2007 Mystery Hunt

[tahnan]

I haven't written anything about this year's Mystery Hunt. Also, I don't intend to. A lot of other people have probably written about this year's Hunt, but I fell behind on LJ in the last week, and so I haven't read those things. Also, I don't intend to.

I like post-mortems as much as the next guy; no, probably more. But somehow, while I've enjoyed discussing things with my team, I don't feel any particular need to go through the typical recriminations and praises and so forth.

I did read the posts, and comments, on thedan's blog. I didn't reply to any of them, though, tempted as I was. He talked about, and people replied to, thoughts about what he called "sweatshop teams" who, when he visited, looked more focused on solving than socializing. Noah has thoughts about "pure" vs. "shell" metas, about whether the Round VIII Meta was broken, and so forth, and I by and large won't comment on those, either. I will say that he talks about the 2003 hunt having "shell" metas, metapuzzles in which you put the words into a framework rather than just using them as words; but that by my count, there were only two rounds of seven where the answers were used as transformations on a grid (I'm not counting Round 2; we could have left out the grid, but that would have only made it harder). (I'll also note with amusement that he complains that, for the Mystery Hunt, "starting out with a round of pop-culture and a round of sports was somewhat inappropriate"; whereas many Hunters complain that having too many crossword-and-wordplay puzzles is inappropriate.)

The discussion about whether the Round VIII Meta is "broken" seems to be typified by two opinions:

• "Anyway, bottom line, two different teams solved the puzzle without any hinting. That's the definition of solvable, and while it could have been calibrated better, that's good enough for me." -Dan
• "Any puzzle that stumps so many smart people for so long is probably broken, even if it is ultimately solveable." -Noah

I think they're both right, in a sense, but also somewhat misguided.

It's possible to solve a broken puzzle. Heck, I've solved broken puzzles. Being able to get to a solution does not in and of itself make a puzzle broken. It's also possible for a puzzle to be hard, even perhaps universally hard, without being "broken".

I don't think the Round VIII Meta was broken. I also don't think the fact that (a few) teams solved it should be taken to indicate that it was somehow "all right". I do think it was terribly underclued (which I think Dan acknowledges in his post). I think it ultimately had the problem which Bridget called the "Taipei Problem" when we were writing the 2003 Hunt, based on this puzzle. Teams solved that puzzle; that wasn't the problem. The Taipei Problem is this: the right thing to do is reasonable, but there are hundreds of other perfectly reasonable-looking wrong things to do. That, I feel, is what happened here: using the Senate floor as a grid is reasonable, but so is listing the senators alphabetically, or by seniority, or...

Anyway.

Comments

[toonhead-npl.livejournal.com]

You probably don't want to go on about Hunt stuff anymore, but if you do, I'd be interested in hearing your opinion about that final meta that stumped everyone forever in 2005.

[tahnan.livejournal.com]

Unfortunately, we were so far behind on that hunt that we never really looked at that, so I don't have any first-hand experience with that one. My understanding was that Setec, faced with everyone failing to solve that meta, wailed, "But it fell quickly in testsolving!" And that's always a danger: your testsolver might happen to get an insight faster than any solvers.

I think in that case, the puzzle itself was perfectly fine, and the only problem was that it wasn't as easy to in fact find it as they intended. Which is also a problem, but of a different sort.

[toonhead-npl.livejournal.com]

My yardstick is just whether or not the answer makes me smack my forehead and say "Duh! How could we not see that?" Or even when you happen upon the possibility and say "this must be it" as you begin to try it out.

For the 2005 Hunt, no argument: it was completely fair and right in front of your face as long as you were paying attention. This year, less so since it was tough, but it didn't feel completely unfair. I sure would be curious about how many different scenarios were tried out by other teams.

I know I never would have come up with the answer on my own.

[rubrick.livejournal.com]

This exactly echoes #1 of my puzzle-constructing commandments (which I went on about at some length during the "perfect puzzles" talk at TexSAcon): "If the solver is on the right track, they should know it."

I think a very good gauge of whether a "what do I do?" puzzle hits the sweet spot is whether, when you finally get it, you want to smack yourself in the head or smack the constructor in the head.

[tahnan.livejournal.com]

That's a good commandment, though it's somewhat vague. At which point does the solver need to know that (s)he is on the right track? And: how knowledgeable/intelligent a solver? If the solver gets "FUTH..." and gives up because it's not forming a word, has the constructor failed, or the solver? If the solver says, "Maybe these are all quotes from 'The Satanic Verses'" (and is right) but doesn't see any on a casual skim of the book, again, who has failed?

Vagueness isn't bad. But it's important to recgonize.

[rubrick.livejournal.com]

If the solver says, "Maybe these are all quotes from 'The Satanic Verses'" (and is right) but doesn't see any on a casual skim of the book, again, who has failed?

Salman Rushdie.

[dr_whom]

I quite disagree about the orange-star meta in the 2005 Hunt, actually. It wasn't "right in front of your face" for the following reason: the content of the early parts of a Hunt sets up criteria for what's reasonable to expect in the later parts of a Hunt; the further into a Hunt some structural fact remains true, the more likely that it holds for the entire Hunt. (This is why when, in the SPIES Hunt, the fifth meta was the first to depend for ordering on the layout of puzzles on the map, we had one of the earlier agents warn teams that there were metas that worked that way.) In the case of Normalville, the orange star meta was the twelfth of twelve metas, and the first eleven had all been what Noah calls "pure" metas, with not a "shell" meta in the bunch. But the twelfth meta turned out to be not only a shell meta, but a shell meta whose shell status was at best very obscurely indicated, as the shell itself was provided separately from the orange star puzzles, with only sparse and oblique clues to indicate the connection. (Compare all the shell metas from this year, from 2001, etc., where the "shell" data needed to solve the metas were provided explicitly as part of the round the meta belonged to.)

