Joe Krozel author page

Anyone who solved my NYT puzzle from Sunday 1/18/2015 knows that I like to add new features to traditional constructs like word ladders. Curiously, though, the current puzzle idea came about when I wasn't actively seeking candidates for a word ladder; the concept 1-A + 44-A + 78-A was mentioned on the news one day, and the daydreaming half of my brain immediately noticed how easily 1-A could transform to 44-A. Suddenly, the analytical half of my brain engaged itself with the challenge of getting from 44-A to 78-A. No real difficulties there except for avoiding repeat words and proper nouns. (Meanwhile … I sometimes wonder what vital information I might be missing all the time when daydreaming along with the news).

The puzzle construction challenge was to lay out eleven four-letter words on successively lower rows of a 15x15 grid. I added the constraint of forbidding any non-thematic Across entries from being four letters long. The thematic content was spread out left-center-right to minimize the impact on the Down entries. So, the grid layout alone offered good chances of acquiring lively surrounding fill. In the end, I was especially happy that the longest entries were lively … and mostly unused before.

A few months after this puzzle was accepted, David Kahn had an NYT puzzle published (on 8/30/2017) involving a similar theme: a word ladder consisting of nine five-letter words with a thematic element at the halfway point. I was happy for David's success, but I thought it meant that my own puzzle would probably stay on hold for about a decade. Fortunately, that wasn't the case. I'm guessing that stylistic differences — including the absence of clues for the ladder words — provided enough freshness to receive a normal publication date.

I was inspired to produce a low-word-count pangram after seeing Patrick Berry's very Scrabbly 7/15/2006 puzzle. The starting strategy was to get at least two of J, Q, X and Z into the same quadrant. I think I tried seeding every quadrant with just the letters J and Q until I produced what is seen in the NE.

The procedure obviously involved judicious autofilling of successive quadrants; At first I used relatively high threshold values for autofill and eliminated clunkers from my word list with each successive fill, but eventually that left me with very few candidates for completed fill regions. Out of necessity I lowered my autofill threshold and looked for underranked entries which might be tolerated. So, for instance, JAILABLE and UNNAILED were allowed because the surrounding fill was quite satisfactory.

As I moved out of the NE quadrant with J, Q, Z, B, C and G checked off my difficult-letter list, I focused on two main issues: 1. Where to put the letter X, and 2. What to do about G and Z which now spanned two quadrants. (The latter issue was actually more urgent since I probably had plenty of words beginning with EX- … or could eventually invent new ones).

So, as I looked at the G, I decided that the last letter of 14-Down should be something common (R, S, T, L, N or vowel) and specifically considered flexible endings like GHT, GLE, GER, GED and GES. Separately, I decided that 26-Down would probably start with ZIP and 32-Across would have many possibilities ending in UP. This turned out to be a good starting strategy even though the final fill strayed from what I had anticipated. So, the SE quadrant was attempted second — because of the ZIP limitation mentioned above — and a few tentative fills were produced. With each of these candidate fills, I checked off additional letters from my difficult-letter list. At the completion of the SE quadrant, G, F and V were then accounted for, and that mainly left K, H, Y and X to be allotted between the two remaining quadrants.

Beyond the choices for 32-Across ending in UP, I expanded my seeding possibilities for the SW quadrant to specifically include TAKEASIP and HAVEASIP … because they immediately accounted for K and H respectively (while allowing 14-Down to end in an easy letter S. The fill accompanying HAVEASIP looked strong and allowed me to account for the letter Y. So, all that remained was to pack X and K into the NW quadrant. This partially involved luck, but I was also forced to accept one entry beginning with RE- and another ending in –ER. I decided I could live with those two specific words: RESOLE and CUSSER … even if the latter was prominently displayed at 1-Across.

I was pondering ways to revisit my NYT puzzle from Friday 1/27/2012 and add a new twist, AND I decided there were other tricks I could play with the central grid entry engulfed by black squares. It must have been a flashback to my childhood — reading a riddle book and asking my mother to explain the answer to "What are kids' two favorite letters of the alphabet?" — which brought 35-A under consideration. Suddenly it occurred to me that there was one particular enclosure (44-D) which made absolute sense for 35-A and the theme magically fell out of the sky. From there, the main challenge would be to get the grid art just right.

Grid art puzzles are always very tricky to make. A 15x15 grid is already pretty low-res to begin with. Add to that a requirement of word interconnectivity, and suddenly the grid art is pushed away from the walls ... into a 9x9 central space. I made at least six grid depictions that looked an awful lot like dogs and horses until I settled upon the current, rather neckless design. So, I'll probably get battered in the solver reviews for not getting it perfect.

