Friday, March 23, 2012

ACPT 2012 Recap

I thought I'd share my 2012 ACPT crossword puzzling experience with anybody who would care to read it...

I finished Puzzle 1 (a 15-minute easy warmup puzzle) in under 4 minutes, which was the second fastest time range. (Six of the elite solvers, including 6-time champion Tyler Hinman and 3-time champion Dan Feyer, solved it in under 3 minutes.) Off to a good start, that meant I entered the competition tied for 7th place (out of just under 600 contestants) - hooray!

Unsurprisingly, as the puzzles got harder, things didn't go quite as well. On Puzzle 2, a 30-minute puzzle, I finished in under 13 minutes, not nearly as fast as the 6 minutes of the top solvers, but still fast enough to keep me at 21st place.

I got stuck a bit on Puzzle 3 (another 30-minute puzzle), but still managed to finish in 16 minutes (the fastest solvers completed this one in 7 minutes -- no, I'm not kidding). Which would have dropped me down to 31st place. If I had completed it cleanly. Sadly, I did not -- I had TRIP instead of GRIP for "Golf coach's concern." Which would have been an OK answer, if the crossing "G" hadn't come from "SPF CHANGS" -- or, in my puzzle, "SPF CHANTS." Oh, well, I was still fast enough to remain near the top 10% (which was my goal for the tournament), in 70th place.

I made back some ground on Puzzle 4, a 20-minute puzzle that I completed in under 4 minutes (I know, right?!) -- of course, again, the top solvers finished this one in under 3 minutes, but you can't complain about a 3.5-minute solve. That pulled me up to 62nd place.

Then there was Puzzle 5. Oh, the drama of Puzzle 5. So first of all, if you're not familiar with the ACPT, they refer to this puzzle as "The Dreaded Puzzle Five." It chews solvers up and spits them out, year after year. I had previously blown two Puzzle 5s -- one year, I had a mistake; one year, I just sat there staring at a whole block that Would Not Yield, and left early after filling in a few random letters, hoping to get at least a couple of bonus minutes. But this year, I positively flew through it, and other than a few panicky "Maybe I can't do this?!" moments, it didn't actually seem all that difficult. I finished the 30-minute puzzle in under 13 minutes, which would have given me a score of 1595, which would have tied for the 13th highest score on that puzzle, two minutes faster than two of the four Mid-Atlantic solvers who beat me overall this year (Doug Hoylman and Erik Agard). You will note that this is all in the conditional tense. Because I had an error. Oh, not the error I thought I might have made, where I wrote "SUR" instead of "SNO for one answer" -- that one, I fixed when I entered the crossing answers. No, this was in the lower right quadrant, where I had "SLANDER" as the answer for "Slight." Which is a perfectly good answer. Or would be, if the "A" in "SLANDER" wasn't supposed to be the "E" in "DIE" ("It can be loaded"). I tried to make the case that "DIA" (the graph drawing software) can also be loaded, which would make it a clean solve, but they weren't buying it. (What, not everybody uses Dia?!) So instead of a beautiful 35th, I only pulled myself up to 49th. (Yes, that's right, with an error on P5, I still moved UP from 62nd to 49th. That's just how hard P5 was, and just how fast my time was.)

Puzzle 7 neither helped nor hurt me -- under 9 minutes on a 30-minute puzzle, fast enough to be in the top 40 or so solvers but not fast enough to make a serious difference. I think I was still around 50th but didn't write down my standing that morning.

Sunday morning, of course, is the final puzzle. Which is huge (a 45-minute Sunday-size puzzle), but never terribly hard. This year, it was terribly easy. So easy that I was really just limited by how fast I could write, and how much stamina I had to stay focused. I finished in under 10 minutes (I'm not making this up!), slower than only 14 other solvers. (The fastest solver, Dan Feyer, finished in 6 minutes!)

Here's the thing, though -- when I went to check my final standings later in the morning, I was 110th, with a score of 1550 (insteadof 2425!!) on Puzzle 7. Clearly an error had been made -- it turned out they had scanned somebody else's puzzle (it's still wrong -- you can see that the handwriting isn't mine if you look at the scan: ),and it must have been somebody who had no time bonus (that score differential is exactly equal to 35 minutes of bonus time). (I was told later that actually the problem was that my bonus time was accidentally entered into the missing-letters box.)

