New York Times, Friday, October 25, 2013

Joe Krozel mentioned recently that the whole trick of constructing quad-stack grids was to create as many independent mini 15x4 stacks as possible. And from them, try to build larger 15x15 puzzles if possible. However, he was not talking about complete mini 15x4 puzzles, rather he was referring to 15x4 units that have some legit 4-letter crossers mixed with 4-letter sequences that look like they have the potential to get extended into real words. This is why computers (and the dreaded autofill) play a smaller role in constructing these puzzles than many solvers may think.

For example, if you look closely at the top stack in today's grid, you find some good complete 4-letter words (MELS, DOSE, TAME, ORES) along with some non-words: ENAC, IMGA, AFIR, and MONI. So this top stack can only work, if a grid pattern can be found that extends those non-words into real words (for example: ENACT-, I'M GAME, AFIRE/AFIRST/MONIED, etc.

Usually at this point, the puzzle goes in my unfinished "scrap heap" pile, because a (good) matching bottom stack is next to impossible to find. However, in this case I got very, very lucky and found an earlier orphaned lower 15x4 set that fit quite nicely. So I sent a 66-word version off to Will, who rejected it saying that the stacks were "lovely", but he felt the central area of the grid was too blocky-looking and closed off. Luckily I found a solution, and an extra 15-letter entry to boot as you can see in the final puzzle.

Finally, I know there are a few too many partials in the lower half of the grid, and the "ME AT"/"MINCEMEAT" dupe sucks. But, on the plus side the 15s are reasonably lively... and there are none of those dreaded "ONES" phrases... for a change!

Any crossword with a "quad-stack" is impressive, and this one has two. Amazing! On the downside, MEAT at 50D duplicates part of 1A, and ODE (21A) really shouldn't appear in the same puzzle as ODIST (47D). But I clued the short words carefully. And overall the construction was too nice not to say yes to.

A couple of months ago, a non-crossword friend of mine pulled me aside and said I had to see something. He took out a newspaper he had saved and held up a NYT crossword, a themeless with two triple-stacks and very few black squares. "This is the coolest thing ever," he said. That's a pretty powerful statement from someone who doesn't even like crosswords.

I realize some people are going to hate today's crossword, but I say, haters gonna hate. I'm not a fan of construction feats for the sake of record-breaking, but I was so wowed by the visual of today's puzzle, with those wide-open spaces, that I had to give it the POW. Well done, MAS!

The usual knock on triple-stacks is two-fold (and these apply even more strongly for quad-stacks). First, because of their construction difficulty, they tend to use phrases which aren't very snappy, use too many common letters (RSTLN E), or incorporate ONES (A LOT ON ONES PLATE being the most notorious). MAS shines here, using eight really nice entries. At first I was put off by CHANSONS DE GESTE, but after looking it up, I decided I enjoyed learning about a term that's gridworthy; a deficiency in my knowledge base. It's just amazing that MAS managed to find two separate sets of four good entries for his stacks.

The second knock is that the crossing down entries tend to be tortured, giving the solver an equally tortured solving experience filled with a gamut of partials, abbreviations, esoteric names, roll-your-own words, etc. On this front MAS doesn't do quite as well. In general, there are a reasonable number of ugly entries, but there are so many of one type, partials, that it was noticeable during my solve. It's unfortunate that ME AT had to be clued as a partial to avoid the dupe with MADE MINCEMEAT OF, as that would have helped the issue. Should we give constructors a break, allowing for extra ugly fill when a feat is as cool as clean quad stacks? I think a little leeway is reasonable to expect, but the high number of partials did detract from my personal solve today.

So overall, very impressive work and an enjoyable puzzle. I'm usually dead set against stunt grids for their own sake, but in my e-mail exchange with MAS, I really appreciated hearing how he tossed out a whole bunch of triple and quad-stacks before settling on something he thought solvers would enjoy. Solver first, constructor second, that's what I love to hear.