Log in

No account? Create an account
27 August 2011 @ 12:33 pm
USPC Results  
I'm sure everyone found that hard in places. I certainly suffered from two big mistakes - taking too long to spot the 10th difference (and then not even getting it to start off the test), and then finishing Word Connection 2 minutes over so it was not worth submitting. I think I have everything else but I'll give myself just 5 points tentatively on my count of XX (revealed much later after the test).

So 354-364 it looks like as a ceiling. Exactly where I normally am.

Lots of good puzzles, and I'll take some time to digest them and go back over each of them after I grab lunch.


ETA: More likely 349 as a ceiling with no counting sub-bonus even. I'll at least say the lack of penalty made trying the counting puzzle more likely than before although perhaps I should have saved it for later?

Day didn't start off great with jackhammering on the sidewalk next door for some repair. I normally have music on when I take these tests, but I had to switch to headphones on with my iPod to better block the noise but it wasn't perfect. Then when I got the test printout open I was having a heck of a time spotting differences (except for the obvious same way facing thing that didn't get reflected). A fair bit of time finally got me to about 7, but I wasn't getting far so I went to the other side of the page.

Puzzle 3, as I expected before the test and only vaguely stated, was a word search/criss cross variant that was much more about the criss cross side of things. I'd half expected the "reverse word search" gimmick as I'd seen it before but without all the loops. So I knew what to expect and it still was just solving it. That in the can, I went back to the differences and got 2 more. None of mine were in the lower 2 rows which was odd. So I stared there for awhile. Only after I had lunch would I find the chick had moved. That chick alone would cost me the chance to equal Melon.

Puzzle 1 was a slightly harder battleships than usual but I followed the steps I described before the test, with sure ships and speculative others. Pretty sure the battleship was vertical, so it was fitting in the two cruisers that was the tweak I really needed to make. Pretty efficient, followed by a hard but good sudoku.

About 25 minutes gone for just 54 points. I bet on a graph of progress this was my lowest slope by half hour and that is not what the first half hour should be.

Off to the back of the test to solve forwards. The Dynasty Sudoku started a lot easier than my practice test but I for some reason missed one of the simple innie/outie placements of an empty square for awhile and then didn't get the right blocks before I felt like guessing. I can't recall if I forced it or did it legit, but it eventually fell. Not perfectly efficient, but a very nice puzzle as expected from the Adam R. Wood/zotmeister.

Hungarian Tapa also started very well until I felt I was running into contradictions in the lower-right with the 8 clue and the 14 from above. Dipping a cell into the last row eventually fixed that, but not before I'd solved around it to have a 43 flip up top in column 7 that was getting all the sums off by 1 on the left for the 22 and 24. I fixed that too, but having two heavy tweaks in a puzzle made me take extra time to check everything out. Still, a stellar Tapa from Serkan.

Jumping Crossword broke for me with the intersecting 8's in the center. FINAL and LOFTY needed to intersect and COSTS and SORTS. The question was which way. I think I ended up being a bit lucky. This type is always very doable for me, just takes some time.

Kaku Rogue was the Michael Rios Kakuro variation we all love to see happen. Very nice progressive solve from finding columns or rows that need negatives, then eventually placing the negatives for sure, then solving slowly but surely with really great forced extremes throughout. Very nice design with a good deductive path with the upper-right corner being both where I ended up and where I probably didn't figure out what to do to resolve it so much as plug in digits for a bit.

Corral Crates was another gem from zotmeister who was also the expected author since he'd made this variation even before Melon had, circa 2006. This one had more than the one internal island in the example I solved from him last night, and was fun up until the square counting for answer entry.

Hopper was an intuitive solve and I think I had several of the right connections (the starting piece with the west virginia piece, for example) on the first go. I needed to get the 2x3 rectangle somewhere though as I had the rest working. Seeing that path could come from the right side of the top of my structure, I rejiggered once and was done. I certainly made a much harder Hopper than the USPC puzzle, so apologies if I scared people off the type. It still is very much a piece assembly/be greedy kind of puzzle where you need to find neighbors for the 6's and 1's most of all, which is what my example taught me, as well the fact that checkerboarding (even and odd squares will alternate on the final grid) could help if you really needed to think about packing constraints which was less relevant here with 50%+ open space.

Skipped 15/16 at the moment as I didn't see a 10 second work-in on either and I was finally solving efficiently so I wanted to bank logic points.

Blocks was a very nice maze variation with tetrominoes (my expectation) from Serkan. Very smooth solve with figuring out how to hide pieces best from the end, then the I and O, then working out the S/Z in the top.

Corral was a quick 10 points, again slowed mostly from having to count for answer entry which still feels the worst of the answer entries we use here compared to the LMI standard which would probably be "connected whites in targeted rows". I don't recall seeing the situation like the 7th row in this puzzle before, not that it was hard to work off of, but the absolute chasm that built very early there was different.

