Matt Ginsberg author page

Let's see. I don't have a lot to say. Here's a video of Dr.Fill solving the puzzle.

Needless to say, DF has no trouble here — all the entries are actual words, relatively uncommon for a Thursday! DF was glad to get the break. :) [Not to mention the fact that the words were obviously all in my dictionary as well. I do try not to reuse clues from my clue database, available for free.]

As usual, Will Shortz changed a decent number of the clues. In general, he improved things (as always!). Still, I was sorry to see [Lady Macbeth's concern] disappear for 1-Down and was pleased with my original clues for the 5-Down/6-Down pair, which were [Medically dangerous fibrous material] and [Medically useful fibrous material]. I also had [Muffin type?] for 37-Down; is that allowed in the current political climate?

One final tidbit: I have never put a fill-in-the-blank clue in any puzzle I've submitted to the Times. I figure if Will wants a FITB clue, it's always easy to put one in. So any time you see an FITB clue in a puzzle of mine, it's something that the editorial staff provided.

I hope everyone enjoyed the puzzle!

There's not a whole lot of history here; I think I was watching "Now You See Me" on TV when the idea for the theme came to me. Will changed sort of the usual number of clues; the only one I was really sorry to see go was [Fantine / King commonality] for DREAM. Love the song, love the speech.

You can see Dr.Fill solving this puzzle here. The purple squares at the beginning are what you see as DF analyzes the clues; then it blasts along, getting most everything right except for a couple squares in MAGI MOMENT that it fixes up soon enough. 100% correct!

Now onto the theme analysis. It doesn't know about these mixed themes, and there are more acrosses than downs, so it decides that the theme is "add a C." I can live with that. Or so I think: it reworks the puzzle, and decides to replace MODERN DANE, which has no C and is therefore clearly not adding a C to anything, with MODERN CANE, which is adding a C to MODERN ANE, whatever that is. TICAL is, sadly, an old word for the Thai baht. You can see on the right that it thought TIDAL was far better (scoring 5.40 instead of 10.01; lower scores are better), but that C thing really nailed it.

Sigh. I hate computers.

I've always wanted to make a Schrödinger puzzle. The old CLINTON/BOB DOLE puzzle was just so amazing. What I really want to make is a puzzle with two completely different solutions, but I think I need to be a better clue writer to do that. And to have more time on my hands!

As far as this puzzle goes, I often tell people (truthfully) that when my puzzles come out in The Times, I can't do them. That's especially true today; Will changed the vast majority of my clues. I'm confident that he thinks all of his modifications are better; I certainly think some of them are. We've discussed this briefly, and he seemed to think that my clues were just too weird. So now I'm trying to figure out if he's right. I had originally clued 60-A, for example, as "What Mrs. Ed would be" and had clued 61-A as "Tiller operator's drink, supposedly." Both of those clues require larger mental leaps than the clues that were eventually used. What do people think?

I will confess that I thought the gimmick here was localized enough that the puzzle itself would be pretty easy, which is why I tried to make the clues a bit harder than usual. But maybe I overdid it, or got it wrong, or what have you. Dr.Fill gets this puzzle right basically immediately (it doesn't get the Schrödinger element, of course, just filling in one of the two possible letters), and gives Will's clues a difficulty score of 54. It gets it right with my clues as well, but has to rework the answer twice to figure it out (which it does automatically); my clues get a difficulty score of 72.03. So my clues are harder, but that doesn't mean that they are either better or fair. Like I said, I'd be interested in comments from solvers.

My clues or Will's, though, I hope everyone found the puzzle fun.

This was such a fun puzzle to make; I hope it was as much fun to solve! Since everything is a word, Dr.Fill rips through it with no mistakes in a few seconds.

Will was very kind to my clues this time, using the vast majority of them and, as always, I found some of his changes interesting. I had gone with the slightly edgier [Bridgegate locale] for 19-D and with [Death of a telethon host?] for 35-A. Not sure what people will think of those. Will also removed my slightly political [Company that recalled 18 million products after outsourcing production to China in 2002] for 40-A. I had clued 9-D as [Start of something big] and was sorry to see that changed.

I never know where to clue fill as an abbreviation. So, for example, my clue for 100-D involved a couple of actresses. 110-D was [Am partner], and 106-A was [Baby/pigeon commonality]. I'd love to hear what people think of these changes, and I'll hopefully learn something!

Mostly, though, I had a great time finding the ever-so-painful puns and then subjecting everyone where I work to them. Hopefully, all of you groaned as much as they did. I guess I'd be remiss not to include some of the words that wound up on the cutting room floor. Here are the clues; the answers are below in a random order.

• What a wild stallion may do
• Asked for help
• Like "go fly a kite"
• Plain ol' spud
• Drag Karl around
• Passover celebration for the staff

• COMMENTATOR
• SAUTEED
• CRUSADER
• BUXOM
• HALLMARKS
• FORWARDS

I made this puzzle shortly after this year's ACPT, when all of the things I had done to explain "geometry" themes to Dr.Fill were fresh in my mind. Thinking about all the things that other constructors had done with words twisting this way and that, this seemed like a natural idea (and one that would test Dr.Fill's abilities in this area).

