MGWCC #752

crossword 2:34
meta DNF 


hello and welcome to episode #752 of matt gaffney’s weekly crossword contest, “Costume Change”. for this week 4 puzzle, the instructions ask for a common 5-letter Halloween costume. all right. what are the theme answers?

i don’t know. it’s a really weird grid, with no long answers, a huge number of black squares, and most unusually, no all-over interlock: there are three totally disconnected sections of the grid, and several other places where multiple sections are connected only through a single letter. there is also some slight asymmetry, with square 38 being a white square but its symmetric partner a black square at the end of 28a/5d. that one white square allows 38d {Scottish jackets} ARGYLLS, the only answer in the grid longer than even 6 letters. (i did not know this word, by the way.)

i did note several answers in the grid that could fit clues for different entries:

  • {“Fire and Ice” poet} is robert FROST, but FROST could also be a {Plane grounder} like the FAA, and arguably also {Become hard to see through} FOG UP.
  • {Action movie character, often} SPY could also be the answer to {Watch closely} EYE.
  • {Use a cleaver on} CHOP / {Snip with scissors} CUT UP and maybe {Tiniest bit of evidence} SHRED are to some extent interchangeable; neither of those other answers could fit the SHRED clue, but SHRED could be the answer to either of the other two, and CUT UP could certainly be the answer to the cleaver clue.
  • {Walk casually} LOPE could also be the answer to {Easy pace} TROT, and perhaps vice versa.

i can’t see what to do with these, even though i suspect they are relevant. i also think we need to be looking at ways of altering the grid to make it connected like a normal crossword.

the fact that the final across answer was {It’s sometimes made of silk} SHEET certainly caught my attention, given that we’re looking for a 5-letter halloween costume. if nothing else comes up, i’ll try GHOST but without any real conviction. but this probably also wants to be a clue for TIE or NECKTIE—neither of which is in the grid.

what about the alternate title? “anybody want a refill”? this didn’t help me. i’d love a refill, and i guess this does suggest more strongly changing the grid—but surely matt can’t want us to entirely refill the grid. i would really, really like to put some letters into black squares to give the grid connectivity, but i just don’t see where to do it without creating nonsense entries.

well, i’m about out of time and out of ideas. i guess i’m trying ghost, but i have no idea what’s going on in this meta. somebody let me know in the comments.

(Matt here, adding the solution grid to Joon’s review. You take the nine 5-letter words from the outer two areas, refill them into the ten 5-letter entries in the middle of the grid, and CLOWN appears across the center).

This entry was posted in Contests and tagged . Bookmark the permalink.

20 Responses to MGWCC #752

  1. Garrett says:

    I had a hail Mary guess, but could not get the answer submitted on the website.

  2. Mikey G says:

    You get my random thoughts, in real-time, haha.

    1. Bizarre grid. What’s up with all those squares? And then I realize, “OH YEAH NOT CONNECTED.”

    2. 5-letter Halloween costume, so clearly something with “5.” Early ideas include CLOWN, FAIRY, and GHOST, and the latter simply can’t be (can it???).

    3. Note the asymmetry as well: this prevents an extra 5-letter word in NW (or maybe adds one in SE). So something’s up with that.

    4. There are 4 5-letter entries in NW, 5 in SE, and 10 (5 each way) in the middle. That middle 5-by-5 chunk seems suspicious. Let’s try stacking it:


    Yeah, nothing.

    5. So we have 9 entries “outside” the center and 10 inside it. Do we match up one to each word in the center and see what remains?

    6. Title: Changing costumes. Changing initial letters? Superimposing these in certain places?

    7. Just saw SHREK in the middle perhaps? (Tough metas make me go “OH GRRR.” Ogre. That was terrible, moving on.)

    8. Deleting the entire grid. Retyping it. Changing a letter? SHRED to SHREK, AERIE to EERIE, and so forth. Nothing much – CANID is kind of weird too, and that won’t give us anything. That 5-letter reminder in the meta prompt needs to have something to do with these 5 letter words. Hmmm.

