Other people wrote some puzzles

[dr_whom]

This is the third in my series of post-Hunt wrapup posts. In my last post, I discussed the puzzles I wrote; but since I was lead puzzle editor, I had the privilege of witnessing the entire writing and testsolving process on the puzzles I didn't write, as well. Here I discuss some of those puzzles I didn't write, about which I have some comments or anecdotes to share.

• English Expectations. Although I'm (correctly) not credited as an author on this one, I see the authors did give me a "with a bunch of help from" credit—mainly because I was able to get translations into Arabic, Czech, German, Hebrew, and Italian from my vast network of linguist contacts. I'm a much bigger fan of thematic flavortext than a lot of my teammates are, and I was pushing for something along the lines of how when Mario climbs the staircase at the end of a level he feels like the king of the world, or something (thus alluding both to the answer JACK DAWSON and to flags, which are what's located after the end-of-level stairs in Super Mario Bros.), but it worked fine without the flavortext anyway.
  
• Pattern Recognition. This spent a lot of time in testsolving with testers jumping to the conclusion that each group was supposed to contain only one pattern and then reporting the other patterns they saw as red herrings that the authors should eliminate! Did the same thing happen in the Hunt itself?
  
• Two Heads Are Better Than One. As I noted elsewhere, okosut wrote the first puzzle to get testsolved in both this Hunt and SPIES. This time around, it was this puzzle. (In SPIES, it was this one.)
  
• Hammer Time. As I noted elsewhere, I think one of our mistakes in writing this Hunt was that we spent way too much time writing metas. However, that did give us a stock of ready-made puzzle concepts that we could plug in as regular puzzles. This is one of them; others include One More Try, Famous Faces, Puzzle Box, and of course four of the five subpuzzles of Meta testing! Hammer Time dates back to an attempt to base the three Mario metas around SUPER, MARIO, and BROTHERS instead of Mushroom, Flower, and Star; later, this puzzle was part of an intriguing but abortive attempt to have each puzzle in the Mario world themed around a particular Mario bad guy. We were very lucky with the answer HANK AARON; originally we were thinking of assigning this answer to one of our sports puzzles, but when we realized that Hank Aaron is, so to speak, a Hammer Brother himself, we assigned it here. The set of letters we had available for answer extraction was pretty constrained, but I think we got an okay and thematic cluephrase out of them anyhow.
  
• Flat Head. One of the most controversial puzzles among the editors. The editors who attempted to testsolve it completely bounced off it and argued strenuously for cutting it; when it went in anyway (due to strong endorsement from another testsolver), one predicted it would be the most hated puzzle in the Hunt. It seems to have turned out well-liked, which is somewhat gratifying.
  
• Timbales. Ricky was our team's power testsolver; he'd rip through puzzles in an hour and a half that other testers remained stumped on for days. As a result of this, I was kind of expecting him to write mostly hard-core puzzly puzzles, and he did write a couple of those in Mountain Pass and Nik-holey; but as it turned out (and as he noted elsewhere) most of his puzzles are fairly lightweight puzzles on humorous topics—not only did he write the puzzle on Latin puns, but also the ones on Garfield, My Little Pony, -er jokes, and so on. Ricky also seems to be one of the only other members of Metaphysical who favors rich thematic flavortext cluing, although Timbales ended up not really using any since a testsolver found that in this case it made the puzzle too easy. Anyhow, we got really lucky on this puzzle: the Bebop Man meta we ended up using required a ten-letter answer ending in LATI, and the best we could do for that was either PEDICULATI or ACTA PILATI, either of which is a terrible puzzle answer. And then this puzzle appeared for which an awkward Latin phrase would be a perfectly plausible answer! It's no wonder that many testsolvers (and solvers) thought that the answer inspired the puzzle, but in fact they were completely independent! Overall I think we did a pretty good job finding homes for some of our more awkward puzzle answers, including not only this but also FOREIGN DISASTER RELIEF, SYMPATHECTOMY, and to some extent even MULUGETA WENDIMU.
  
• Stuff Nerd People Like. Our backup puzzle, stuck into the Hunt when Funny Farm crashed our server. I'm still hoping to get Funny Farm posted in the Hunt archive anyway in an "outtakes" section. Anyhow, my other reason for mentioning it here is to point out that redcat9 is in fact continuing the Stuff Nerd People Like blog as an independent project post-Hunt; the first non–puzzle-related post is already up.
  