The puzzle worked, but Dr.Fill couldn't solve it! (As I only just now learned; I always take a break from Dr.Fill after the ACPT because I'm burned out). It doesn't understand the theme at all, even though it should, and winds up making six mistakes in its usual fast solve. I'll have to figure that out before the 2018 tournament!

As usual, Will in general made things better. My original didn't mark the theme entries but I think Will was right to change that. The only thing I was really sad about was seeing the clue for 52-Across change from [Parisian woman?]. I really liked that clue, figuring everyone would think of a *different* misdirection and happily (and with great confidence) put FEMME in the grid.

I hope everyone enjoyed the puzzle!

MATT: This puzzle has been a long time in the making, going through a variety of revisions since I first submitted it (and Will rejected it) in February. He seemed to think it had promise, but didn't like my selection of theme entries. So I found a new set of theme entries and, before rebuilding the whole puzzle, asked if they were better. They were.

So I built a couple of very different versions of the puzzle and didn't clue them, instead asking Will which he preferred. As I clued the one he preferred, I discovered that while the fill was reasonably clean, it had so many Spanish entries that it might as well be a Spanish puzzle! So even though Will had accepted it, I changed it around quite a bit (there is still quite a bit of Spanish in it, I fear), and Will took the modified version.

There are a lot of ways that we could have dealt with the mixed up entries. Will chose what I suspect is the most benign, cluing PLEA BARGAINS = LAG + PEABRAINS as [Some compromises = Delay + dodos]. The nastiest mechanism (which I preferred, of course), would clue it simply as [Drop knuckleheads behind] where the clues for LAG [Drop behind] and PEABRAINS [knuckleheads] are interleaved, just like the answer is. You can uninterleave them by realizing that one clue is italicized and the other is not. (I guess it would have been even nastier to not even have the varying font, but that struck me as *too* nasty.)

My favorite clues that didn't make the grade: [Unsur] for NORTE and [Shortest sentence in English although, on sad occasion, also the longest] for IDO, [Grass opposite] for TURF, and especially [Like Doyle Lonnegon] for STUNG.

On the Dr. Fill side, bottom line is that it found the puzzle very easy, since all the words were in its dictionary. But it made a mistake! At 57-D, GOTAB has seen a reasonable number of appearances in high-quality puzzles, but GETAB has only appeared in less popular places. (You can check this out via my free clue database). So even though DEER is significantly better then DEOR at 62-A, it's so freaked out by the fact that no one "famous" has used GETAB that it decides to stick an O in the intersection square instead of an E. As always, it has no real idea what's going on and doesn't realize (for example) that GETAB and GOTAB are the same except for tense.

I hope everyone enjoyed the puzzle!

DR. FILL: I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal.

Will definitely saved me on this one. I sent him two very different versions — one with the minitheme that you see here, and one with a ton of examplars like this, but with unionized still far and away the best. He wisely decided to stick with what you see, and I hope everyone liked it!

As I often do, a video of Dr.Fill solving today's puzzle. It has no trouble with it, which isn't a total surprise because all of the words in the puzzle are obviously in my database.

Speaking of which, I've been working on both Dr.Fill and the my word list recently, trying to use some additional techniques from machine learning to improve performance on both sides. The word list is definitely better, which should help Dr.Fill as well, since one of its criteria when it's a bit confused is just trying to guess how good any specific fill is from a crossword perspective. The machine learning stuff on clue analysis is a bit more complicated, and I can't tell yet if it's going to be effective or not. I've got lots of time to get ready to challenge the humans next spring at Stamford, though!

It's always interesting to see what happens to my puzzles between submission and appearance. On this puzzle, I clued all the horizontal words with circles as (SMALL SPOILER ALERT) simply [See circled letters]. But Will decided to make it a bit easier. Not sure who was right; my clue had been used in a similar "Kangaroo words" puzzle in the ACPT one year, although that puzzle was much somewhat more straightforward, cluing for example "israELi AirLine" as [See circled letters]. So this would have been a bit trickier.

I thought of the theme itself while sitting in church one Sunday. Then I spent the rest of the service trying to think of examples to use. This was good because I stayed awake but bad because I didn't pay attention to the sermon. There was no test afterwards, so it all worked out ok.

Meanwhile, I seem to be losing my touch a bit on the cluing side, as many of the clues that I thought were clever or topical didn't make the editorial cut. I used [Seahawks' star Marshawn] for 21-Down and [Berry, repeatedly] for 44-Down. I used [Disney actor] for 67-Down in light of the (fairly) recent film. Perhaps I am simply too old to recognize topical when I see it. (I must confess that I transitioned from one ACPT age division to another a couple of weeks ago.)