    9. That semidupe FOGUP/CUTUP. I don’t like that. Do I “superimpose” FOGUP on CUTUP…and then FOGUP and can connect with FROST going down…FROST to SHEET. WAIT A MINUTE. We are supposed to replace that “five-by-five” grid in the middle with all the other five-letter words elsewhere in the grid.


    Someone else can do it better than I can, haha. But basically, replace the middle 5-letter words in the center (there are 10) with the other 9 5-letter words elsewhere in the grid (some of the words, such as FOGUP and SHRED, lock in what needs to occur). When you do this, the obligatory 10th word that must form, CLOWN, appears across the middle, the meta answer.

    My clown opens up doors for me. It was a nice jester!

  3. You don’t need to refill the entire grid — just the middle section.

    Erase all the 5-letter words in the middle and refill that part of the grid with the 5-letter words from the outer sections. You’ll be able to create a new valid grid … with one extra word which is the final answer.

    ETA: Mikey G beat me to it but that’s the shorthand version of what to do. :)

  4. Matt Gaffney says:

    Thanks, Joon — 135 correct answers, of which just 23 were solo solves.

  5. Stribbs says:

    Didn’t get it, lost a biggish streak, might be a little sad about some of the clues leading me down rabbit holes similar to Joon’s, and maybe can see some thinking that the focus on 5-letter words is a bit arbitrary…

    But the clues were there, the quirks in the puzzle shape all served to support the solving path, and the construction is super impressive… how anyone is voting this below a 4 is baffling to me.

  6. Mikie says:

    Explains the whooshing sound I heard this weekend…this meta going right past me. Impressive construction.

  7. Hector says:

    DNF, oh well. Neat concept!

    Follow-up quiz: here are two very similar grids. Where are they from?

    (That should be a link to an image file on my google drive. Apologies in advance if that doesn’t work here!)

    • Adam Rosenfield says:

      Week 2’s puzzle this month also had a grid that was suspiciously similar to week 4’s, sharing 30 black squares in the same locations (out of a total of 38 in week 2 or 61 in week 4). Week 1’s puzzle also shared 22 black squares with week 4. (But week 3 wasn’t a 15×15.)

      I spent some time down a rabbit hole trying to match up and combine grids—one theory was that the black squares in week 4 were the union of the black squares in two other puzzles, which would explain the big blocks of 3×2 black squares, 3 separated components, and narrow choke points between regions, but I couldn’t ever make that theory work out.

  8. Seth says:

    Holy. Crap. That is one of the hardest metas I’ve ever seen. I mean, it’s really cool and really impressive and I’m continuously astounded at Matt’s ability to come up with and make these, but wow that’s hard. HOW DO PEOPLE SOLVE THESE THINGS???

    Earnest question to anyone who regularly solves week 4s and week 5s: Is there any chance we could correspond by email a bit for the next few week 4s and 5s for me to maybe get some nudges? I know no one else who solves these metas, so I’m always solving in a vacuum, and I’d love to get a glimpse at the process of actually figuring out some of these really hard ones.

    • C. Y. Hollander says:

      Repost of my original reply, which promptly disappeared (perhaps caught by some filter)
      I usually solve them, but I missed this one, so I might not be the guy you’re looking for, but for what it’s worth, I’ll offer my retrospective thoughts on how I should have solved it (a worthwhile exercise for getting better, IMHO).

      For one thing, I ended up a little too committed to rabbit holes that weren’t panning out. That’s such a commonplace it almost goes without saying, but specifically, this reminded me that “two is a coincidence, three is a trend” is quite a solid heuristic. (This time, I had a theory about matching nouns that might be part of costumes with verbs that might be associated with them (à la the phrase “costume change”. I’d associated PESTLE with BEAT and CUT UP with SHEET, so once I noticed that fusing the two words gave me BEATLE and SHUT UP, and the grid contained entries that were one off from RINGO and CAN IT, I had a hard time convincing myself that it was coincidence after all).