In other words, the orange star meta was flawed not because it was a broken puzzle on its own account, but because as a puzzle it broke the rules that the Normalville Hunt had apparently set for itself internally.

[tablesaw]

Actually, the Yellow and Green "supermetas" used the map as the basis for shell metas.

[foggyb.livejournal.com]

I've talked about this with others, but the two obstacles I had faced with the orange meta was using a Mac and Internet Explorer at the time, which surprisingly did not support the necessary tagging to reveal the captions. The other was that early descriptions of what to do when you've got all six parts of your costume implied something about a graveyard. Since the necessary part of the orange meta was in black, and kind of graveyardish, that part of the map seemed more necessary in the endgame.

[(Anonymous)]

On top of this technical issue, there was also the unfortunate choice to have one of the names in the graveyard clue "red herring."

Very subtle problems all of them, and things that probably never came up in testsolving (because its hard to imagine that such a small team could have testsolvers who didn't know that the graveryard wasn't used for anything else), but surprisingly big problems in practice.

--noah

[dr_whom]

Ah! Yes, that's true. What I should have said was, none of the metas before the orange star employed any data other than (a) the puzzle answers and (b) the positions of the puzzles on the map. And the pattern of using the puzzles' positions on the map was established early; the yellow and blue dot metas both used the map position of their puzzles. What the orange star meta did was require data from something (apparently) completely unrelated.

The technical issue cited by foggyb didn't help much either, of course.

[(Anonymous)]

There was one very subtle exception to the rule which Aaron is explaining, that is that for the green supermeta the shape of the roads was important. However, this is only a very small departure from Aaron's rule, and it was clued very strongly by the fact that nowhere else on the map are their circular roads.

--noah

[dr_whom]

...The shape of the roads that the green puzzles were located on. If the orange puzzles had been located in such a way as to suggest an association with the "graveyard", even that could have been okay.

[thedan.livejournal.com]

I think this (the discussion across all the blogs) has become a silly syntax debate about what "broken" means. In my enigmatological lexicon, a puzzle can be very flawed without being broken. Broken implies something severely wrong, something that either can't be solved at all, or can't be solved without guessing, backsolving, or otherwise working outside the "rules" of puzzles. A puzzle littered with factual errors is broken, even if you can determine the final phrase by guessing which entries have mistakes. A logic puzzle with a nonunique solution is broken, even if it has two solutions and you can see which one gives a final answer. And a puzzle with no intended path for the solver to reach the a-ha is broken.

The right thing to do in Meta VIII was clued, but not as well as it should have been. That's a nontrivial flaw, and if Mike's line in the video about seats were more isolated or indicated, the puzzle would be significantly improved. However, multiple teams managed to find the intended path of reasoning and correctly apply it, without having to account for a mistake by the constructors. Thus, by my definition, the meta is flawed, but not broken.

Basically, I think calling the meta "broken" implies that it ruined the Hunt. I don't believe it did. However, your mileage (and definition of broken) may vary.

[tahnan.livejournal.com]

Semantics, not syntax. But anyway, yes, it does depend on the definition of "broken".

[thedan.livejournal.com]

I initially wrote semantics and then changed it because I thought syntax was the meaning of a single word. But you know better than I do.

[(Anonymous)]

Yes, the "Taipei Problem" is exactly what I meant, and is definitely a more accurate description than broken or non-broken. I'd seen someone (maybe you?) mention that term when i was searching after hunt last year, but I'd forgotten about it. I don't care whether a puzzle with the "Taipei Problem" is called broken or not, as long as we all agree that it's bad and not just "I didn't know that broken was a synonym for too hard for me to solve." (Not to pick on Tyler, whose recent Onion puzzles have been delightful.)

--noah

[(Anonymous)]

Oh, and a good name for the other kind of problem that you mention of a puzzle that is universally hard without being "broken" is the "American Championship" problem. The original version of that puzzle (without the commentary column) was totally fair: the aha was easy, it confirmed well (the pattern in the first player's answers), and it had a unique solution (checked by computer). Nonetheless it was way way way too hard, verging on unsolveable.

--noah

[dr_whom]

I'm not sure "'American Championship' problem" makes that much sense, since we after all caught the problem on "American Championship" and (I hope!) fixed it before the Hunt. Maybe "'Cock Conundrum' problem" would be better....

[(Anonymous)]

Ooh, right, better example. [looks sheepish]

--noah

[dr_whom]

I'd hoped to chat with you about the Hunt on the train, but I'm glad to see you at least seem to have gotten back to Philadelphia safely. I'd be interested in hearing your opinions later, though, if we find time.

[tahnan.livejournal.com]

We should find time. I mean, how busy can I possibly be these days...?

[jenaflynn.livejournal.com]

A couple of people at work were talking about this today during lunch. Not sure which team they were on, but they mentioned that they had a blast doing it.

God, this town has some neato stuff.

[tahnan.livejournal.com]

Yeah, we like it. :-)

If you find out what team they were on, let me know, 'cause now I'm curious. Not knowing, after all, where you work.

[jenaflynn.livejournal.com]

GE Healthcare. It's in Lawrence, although I think one or two of the people talking bout it might have been from the GE Downtown Boston location. (We trade employees with them quite frequently.) One of the fellows is French Canadian.. I do believe Mico is his name. Not sure about the rest. (I'm new, and so terrible with learning my co-workers names)