So I emailed Will Shortz on the tournament page, explained the situation, and got an email saying yes, my solve was clean and I had 35 minutes remaining, so they would change it. At which point I went to check the standings to see if they'd already updated it -- and my score on P7 had dropped, to 1050, which put me in 154th place! It finally got sorted out on Thursday and the correct results have been posted:

  • Puzzle scores: P1 - 1165, P2 - 1575, P3 - 1485, P4 - 1330, P5 - 1400, P6 - 1895, P7 - 2425
  • Overall score: 11,275
  • Overall ranking: 44th
  • B division ranking: 22nd (B division includes all players who haven't been in the A or B finals in the last 7 years)
  • Mid-Atlantic ranking: 5th (top female solver in the Mid-Atlantic)
  • Rank among female solvers: 9th

I also did an Alternate Reality Analysis, a world in which I didn't make those two stupid mistakes on Puzzles 3 and 5, just for yuks. Of course, this assumes that everybody else still made their mistakes!

  • Adjusted puzzle scores: P3 - 1680, P5 - 1595
  • Overall score: 11,665
  • Overall ranking: 21st
  • B division ranking: 5th (just two minutes shy of the B finals!)
  • Mid-Atlantic ranking: 4th
  • Ranking among female solvers: 4th

Which is all kinda cool, even if it's not reality. (Oh, and I would have won the 70s and Seniors (80s+) divisions with those scores, so if I can just keep it up for another 25 years, I'm golden!)

The interesting thing is that I felt as though I was wasting a fair amount of time on a lot of the puzzles, jumping around a bit inefficiently. So I believe I have the capacity to get faster. The question is whether I have the capacity to stop making stupid mistakes, which is much less certain!!

A couple of people have asked me how I prepared for the tournament, and how I managed to improve so much (the first year I went, I was ranked 145th, just shy of the top 20% -- that's great for a rookie but still much lower than the top 7.5% ranking of this year). Here's my advice for those who want to go into training:

  • Do lots of crossword puzzles, of all levels of difficulty.
  • You don't always have to race the clock, but it's good to spend some time regularly solving as fast as you can. I solve the NYT puzzle online so I can gauge my performance -- your relative rank in this community is reasonably well correlated with what your rank would be at ACPT.
  • Don't ever give up on a puzzle. I used to peek in the back when I couldn't get an answer, figuring I'd learn a new thing.
  • That habit trained me into psychologically being unprepared for a situation where I couldn't peek. Once I forced myself to stop doing this, I was surprised at how often I could break what seemed to be an impossible logjam. Sometimes I set a puzzle aside and come back to it, and then bam! I can solve it.
  • If you are really stuck, force yourself to make your best possible guess -- what would you write if it were the tournament? -- and actually write it down. Commit! Then, and only then, can you look to see if you guessed right. Whether you did or not, go look up that bit of knowledge on the Internet, so that you can cement it into your head.
  • Don't ever do the previous step (commit then check) until you've finished the entire rest of the puzzle. Often there will be some tricky bit of theme or wordplay elsewhere in the puzzle that will let you break through in a place where you thought you were stuck. So train yourself not to give up prematurely.
  • Be sure you do puzzles on paper, not just online, since the tournament is on paper. I'm actually much faster on paper, but I know a number of strong solvers who are faster online, and have heard them say they need to spend more time training on paper puzzles.
  • And of course, it's worth re-emphasizing: do lots of crossword puzzles, of all difficulty.

Some things I don't do, but I could imagine would be useful:

  • Learn who the key crossword puzzle constructors are, and get to know their style by solving lots of their puzzles. (I have a hard time actually paying attention to who constructed what puzzle, so this is a bit hopeless for me.)
  • Have a daily routine of doing lots of newspaper puzzles (I only do the NYT and sometimes the Washington Post because I subscribe; the other puzzles I do are almost all in books, though I sometimes do Brendan Emmett Quigley's puzzles, and always try to do Matt Gaffney's competition puzzle -- though I'm hopeless at the metas).
  • Read the almanac, dictionary, and encyclopedia for fun.