• Life of the Party. My only comment here is that I think it's amusing that both this Hunt and SPIES had a puzzle about unusual manners of death.
  
• Pesky Bugs. So, I really wanted this to be an audio perception puzzle rather than a wav-file manipulation puzzle. The original version of the puzzle had the six "bugs" in six different sound files, and I was hoping people would solve it by just listening to the six of them and figuring out their flight paths by ear rather than by calculation. Unfortunately (so to speak) a testsolver was able to solve the puzzle in five minutes by doing that, so we resorted to this version, for which solving by ear is much less plausible.
  
• The Bio Man meta. Hoo boy, this year's Orange Star meta: as a result of differences between the testsolving and Hunt solving circumstances, this meta ended up broken, and no team (I believe) forward-solved it during the Hunt. In first-pass testsolving, the virus diagram looked like this: the tail fibers were a different color than the rest of the diagram, the five rows were more or less the same height, the interior was white, the heavy lines separating rows were fairly clear—and the diagram overall was pretty hideous. It was redrawn to look a little prettier for the Hunt itself, but that redrawing lost a lot of the subtle cues that had helped testers figure out what to do with it. In fact, our blind testsolvers never solved it—but that wasn't really what we were looking for from the blind testsolvers, since they did solve the Mega Man supermeta, and we didn't require blindsolvers to solve all the puzzles in the limited time they had. (Indeed, the one saving grace of this broken meta was that it was not an absolute roadblock; it was still possible to finish the Hunt without solving it.) I did notice the problem a week or two before the Hunt and put it on my list of things to deal with, but somehow it slipped through the cracks and never got fixed; I take full responsibility and I apologize.
  
• N-tris. This puzzle has an odd history. callyperry originally proposed 'let's have a QR code puzzle—maybe knitting' and never really fleshed the idea out; ordinarily we wouldn't accept that sketchy a puzzle proposal, but I noticed we had the answer QUEUER in the Blackberry round and thought it was too perfect not to use. James and Mira wrote a knitting puzzle for it, but neither the authors nor testsolvers were really satisfied with it. Then a few weeks before the Hunt okosut proposed this completely different QR code puzzle that everyone was happy with, and it seemed all was well—until Manic Sages called in to say that they had found an unintentional alternate solution to the grid that wasn't a valid QR code (and hadn't been discovered in testsolving). okosut did a heroic job of revising the puzzle on the fly to eliminate the ambiguity; I salute him for it.
  
• The Minus Man meta. Possibly my favorite meta; both thematic and extremely streamlined. SW(AY)INGS was perhaps a little weak as an entry here, but the best alternative was OK(AY), which is really two spellings of the same word, not actual synonyms. I'm glad I was able to convince the meta committee that JOSE(PH) was a valid entry, though.
  
• Pointillisme. Like Pipe Dream 2, a puzzle that Hunt solvers had way more trouble with than testsolvers, for reasons I can't quite identify. Testers didn't find it easy, but they solved it relatively smoothly and many of them liked it a great deal; but Hunt teams by and large failed to find the aha. I'm not sure why. (The first round of testsolving used Google Street View images, which were replaced by photographs actually taken by Julian at each of the mural locations on the grounds that the latter were much more awesome to use; but it did get a second full testsolve after the replacement.) On another note, a couple of teams reported as an erratum that the numbers on the last PbN grid didn't match (i.e., the row numbers and column numbers didn't add up to the same total), but I think what was going on was just that the grid was just a little too wide, and one of the 1s on the right side of the grid ended up off the page when the puzzle was printed.
  
• The Zelda world structure. Teams seem to have had more difficulty getting a handle on how the three-answer puzzle system worked than we expected. Various commenters have said that their teams were put off in one way or another by, e.g., the different structures of the three-answer puzzles: in some you get all the answers from a single answer extraction; in some there are three different answer extractions on fundamentally the same puzzle; and in some there are in effect three different puzzles within a single data stream. This is in part because of disagreements among the editors with respect to which of these was the best model for a three-answer puzzle, and in part because those were just the ideas for three-answer puzzles that were proposed; but I liked the diversity because it allowed us to explore the space of possible three-answer puzzles a bit more than we could have if they'd all stuck to a single model. But apparently teams were thrown off by this to some extent, expecting one puzzle's three answers to bear the same structural relationship to each other that another puzzle's three answers did. In restrospect, this was also a failure of blindsolving: we didn't (by this point) ask blind testsolvers to solve individual puzzles, because we were focusing on getting their reactions to the world structure and metas overall; but in this case, solving the puzzles was necessary for getting a feel for the overall world structure. Even so, though, enough blindsolvers had written, or proposed to write, puzzles for the Zelda round that they might already have been somewhat spoiled on the way individual puzzles could be structured.
  