One last perspective on the grid art: those black square stacks on the sides and bottom depict the ends of bats of six individuals waiting for their turn with the PINATA. Wasn't that enough of a visual hint?

Today's crossword originated with a study of rebus setups involving word fragments in some relative position. It just so happened that the first three examples I generated involved two side-by-side words rather than word fragments, so those words could be clued independently. Next, it occurred to me that each rebus needed added words to complete — or reveal — the rebus thought, and it turned out to be rather convenient to tag on these revealers immediately after the rebus words. So, the three-consecutive-entry layout was shaping up nicely from the start. Not long after that, I realized that the revealer words themselves were synonyms suggesting a rift — well, duh! — so I imposed that long rift of black squares upon the grid as an additional novelty.

For a while, I pondered whether any further theme embellishments were possible. Clearly TED, ANA and KEN have the name connection, but nothing elegant emerged from that observation. I also noticed that BUS and BAN could act as imperative verb forms, and I might have pursued that if BRO were the same … but alas, that was another potential connection which didn't pan out. In the end, not all ideas could be exploited, but those that were, made for a pretty decent theme on their own.

Finally, some solvers may know the "Rebus" category of crosswords and will recognize that this puzzle is not within that classification; I generally cannot solve crosswords requiring more than one letter per square, so — after 84 crossword contributions — I still refuse to construct anything like that. Instead, I decided I would produce an "anti-Rebus" rebus.

JOE: Tim bounced a preliminary idea off of me for a pyramid puzzle, and I responded with suggestions on how he might cut down the black squares and incorporate additional theme. Later, I was delighted when he came back to me and asked to make this a collaboration. I have thoroughly enjoyed the collaborations I've had with Tim ... and all the other constructors I've made puzzles with over the years (Peter, Martin, Victor).

TIM: Thanks very much to Joe for working with me on this.

I'd built a mirror grid with a smaller pyramid/entombed mummy and the four theme answers, but it felt like the puzzle was in this weird area between the theme's not being rich enough and the grid's not being visually distinctive enough. There was too much noise for the pyramid to stand out; too little theme to compensate.

Assistance was required. On went the Bat-Signal.

Joe liked the idea but agreed that the grid needed revamping. He expanded it vertically while adding in the larger pyramid at the bottom and the three others around the sides.

Incidentally, King Tut's tomb was discovered in the Valley of the Kings, not in a pyramid. We weren't sure whether to include him anyway, but decided to after seeing that he could be tucked in under the barrow-like central block formation.

Thank you also to Will and to the enfant terrible of crosswords, David Steinberg, for their help in revising the theme. Specifically: how to link MUMMY with the outside fill. The notepad/numbers idea was all theirs.

I had been collecting companion entries, e.g. MOIL and TOIL, for making the Schrodinger category of puzzle for what seemed like ages. Much of that search was facilitated by Matt Ginsberg's clue database which displays alternate entries with the same clue. (For instance, if I were to look up SIT, one of the clues [Meet] would also apply to FIT ... which differs by a single letter). Short-entry pairs like that were easy to come by, so I scratched my head to find something longer and stumbled upon 10-D and 25-D ... which differ by two letters. Good enough, I said.

Oddly enough, the placement of those long Schrodinger entries into a thematic presentation came only gradually; I initially thought the puzzle would be themeless. Even so, it quickly made sense to place a single entry vertically because of the orientation of these CAVERN growths in nature. From there it took a long while before I decided to depict them as two distinct entries. First, it seemed redundant, but the other mental block was having two, long, nearly-the-same, single-word entries in a puzzle: very taboo. Eventually the theme notion struck me, and it carried the day.

Now, having said all that, I should point out that the puzzle isn't the sort of Schrodinger puzzle where either theme entries is correct in either location: they make the best sense with one particular arrangement ... and I'm so naughty for this sly prank: Someone should look up what percentage of the population doesn't know which is which.

JOE: This puzzle concept came to me when I noticed that the two main theme entries were equal length: constructors lo-o-ove when that happens! The literal interpretation of them virtually begged running them in opposite directions. To build up the theme further, I decided to run the corresponding background fill the same direction. Next, the theme suggested that I might enlist a collaborator, so I called upon Pete. From there, I would fill one half of the puzzle, flip it around, and ask Pete to fill the other half. So that was the basic logistics.

