Too Big to Solve?

[motris]

Not my tagline, but a good description for the Mystery Hunt that just happened. One line of dialogue after last year's Hunt that I led with in my wrap-up was a question of when is too soon for a Hunt to end. I said, in this era of a few competitive teams trying to grow to get over the winning hurdle, constructors aiming bigger was a mistake. The

Comments

[zotmeister]

I wanted to remark on the seeming change in attitude you present between this entry and your previous "that happened" ones, but it occurs to me that this is quite possibly entirely the result of my having read this one in YOUR voice rather than my own. - ZM

[motris]

I'm slowly getting the entries shorter so that I can just tweet "so XXX happened" in the future with a link and not rant for paragraphs again. But the phrase is certainly a loaded part of my lexicon, after WSCs and Philly Sudoku tournaments and such past.

[ext_209009]

I'm not expecting a whole lot of positive reviews given that the difficulty fell so far outside of our expectations. I hope Portals was enjoyable at least - of the half a dozen puzzles I made, that's probably the only one I was proud of. But now that things are over I also realize it was probably part of the length problem.

I can't really say anything about how the length debacle happened since even now I still only vaguely know about how one of our supermetas worked, and I only saw a few dozen non-metapuzzles in all. Conjectures from the organizing end about how we were so off-target will have to come from someone else.

[motris]

Portals fits the space of great puzzle, but great Hunt puzzle?!? as I felt about the Kakuro from last year. Yes, it was certainly a very impressive feat of puzzle construction and had many of the same joys of your last marathon from your blog. But I saw no easy way to parallelize the process, and no easy way after 3 hours of getting to an almost answer that incorrectly copied the wrong snake shape into the Statue Park and then built off of it, to hand off or get help from anyone else except to dig myself out of my own hole. Once I got the Statue Park to "feel" right to constrain the lower-right corner, I knew I'd get to the end soon, but I was not expecting to sign up for a six hour puzzle experience.

[ext_209009]

Actually I deliberately constructed it be parallelizable. There were about 10 break-ins to that puzzle and they did not merge for a long time. Guess that didn't work out so well in practice?

The inability to correct mistakes is certainly a problem though. In hindsight I might have fixed that by finding an extraction mechanism that could let you self-correct or self-confirm your progress.

[motris]

My comment on not splitting may also have been affected by two other things: 1) we got this ~1-2 AM and people I could have worked with were asleep and 2) I am paranoid about working on such a big task and having mistakes pop in not of my own making.

Compared to the ~10 independent path puzzles or the 8 independent sudoku, or several hunts ago my 8 unknown Kakuro variants with Dan, that each split into a "you do this page, I do this page" kind of thing, having 10 different break-ins does not help if it means pencils are crossing across puzzles to use them or track progress on two sets of sheets. So splittable I guess, but not 100% in practice to split only other routes.

[anonymous]

In all honesty, I was looking forward to one of Palmer's puzzles on the hunt, and when I saw Portals, I knew it was going to be good. Not that I made a lot of progress at all on it, but now that the hunt is over, I can solve it my way

[anonymous]

I certainly have some sympathy for what you must have gone through. We obviously failed spectacularly at estimating difficulty/length. We also had an initial plan to keep the number of open puzzles reasonable, to try to avoid favoring large teams; you probably noticed that on Friday. But by Saturday, *everyone* was behind the time curve (we simply had to release puzzles at a certain rate so they would all be out by Sunday), and with 30+ puzzles open, large teams had an obvious advantage. My impression was that the less serious teams loved having a vast array of puzzles to pick and choose from. But it just wasn't a good competitive Hunt, given the crazy amount of hints and "options" (free puzzles) we had to give out in a desperate (failed) attempt to end the Hunt by Monday morning

[motris]

There are many valid interpretations of the flavor including don't cut out. The meta clearly presented snakes with the letters in the answer words and the concept of knotting and just seemed to obfuscate snake identity to be mean. One can imagine a single ouroboros assembled in one loop so that going head to tail around it all the letters can be crossed off in order uniquely in the answer words leaving the leftovers. That would have a nice word and logic component to match the other parts. It was a much more fun idea that almost was fully constrained. One can imagine knitting a head to tail snake group like a potholder to spell out all answer words with overs and unders and edge leftovers read cw or ccw for extract. Certainly one can even with sleep read not cut out to mean not cut out because those are the words in a row.

I will hate all Hunt construction teams that have any contingency that needs to unlock a round of 20(+6) puzzles all at once to a competive team of any size as happened to my team at midnight to 4 AM with Rubik.

[anonymous]

Wait, how can you make a giant Ouroboros loop without cutting anything out? (And where would the "1/2" markers fit in, or were you just ignoring them?) I don't understand your "pot holder" description at all. I was the author, btw, so I probably shouldn't try to defend this universally hated puzzle. I just thought disassembling one set of snakes to form another set of snakes was neat. Sigh.

Incidentally, I just read Eric Berlin's writeup ( http://ericberlin.com/?p=5228 ). Palindrome also considered themselves "stuck" on the three supermetas, which is interesting because I think they were by far the fastest to get them (relative to when they'd solved enough metas to have a chance). They really caught fire once they got serious on Sunday. They were the only ones to solve Rubik, and [Atlas Shrugged] were the only ones to solve Indiana Jones.

If you want some cheering up, Luck was the first to solve a meta in the last three rounds (the Snake logic puzzle).

- Derek

[ericberlin]

I'm pretty sure our experience with those three supermetas was the very definition of "stuck," regardless of whether or not we got ourselves unstuck faster than other teams.

[motris]

The answer I've heard for years, when solving or constructing, is that tradition is not to confirm partials as is done at Microsoft Hunts or Google or elsewhere.

I've wanted an automated system that confirms partials (and maybe pushes "keep going" or eventually hints) since we ran 2009's Zyzzlvaria Hunt. I had no traction with my last team, but if I ever have my own team and win I'll modernize many aspects of Hunt including partial confirmations.

[zotmeister]

As I suggested during the Hunt (and you apparently responded to, although at the time I wasn't able to catch your entire response), the three distinct portions of every puzzle (realization, execution, extraction) should be separately hintable. Using "magic bullets" (free answers) is a last-resort failsafe that in theory should only be needed if a puzzle is broken in the first place... which of course means it should NEVER be needed, rather only a strategic option to be employed after being earned. Two or rarely three bullets should be far more than enough for an entire Hunt. When I left Sunday night, we'd already fired two full chambers of magic bullets, and Manic Sages kept reloading the revolver. What started as a Hunt became an absolute farce; forget what other teams thought of the Hunt, the Sages themselves clearly gave up on their own creation. Quod erat demonstrandum