• Technological Crisis at Shikakuro Farms. Nikoli-style puzzles aren't really my thing, and when Jason produced a proof-of-concept mockup, it took me two days to solve it, but I definitely enjoyed this one: there are so many distinct interacting constraints that you have to keep changing the direction in which you're thinking about it to make progress. I kind of would have liked the two grids to produce IN and GRID rather than ING and RID, to play up the pun a bit more, but that probably would have led to red herrings (and been too crowded to construct, in the case of the second one).
  
• Amateur Hour. It was after this puzzle had already been testsolved when we realized that, although the puzzle extracts to ROCK BAND MOTD, the actual answer was going to be posted in the Rock Band 3 MOTD. The authors inserted another clue into the puzzle to produce a 3, but the only testsolver who saw it after that reasonably tried to extract the first letter and got ROCK BAND TMOTD instead. Since that didn't have the desired effect, I insisted that the authors just stick a big ol' 3 into the video at the appropriate place; they had plenty of other stuff to do at the time, but I think it helped the puzzle. It's also pretty hilarious that the answer extraction phrase in the Rock Band 3 MOTD was helpfully labeled "SECRET CLUE".
  
• Fascinating Kids. This was the Calumny Challenge of this year's Hunt: a puzzle which went into the Hunt despite not having a full solve from any testsolver (although editors and testing-admins agreed that there was nothing wrong with it and the only reason testers hadn't completed it was not having the patience to double-check their data), and then didn't get solved by any team in the Hunt—in part because the aha was hard, and in part because it was too late in the Hunt and in its round for solvers to need to or want to go back to it once they'd solved enough puzzles to get around it. That said, Hunt solvers got hung up on this earlier than testsolvers did—we did have at least one tester who figured out that writing grade level was what was important and didn't take that long to do so, and the only reason it wasn't solved was because of errors in word counting; I would have expected at least one solving team to have that idea. Oh well. But in any case, this puzzle and Calumny Challenge can serve as a cautionary tale.
  
• The Granary of Ur meta. I'm (correctly) not listed as an author on this one, but I did write the title and flavortext for it and I'm really proud of how that turned out. (In fact, I think I wrote all of the titles and most of the flavortexts for the Civ metas.) What I had to work from was "a Braille puzzle whose answer is THE FED which has to be the historically earliest of six Wonders of the World", which on the face of it are three elements that have nothing to do with each other. But when I thought of the "Granary" idea, the whole thing just sort of snapped together—both possible readings of THE FED are justified by the flavortext on "keeping track of the city's wealth and ensuring there was enough grain to feed the people", which is exactly what early writing systems were used for; and cuneiform seems like a pretty reasonable clue for Braille. I don't think there actually was a historically notable Granary of Ur (though I'm pretty sure Ur must have had a granary), but it sounds like a plausible name for a Wonder anyway.
  
• The Wall Street meta. This one went through a lot of proposed titles and flavortexts: Golden Gate Bridge ("suspended between two identical pillars"), United Nations ("prevent history from repeating itself"), and what we finally ended up with. Early rounds of testsolvers found this one really easy with the UN and Golden Gate flavortext; Seth was pushing for having this put in the Hunt with no flavortext at all, but we never got around to having it testsolved in that form. Anyhow, since this proved to be a pretty easy meta, I figured we could afford to give this one a complicated method for putting the puzzles in order (rather than the "by length of tech name"–type orderings used for the other Civ wonders). As a result of having read this solution file from the 2002 Hunt, I was worried about people mistaking the ordering puzzle for the actual metapuzzle, which is one of the reasons the ordering puzzle is on a separate page; that's also one of the reasons I wasn't that interested in getting rid of the flavortext, since without a box there for the flavortext it would look even more like the ordering puzzle was supposed to be the meta itself. I don't know how well this plan succeeded, though.
  