When I construct, I think about fairness to the solver. I guessed there would be solvers who would be very unhappy about the trick in the lower half of the puzzle, so I addressed this in two ways. First, I included the helper entry BACKWARD in the top half of the puzzle ... preparing the solver ahead of time. Second, I made sure that the theme was completely absent from the down fill so that solvers could opt out in that way. That was the least I could offer.

By the way, the other "helper" entry DRAWKCAB was more of a joke than anything else ... not much of any help to the solver I suspect. But it was fun to claim that — technically — the same word appears twice in the puzzle!

PETE: Yes, this one was definitely Joe's idea. When he asked me to go along for the ride, I was more than happy to.

Beyond what Joe said, I'll just add a couple of things. Did you notice the necessary palindromes (ESSE/MOM/ATTA) through the horizontal midline of the puzzle, and BOB vertically intersecting MOM? My clue for BOB was "Go up and down", which I thought was appropriate given the circumstances.

Second, did you notice PETER in the top half of the puzzle and JOE in the bottom? Of course you did.

TIM: Joe and I found two theme arrangements that allowed us to finish this puzzle. Because we weren't sure which grid Will and Joel would prefer if they liked the theme, we submitted both. Each has its pluses and minuses.

The alternative grid's biggest virtue is that its theme answers are well distributed throughout. The greater length of SHANTIES (8) over CABANAS (6) is another advantage.

There's a bunch of bouncy stuff (TRADE WAR, LIVE AMMO, ROOKIE MISTAKES, TAMPER RESISTANT, RUM RAISIN, SCIENTIST), but there's also a fair accumulation of junk (ADAH, RST, TSE, A KING, ATL, ENGR). From an aesthetics standpoint, SSSS looks much worse than SSS, while the triplication (and crossing) of SEASONED TO TASTE, REASON TO LIVE and TURN TO STONE is hardly ideal. And PARENTIS, a slice of abstruse legalese, might have been the worst submitted entry since the time some nitwit (that would be me) sent in a puzzle containing ADEEM a couple of years ago.

On the other hand, the spacing of themers in CABANAS isn't perfect, even if the placement of a theme answer at 1-A (one of Joe's original, starry-eyed ambitions) helps to make up for it. A few clunkers of its own aside, the overall fill of CABANAS feels a touch zippier, and I think both of us prefer it now that we're removed enough to judge through more objective eyes.

If nothing else, the opportunity to clue SPINAL TAP made this whole endeavor worthwhile.

I noticed that when acronyms and initialisms appear in clues they automatically pin things down. For instance, the clue [Part of GPS] readily identifies "Global Positioning System" and guides the solver to an entry like GLOBAL or SYSTEM or SYST. So, I wondered how a solver might react to a more generalized clue like [Part of ?PS] which could refer to FPS, GPS, IPS, PPS or UPS. This was the basis for today's theme (and the stars in the clues eliminated the need to repeat "part of" in every clue ... perhaps even adding a mystique).

Most acronyms and initialisms contain short words — VAT, for instance, could yield VALUE or ADDED or TAX — and these words wouldn't normally form theme entries by themselves. But I had pioneered a puzzle style in my 12/8/2009 NYT puzzle which uses short-answers-having-something-in-common in all the across entries, so the current theme seemed like a natural fit. What I like about this type of theme representation is that solvers can opt out of the theme entirely by solving only the Down clues, but I like to think that solvers would be delighted by the nature of the theme once they actually did catch on to it.

One last thought: kudos to anyone who solved the entire puzzle with only the Across clues!

My last revision of this puzzle was submitted almost two years prior to the publication date, so when I sat down to solve this final edit I had long forgotten the fill, and Will had vastly livened up most of the clues. Hence, this would be as unbiased a solving experience as I could contrive for myself. For me, I had to confirm that a "No-Google" solver could indeed complete it, and I am happy to report that I did ... completely Google-free.

I consider the "No-Google" solvers to be one critical audience for a puzzle, and the best way to ensure their satisfaction is to minimize the density of unfamiliar content. My greatest concern during construction must have been the proximity of CHEROOT and REYNOSA in the Northeast and the crossing of TCCHEN with HSIA in the West, but they were only minor obstacles for my own solving experience. So, I'm satisfied with the way the puzzle turned out.