Takeout was the same type of puzzle as last year, except for the "highlight" of finding that last word was actually backwards in the letters in the second unfilled row. I had been looking for it for awhile so to find it there was amusing.

I hit the first (not second) of the Flash Cards on my return trip by figuring out which two products would even be close. Then did the Crack It On. The instructions were ambiguous yesterday about "grid" and "diagonals" so that I thought there could be several as this puzzle is usually presented. I had practiced with 3x's but this was a good hard 2x. My break-in was looking at columns 23 on the left hand grid and trying to match X1X23 and X123X in any words. Sadly it was a word at the very end of the list, TITLE, that gave me what I needed but once I had that sure work-in it was rather straight-forward.

The Square Count got a quick answer, then I think 3 more quick answers as I corrected certain subcounts. Still not interested in being sure of my answer, or if Nick actually meant subcounts descending in square size order or number order but I went with the former.

The Masyu was incredibly nice and used a familiar theme in the black sections, including with the long vertical that will go between them. Melon seems to think its a juno but I know a few other authors (not many) who could have done it. Hopefully we'll eventually learn but from the USPC to this point "Nikoli" is all we ever see.

Times Like These was the lone Friedman puzzle this year it seems which I'm very happy with. I realized it would solve by figuring out how to bury 0's in certain products, like X0Y in the first row. Certainly a fun new design with good math teaching properties. Wonder what more can be done with it?

Easy as ABCD was easy as ABCD, even without any croco-puzzle practice the last month.

That left the STD (unlikely to return to), the second Flash Cards, and two word puzzles. I figured I'd go in return order starting at the Flash Cards which just needed second level thinking about what similar types of fractions (that add to an integer) get flipped. ?/27 and ?/9 seemed pretty likely as working out when 6/27 reduced to a 2/9 and it did. I am so thankful that both of these puzzles used just 4 cards where the example was WAY TOO HARD! Bad Example! Bad!

While I missed a work-in on the hex words earlier, I got a very promising triangle in the lower-left with UPDATED, STELE, and BRITT with nucleated nicely for the rest of the way. A bit of work, and my word marking was sort of sloppy so it took extra time to check that I had the right missing thing.

Onto the last puzzle, the Word Connection, which is probably my favorite newish-feeling puzzle of this test. Serkan had pointed to a link this morning from the Japanese Puzzle Championship in 2002 with SUN, MON, ..., SAT and I was expecting something like that (full day names) or possibly months. Even expecting it might be months, I did not correctly bank the letter data in the puzzle until I'd gotten very close to the answer. Specifically, I knew there were 11 months with letters + JUNE, but I never thought about MAY until the end. That would have been a lot better to spot early, as the APRIL MAY column with JUNE and JULY on the back-swing was pretty obvious to see after I'd done the hard work of getting the months on the left correct. Unfortunately, by the time I was close to the answer orbit, time was out, and I never got to a sure enough state (certainly 80% sure in the second penalty minute) to even try for the 25-20=5 points.

So, a slightly off day, but a very very fun test. My worry was that with so many good LMI tests these days that the USPC would lose the luster somewhat. This test raised the bar in many many good ways. Not a single bad or unfair puzzle. Only a lot of variety, some familiar but well-executed classics, and some very fun new variations. Thanks to all the puzzle authors. And thanks to Nick (and Will) for the continued work to put together a stellar test, and to consider rules updates and such as I'm sure people will agree this is certainly the best USPC yet on almost every level.

I'm saddened that I didn't solve at my absolute best, and seeing Melon solving essentially at my speed on Magic Cube and Japanese Puzzle Land in the last month I expected he would be the rival to watch. On those two LMI tests he proved he can't submit answers. We'll see if that is the case this time. But completing the test early when I have my tentative finish time at 2:37 gives him a 15 minute (10%) whooping of the old champion modulo mistakes by either of us. I've always said I would go to the WPC for sure if I won the USPC, and think about sitting out if I did not. I think I now have to sort out if I actually have 10 days of time when I'm starting a new job only a couple weeks before this trip. There's also the "only do one of the WSC/WPC" option which is the most likely compromise, but if so that will certainly be competing at the WPC and not the WSC.

Palmer is to be congratulated whatever the scores eventually report. There are other unknown solvers, but I'd put money on him right now being back on team USA. The fourth spot is really up for grabs until the results are in.
thedan on August 27th, 2011 09:23 pm (UTC)
Re: oops
I also missed one of the 2x2 diagonals, and both of the 3x3 diagonals.

Well, maybe now I won't have to stress about whether I can afford to miss a week a work. :-/
THrpipuzzleguy on August 27th, 2011 09:29 pm (UTC)
Re: oops
At least you got into the thirties. Boy, did I screw the pooch on this one.