• The Civilization supermeta. This is the one that we were really worried about having a Normalville-style "orange star supermeta" problem, and not just because it was a "supermeta" whose answer is ORANGE. It had all the hallmarks of an orange-star meta problem: it went into the Hunt with the data organized in a different way than it had had in testsolving, and in fact blindsolvers never solved it at all. First-pass testers had basically been told "the Princess letter in this round tells you that the supermeta has no shell, but there's a picture of Mario dressed as Napoleon"—and those things were true in the Hunt as well, but of course there was no particular indication that "Mario dressed as Napoleon" was relevant to the meta in any way. Of course, this wasn't quite as bad as the orange-star meta situation, since the information whose relevance to the meta wasn't clear was merely a flavortext clue, without which the meta was still perfectly solvable; but the fact that no blindsolver had made more than trivial progress on this supermeta was a source of great angst to me. Eventually (by which I mean during the Hunt) okosut and I, over the objections of the author and various other people, decided that we had to take the blunt-instrument approach and just stick the word "French" into the Princess letter. (That version also hadn't been successfully blindsolved, but it was the best we could do.) I'm still pretty sure the meta would still have been solvable if we hadn't done that, but at some point you just have to take your testsolvers seriously; and I don't think the fact that we added "French" to the flavortext made the puzzle unwarrantedly easy, either.
  
• Losing My Nerve. This was one of like three or four independent proposals for a puzzle fitting the description of "Rubik's cube cryptogram". I guess it was just the time for that puzzle idea to happen? Anyhow, I thought the idea of using RUBIKQ as the six fixed-point letters was really cute, but the only available answer we had at the time that didn't have any of RUBIKQ in it was SYMPATHECTOMY—but then it turned out SYMPATHECTOMY was actually really appropriate to the puzzle!
  
• Unnatural Law. I really like this puzzle, but it was stuck in testsolving for a long time—people would notice the scientific laws, not notice the Constitutional amendments, and then get stuck, thinking they were stuck on answer extraction and not realizing that they just hadn't solved the puzzle yet. The way we eventually got the puzzle testsolved, of course, was by making sure it got showed to enough people that one of them spotted the amendments first.
  
• E Pluribus Unum. Originally this puzzle had cute paired clues for the grid entries, like the planets and RETURNER being "those who wander" and "one who comes back". It went all the way through testsolving in that form, and then in December, David said "I wonder if this puzzle would work with no clues at all?" And it did!
  
• Electronic Love. The other puzzle that no one solved. In this case, it wasn't so much because of problems with the puzzle or an unclued aha or anything—it's just that it was a mistake to put a long and work-intensive puzzle in the last round of the Hunt. I think we hadn't yet decided that Katamari was going to be the last round when we assigned it there? Anyhow, I feel kind of bad for not moving it once we realized that; it's a little unfair to the puzzle and its author.
  
• The Higher They Climb, the Farther They Fall. This puzzle was probably actually a lot easier in the actual Hunt than it was in testsolving. It was written and testsolved after (obviously) the elections this past November, but before the new Senate was sworn in at the beginning of January. What this meant was that during writing and testsolving, the full seniority list of the incoming Senate had to be calculated by hand. By the time of the Hunt itself, the new Senate had been sworn in and the full updated seniority list was available on Wikipedia, saving solvers a lot of effort.

Overall I think my personal favorite puzzles that I didn't write in this Hunt include the following: Squared Key, Pattern Recognition, Hammer Time, Timbales, N-tris, Good Times in the Casino, the Minus Man meta, Build Your Own Acrostic, The Crypt, A Modern Palimpsest, Technological Crisis at Shikakuro Farms, The Sport of Princesses, the Palace of Versailles meta, Losing My Nerve, and Unnatural Law.

Comments

[gwillen.livejournal.com]

"As a result of having read this solution file from the 2002 Hunt, I was worried about people mistaking the ordering puzzle for the actual metapuzzle, which is one of the reasons the ordering puzzle is on a separate page; that's also one of the reasons I wasn't that interested in getting rid of the flavortext, since without a box there for the flavortext it would look even more like the ordering puzzle was supposed to be the meta itself. I don't know how well this plan succeeded, though."

FWIW, when I looked at this (about ten seconds after it was released to us) it seemed absolutely blatantly obvious that this was not the metapuzzle, but a puzzle to give us the ordering. Possibly someone who hadn't stared at a lot of metas could disagree, but since the first two things I am accustomed to looking for on a meta are 1) how to order answers, and 2) how to extract letters, if the page text says "CLICK HERE TO SEE HOW TO ORDER ANSWERS" I'm going to take it at its word.