themed logically again. I don't have many comments on this particular puzzle, other than that it's laid out in a very specific way to have a very specific deduction involved, that's well worth finding.

Tuesday, October 27, 2020

Puzzles #237, #238 & #239 - Sagittarius (Sashigane Fillomino)

As Sagittarius is an arrow, I knew that for the Puzzle Zodiac I needed a type with pure arrow clues. The most common type like that is Sashigane, and as I wanted a common type, uncommon type and variant in each class of puzzle, I decided to combine it with the Fillomino rule. I've seen a few puzzles with this variant before and so I knew it added some interesting depth.
I tried for a while to make a 6x6 example without much success- like Foursight, this type really needs some room to move in order to work.
Puzzle 17 on the test, worth 30 points, themed on the Sagittarius sign. I wasn't sure if I would be able to manage making the puzzle with just an arrow out of arrows, but after the top left worked out I really did try. I ended up with about 6 solutions, and so added another arrow in the corner to resolve the rest uniquely.
Puzzle 18 on the test, worth 40 points, and a failed attempt at an anti-symmetric (sum to 10) theme. I thought I had it at one point but there were two solutions, so I adjusted the back half of the solve, and made sure to maintain the focus on only circled number clues as a counterpoint to the prior puzzle. I love it when a puzzle genre can be used to make such different puzzles as these two, and I think that shows through most of this set.

Monday, October 26, 2020

Puzzles #234, #235 & #236 - Scorpio (Cave)

If you couldn't tell by now Cave is one of my favorite types, and it fit well with Scorpio for the Puzzle Zodiac. I apparently have a bit of a knack for constructing these, as I provided one for the Typed Logic contest as well as a couple I can't share yet for other sources.

Originally, I tried to have 8 unique numbers for the example puzzle, but after tinkering with it a bit I settled on this arrangement. A pretty open solve, again to introduce the genre.
Puzzle 15 on the test, worth 20 points, themed on the Scorpio sign and with only odd numbers. Cave is such a flexible type with its visual and logical theming available- I found this one tricky and was leaning towards 30 points until it was testsolved in under a minute. It's still probably the hardest 20 pointer on the test, however.
Puzzle 16 on the test, worth 50 points, and themed in the logical path. This construction was some good practice in specifically trying to force things to happen a certain way. I'm quite pleased with how this turned out as I ran into a lot of problems trying to keep the givens arranged in a way that would facilitate what I was looking for.