21 January 2009

Time for my Mystery Hunt review. I'll discuss individual puzzles in a future post. This year's hunt, Escape from Zyzzlvaria, was put on by the Evil Midnight Bombers What Bomb at Midnight who also wrote the hunt from two year's ago. It was fun, beautiful, and very ambitious. I subscribe to the Setec opinion that your first hunt should be short and clean, and your second hunt should push the envelope. By that metric the Bombers have been a great success. Furthermore this hunt fixed many of the issues that made my team not really enjoy their first hunt (lack of a feeling of progress, too much data mining, not enough puzzles available ever, and no pure metas). Although it went long, we had a lot of fun almost the whole time. Thanks a bunch to the Bombers for putting on a great show.

It's good to have an overambitious hunt once in a while that can raise the bar a bit and have lots of new ideas for other teams to process in their later hunts. Monopoly, Matrix, Time Bandits, and this hunt are the recent ones that had lots of new ideas. Of those Monopoly was perfect, while the other three wold have been improved by dialing it back a bit. It could rightfully be said that most of the structure of our SPIES hunt was just taking a third of the ideas from Time Bandits and doing them well. Of these I think Zyzzlvaria will be remembered as being the most innovative, and its impressive that they managed to be that overambitious while still ending up with an good product (sorry Time Bandits).

The dollarbucks round opening mechanism was really really good. Other teams in the future should just steal it. Really. The first round metas were beautiful, and by using round titles as a clue for each they allowed themselves more flexibility to use ideas that would otherwise have been too difficult (for example, Hiigari). I liked the idea of having a first half with pure metas and a second half with shell metas (although, see rant below). Having some pure metas is important to me as Mystery Hunt is almost the only time I get to work on them, but I'm happy with 5 good ones and shell metas allow for more flexibility. The Bombers used this flexibility very well. The production values raised the bar a lot (an actual game for each of us!). Finally the second half of this hunt had a greater number of interesting structural innovations than the last three hunts combined. It's really fun to figure out how different hunt structures work, and one of the downsides of mystery hunt relative to similar events is that there's often only one structure to figure out which leads to a low puzzle to structure ratio.

The puzzles were hard but fair (but hard!). It seemed to me that there were many fewer easy puzzles than in the Normalville/SPIES/Hell era. I was happy that this year seemed to have less data mining than Hell, but on the other hand it seemed that there were more puzzles where the answer extraction was difficult. I firmly believe that answer extraction should be either interesting or easy. Many puzzles (the hitchhiker's puzzle comes to mind) had fair gettable answer extractions that were still difficult to find and detracted from the overall quality of the puzzle. Nonetheless the puzzles on the whole were excellent. I'll have some shoutouts on particular ones after I get to go through the puzzles I didn't see.

Now for the rant... The second round metas... I joked at some point on Sunday that after everyone complained about the Senate meta the Bombers decided "Well maybe if we just make all our metas like this people will stop complaining and just figure that's how metas are supposed to work." Of the 7 metas we only solved Lazyr Zone. Nonetheless we had also solved Orbital Nexus modulo reading the clue phrase, and for both Astro Jail and Harvoid it was our fault not the puzzle's fault (we were close on Astro Jail but missed the connection with the cards even though we figured this meta used the cards, and on Harvoid we misplaced the box and forgot about it). The remaining three all had major major issues. Of the 4.5 point people from our hunt (Reid, Roger, Aaron, Andrew, and half me) I can't imagine any of us approving any of these three puzzle ideas without modifications. I'm still baffled by how they got in a hunt that otherwise was of such high quality. A checkers puzzle where not all the pieces are on the same colored squares?? Really? How could this possibly be a good idea?

Sometimes in these post hunt discussions I feel like some other people have a puzzle aesthetic that's totally foreign to that of my team. I can already anticipate someone commenting "Yes but the switching colors was clued! So it's fine." I don't care if it was clued, it's still ridiculous. The key property of checkers is that it only takes place on one color, a puzzle that ignores this is a bad puzzle. When you think of checkers as an option the first check about whether it's a sensible theory is to check if the pieces are all on one square.