Just in case anyone was wondering, here are some of my own missteps while solving:

• 13-A NOMADIC
• 26-A MIS
• 28-A DEY
• 34-A SCALE
• 50-A IDO
• 54-A ONICE
• 57-A ALEUTIANISLANDS
• 3-D MARINEENGINEERS
• 12-D PROGRAMDIRECTOR
• 14-D CEDAR
• 59-D CES

I'll be interested in seeing whether the regular solvers had a similar experience.

This theme originated with my observation that SUPERSONIC contains the word PERSON, so I checked the clue and puzzle databases and didn't see that any constructor had connected the two words ... at least in recent history. I was off to a good start. I explored fill-in-the-blanks, e.g., Su____ic, as one way to express this theme, but a theme entry like PERSON would be too short. Then for some reason, I was inspired to interpret Su____ic as a pictoral description. Aha! MISSING PERSON! All I needed was other examples (and, indeed, they would come to me over time).

Many months earlier, I had advocated that constructors cram less theme content into the daily puzzles, but rather expand the theme from, say, five entries to ten and produce a Sunday puzzle instead. Most editors have low Sunday inventory, so packed daily puzzles simply deprive it from growing. This is why I decided upon a Sunday puzzle after having produced just a single one — a collaboration, even — in 2008.

Sunday puzzles are a bear to fill. Duplicate words in the grid are a frustrating obstacle. But I fought hard against that, and I'm particularly happy with the outcome: the fill contains lively long entries cemented with clean short entries. There are no partials in the fill because clues with blanks are reserved solely for the theme entries today.

Overall, this is among my favorite puzzles on account of the success with both theme and fill. Though, I would have liked Will to keep my title "Blank expressions."

As groundwork for this puzzle, I created various Morse code palindromes — first noting that letters E, T, I, R, S, O, P, M, K, X, and H are symmetric in Morse code while letter pairs A/N, D/U, G/W, B/V, L/F and Q/Y are mirror images. But only nonsense emerged: NUN AWAITING NADA; WAIF GO HOWLING; TEAR IN HAIRNET; RING OF SLOW AIR; NET LOSS OF TEA; SEQUIN SAID YES. Really goofy stuff.

So I abandoned the thematic-entry approach and looked toward grid art using black squares to spell out a simple Morse code palindrome: WAITING, FOOTSTOOL or TOP SPOT. I devised numerous odd-sized grids that mostly wouldn't fill before stumbling upon the present grid which would.

Somewhere in the whole process I decided it wouldn't be fair to expect solvers to even know Morse code to get 20-Down, so I double-clued it. As such, the puzzle has only the slightest hint of a theme: one might describe it as more of a themeless with a post-solve conversation piece. (This seems to be one of my signature constructing styles). So, hopefully a few solvers are enticed to ponder the subject of Morse code palindromes for an extra bit of amusement.

Finally, a few words about fill. Given the fixed nature of the grid, it was important to me to introduce as much novel content among the 11- and 15-letter entries; I also wanted the 3- and 4-letter entries to be common stuff. Plus, I wanted all easy content in the vicinity of 20-Down. So, I hope these mini-objectives helped me achieve my goal of an interesting yet smooth solve overall.

Martin:

I like to think of today's double-quad-stack puzzle having a long-lost triple stack little brother from 1996. Do you see any similarity between the grids? In reality, any similarity is just a coincidence, but an intriguing one, nonetheless.

When I started constructing the top stacks for this puzzle, I though it would be fun to use a 15-letter word/phrase that is commonly seen at the very bottom of stacked puzzles: SATELLITE STATES, and placing it at 1-Across, at the very top. In this position, its abundance of low-point (useful word-ending) Scrabble letters, normally an advantage, would be all but useless.

After I finished what looked like a promising top stack, I then contacted my friend Joe Krozel to see if he had an "orphan" stack or two, that we could meld into a finished puzzle. By luck he did. Our first draft had 10-D SLUG, and FAIN in its mirror position. Luckily we were able to open the grid up more (and reduce the word count to 66) by extending both entries to SLUG IT OUT and SAMMY FAIN. We got lucky that day!

Joe:

When constructing my first dual quad-stack puzzle around 2010, I assembled a library of half-filled grids containing a single quad stack at the top or the bottom. So, I had numerous potential matches when Martin presented his upper quad stack to me; I just had to search my library for the closest matches, then Martin and I would move squares around to unite the two puzzle halves with the proper symmetry.

I started the lower quad stack with the lively seed entry AMIGLADTOSEEYOU. The vowels at the front and back ensured that 55-D and 60-D would be words rather than some all-consonant letter strings like SSSS or SSTS.