      On the other hand, I wasn’t rigorous enough about subjecting every theory to the test of whether it explained the two anomalies that joon pointed out: the disconnected grid and the asymmetry. (Notice that while the A of ARGYLL couldn’t be removed without some significant reworking, the corresponding black square could have been quite easily removed by changing CHEX to CHESS.)

      It’s the latter point that I think could have put me on the path to solution. The wording of the prompt gives one to pay attention to the 5-letter entries; indeed I had listed them out and counted them at one point. I hadn’t noticed particular significance to the count, but I should have thought harder about it, because if something were important about that particular count, it would go a long way towards explaining why the asymmetry could not be fixed with CHESS (or any other five-letter word).

      That should have made me think harder about the count, and if I’d done so, I think I could have gotten to the next part: noticing that the 18 words broke down into 4, 5, and 9 instances by section. That’s already enough to suggest the possibility of matching them up the outer and inner groups somehow, and the additional hint given by Matt’s second title (which I finally looked at an hour before the deadline) makes writing the former over the latter a reasonable thing to try. It’s not quite obvious which to write over which, but it would help that the first outer entry (FOGUP) happens to be paired with the first inner entry (CUTUP).
      Anyhow, that’s my story of the road not taken. I hope it didn’t feel pointless and rambling to you, but if it did, I’d nonetheless encourage you to construct your own. Figuring out how you could have done something better is the best way to do it better the next time (again, just my humble opinion, but when it comes to crossword puzzles [meta- and otherwise], at least, supported by experience).

    • David Plass says:

      You should join the “Muggle” community where we have all sorts of meta solvers (including MGWCCs).

  9. Richard K says:

    Yeah, I completely struck out on this meta. I’d be curious to know how many of the 23 solo solves were by crossword constructors.

    The only brief excitement I experienced was when I decided to parse the alternate title as “Anybody Want A-R-E Fill?” There were several entries containing the letters ARE in that order — AROSE, AERIE, AREAS, ACRE, and SABRES — exactly five of them! “Filling” those in gave me OSEIASCSBS, which led to . . . nothing. (Esso basics? Sea is so CBS?)

  10. Jon says:

    My solving group worked really hard on this puzzle & were able to eventually solve it. Once one of us got it, she gave a good hint on the mechanism to solve it on our own. Even with the mechanic given, figuring out the order of substitutions was quite difficult. Which side section 5-letter words went in which particular center section entries was hard to figure out & upon reflection doesn’t appear to have any helpful pattern. Still not sure how I felt about this meta.

    • C. Y. Hollander says:

      With the mechanic given, you’re basically solving a fill-in, which, while not necessarily trivial, is a fairly straightforward exercise. If you realized or surmised that none of the non-five-letter entries would be altered by the refilling, you’d have 12 letters to start you off; otherwise, you’d have a little more work to do.

      • C. Y. Hollander says:

        It just occurred to me that there’s an alternate approach you could have taken if you knew the mechanism and just wanted to use it to get to the solution. For each position from 1-5, look at your word list and extract the letters at that position (i.e., list out all the first letters, all the second letters, all third letters, etc.). Now discard all the letters that pair off, so that you’re left with only the letters that appear [in your word list, for a given position] an odd number of times. Here are the five lists yielded:

        AH EK
        CL UN


        I’ve included the bottom line, which you must complete with one letter from each list so as to spell a common Halloween costume. You get the middle letter for free, since there’s only one of those. The 81 possibilities would still be tedious to wade through if you had to check each one, but you can quickly winnow it down further by observing, for instance, that OOW, OOE, and OOU are highly unlikely to appear in the solution (so you can cross off O from the list of candidates for the second position).

        Matt noted that most of the early solvers “didn’t quite grok the meta fully”. I wonder whether this was the path they took.

        • C. Y. Hollander says:

          Matt noted that most of the early solvers “didn’t quite grok the meta fully”. I wonder whether this was the path they took.

          P.S. If I’m not mistaken, this path is easier if you don’t distinguish between outer and inner entries at all and just count the letters from all the five-letter entries in the crossword. That way, the letters that overlap with non-5-letter entries pair off as well, leaving only CLOWN for the unpaired letters.

Comments are closed.