On Combat Zone, I essentially solved the meta on my walk in on Sunday (my one good idea all hunt, I was mostly off my game), I showed up with a complete theory of how it worked that turned out to be exactly right. Except that the two obvious checks for whether my theory was right failed (the mapping wasn't well-defined because some of the reversed letters occurred multiple times on the same dice, and the ominoes didn't fit into a rectangle). We still tried it, but when we couldn't make out any words (we never had more than 7 puzzles) we assumed that we must be missing something (something in the names of the gods? the "God given order" given by the order of the gods? the pairing between puzzles and shapes?). Any small confirmation and we'd have solved this puzzle. Why not make it a rectangle? Why not make the mapping well-defined? Why not mark the boundary edges of the final shape? Why not put the gods in the same order as the puzzles? Sure the puzzle was solvable as is, but why not make it better?

The Virtual Sectors meta I just don't even want to talk about.

Our team's aesthetic is very much based on Setec's hunts and on the advice they sent us about puzzle writing. These three meta puzzles all fail badly: Was there only one aha? No. Were the steps obviously right in retrospect? No. Is the puzzle solvable without psychoanalyzing weird flavor text? No for two of three.

Again, other than these three puzzles (which didn't spoil our enjoyment of the hunt that much) it was an excellent fun hunt. But I'm really curious about how the second round meta debacles could have happened and how future teams can avoid it. Especially because in my mind it was only this issue that stopped this hunt from being in the argument for best hunt ever.


David Speyer said...

Isn't "debacle" awfully strong? The metas violated my aesthetic standards, but there were 4-6 teams that got each of them. I think we need to focus on how to improve our meta solving, so that we can deal with metas that incorporate multiple steps and outside data.

In this regard, it seems to me that we have two big weakness as a meta-solving team, and this hunt nailed us on both of them.

(1) We tend to focus very much on the answers, and not on the other input data we have. For example, we copied the Wheel of Death onto the white board rather than playing with the physical wheel. I taped a scan of the wheel to the white board, but very few people looked at it. That's why we got SPRMNSDVRSRLMP and not S*P*RM*NS*DV*RS*R**L*MP.

(2) We aren't willing to keep working on an approach that seems ugly. When we figured out how the dice meta worked, we all started brainstorming the most elegant shape for the pieces to form. (A 4x13 grid! A tetris board, clearing rows as they are filled!) Not enough people just started trying to fit the pieces together and see what happened.

Now, I agree with you that I prefer metas that do not use outside data, and are extremely elegant in structure. But, since other teams got these puzzles, I think we should view this as a weakness in our own puzzling style, not in the Bombers' writing style.

Noah said...

I certainly agree that our team needs to get better at pushing through on good ideas that get messy in implementation. For example, on the Thin Red Line I think we found the right extraction very early on but didn't recognize LIVYORH as the start of a clue phrase. Rather than looking for new ideas we should prioritize checking more thoroughly the ideas that we do have.

Nonetheless I don't think debacle is too strong a word. The total amount of time teams spent stuck on these metas is really really large. And given the importance of metas the number of people-hours is just enormous. If they'd fixed these metas the hunt would have ended 12 hours earlier, perfectly on time, and we'd be arguing about how it stacked up against Monopoly.

AJD said...

I'm not sure what it is you don't want to talk about about the Virtual Sectors meta.... As David alludes to above, and Jen mentions elsewhere, the first step of the puzzle is a lot fairer if you use the wheel more directly as the object containing the letters. Obviously I don't know if that would make the answer seem more robust if we had tried it, or if we'd have been more willing to consider a string of consonants as "the right track" if we'd had specific spaces to leave for the vowels. But this part may just have been a failure of imagination on our part on what to treet as primary—the wheel, rather than the answers. I don't know, though: taking the letter after the key vowel in the wheel word, rather than any other letter extraction method, still isn't very well indicated without, as you put it "psychoanalyzing weird flavor text"; and there had already been an 'extract consonants and add vowels' meta in the Hunt. So it still strikes me as poorly thought out.

Looking at the solution to this meta now, though, the thing that bugs me about it the most is how many steps it has. It seems way out of keeping with the rest of the Hunt—once you've extracted the answer the way they want you to, you call it in and then... they give you another puzzle to solve? With another unclued letter extraction? (Seriously: converting colors into numbers?) They couldn't have just had the answer to the meta be the thing you extract from the wheel-o-death spaces?