Not that I will ever compete, of course. And Dr. Fill remains comfortably in the "juniors" bracket.

If you want to see Dr.Fill solving today's puzzle, here it is, below. I apologize for all the video flashing and whatnot; it's courtesy of my Mac's upgrade to Yosemite. I'll try and track it down before the next ACPT.

Not surprisingly, DF does pretty well on this puzzle, since all of the fill consists of actual words or phrases. It makes a few mistakes early on, but has no trouble correcting them. The solution is found in well under a minute, but it then spends bit of time checking its work and looking for a theme or rebus possibility before announcing that it's done. Correctly solved in a bit under two minutes.

I hope everyone enjoyed the puzzle!

My last puzzle in the Times came with a very brief interval between submission and publication of 74 days. This one was nearer the other end of the spectrum, with the interval being 833 days, a bit over two years. (Still, not really so bad!) I'm not sure if it's related, but many more of the clues were changed on this one than on most of my efforts. Some of the clues I included but which never made it into print: [Deserted?] for SERE, [Crime recommended by the Doors in 1967?] for ARSON, [Gabriel Macht and Patrick Adams, as seen on TV] for SUITS, [Pigpen-esque] for MESSY, and [They're in knots] for TOADS. I used Maximum Decimus Meridius instead of Spartacus in 101-A (SLAVE), because I'm a Gladiator fan, and included the bit of trivia [Teddy Roosevelt's was the smallest when he assumed office] for AGE. I used [Burning it doesn't change the name] for ASH and had [Gathers, as Timothy] for HAYS because my wife is an equestrian. Will also saved me by removing a bunch of clunkers.

Cluing is interesting. I've remarked elsewhere that Dr.Fill does better on the NY Times puzzles than it does on those in the ACPT, and I suspect that the cluing is a lot of the reason. Dr.Fill is pretty well dialed in on the cluing style in the Times, but Will edits the ACPT clues much more lightly, and that's why I suspect it has trouble. Speaking of which, you can find a video of Dr.Fill solving today's puzzle below:

This was an interesting one. DF does terribly on its initial pass, but then checks to see if this is a rebus puzzle and concludes that it is (right!) with the rebus AD (also right!). It resolves the puzzle, and does much better. The only mistake is that it thinks that TEENY is better than WEENY for [Itsy-bitsy] and can't figure out what's going on with the existing clue for 103-D. My clue for WEENY was [Diminutive], about the same as Will's; my clue for WINE was [Red or white, but not blue]. DF would probably have liked that clue much better (it knows all about red and white wines), so I'll view this as another of Will's unending and gleeful efforts to cause Dr.Fill fits.

This is a relatively straightforward puzzle, but it reminds me how curious the whole process from construction to publication is. So, for example, this puzzle took 132 days from the time I submitted until it appeared, which seemed fast. Best ever for me is an incredibly quick 27 days; longest is 1377 days — just over 3 3/4 years! I have no idea why some puzzles go quickly and others don't; it just seems to be part of the process.

As always, many of the clues changed. I originally had [Con-dense?] for 46-Across, which I liked, and it took me forever to understand the clue for 9-Down, and I live there! As I often tell people, I can't do my own puzzles when they come out in the Times, and this one was no exception. Dr.Fill, however, had no such problem, solving it correctly in about 15 seconds. It had trouble in the southwest, but eventually managed to patch things up and get a perfect solve. I've put a video of the program solving the puzzle below, where you'll see Dr.Fill mess up the southwest and then fix it before spending the rest of its "ACPT minute" checking its work.

I hope everyone enjoys both the video and the puzzle itself.

I am fighting a battle with Will Shortz that I fear I will never win.

My goal is to put an Easter Egg in a NY Times crossword puzzle. So, for example, in my 9/8/11 effort with Pete Muller, the 72-word fill included the mini-theme KESEY, WOLFE and MERRY PRANKSTERS. Hidden and wrapping around the corners were ELECTRIC, KOOL-AID, ACID and TEST.

Will ran it on a Thursday and circled those letters.

On 2/18/12, I had a puzzle with a long row of diagonal squares. The letters above the squares spelled UPSTAIRS, and those below spelled DOWNSTAIRS. Will added a note in the Notepad: "This puzzle has two bonus answers in appropriate places. Can you find them?"

In today's puzzle, my original clues for 15- 39- and 43-Across were [TV show the last episode of which aired 6/23/99 5/23/10 9/23/11] (I liked the fact that the dates progressed forward in time this way). Will put a little extra in the clues. Sigh.

The reason I can never win this, I have realized, is that Will always gets the last word! And he actually does what is probably the right thing, catering to his broad audience instead of the handful of folks who would discover my Easter Egg.

A lesser man would give up. But not I! My only question is whether I should continue to *tell* Will when I put an Easter Egg in a puzzle. If I do, he'll spoil the surprise. If I don't, he'll be annoyed when he finds out. I'm caught on the horns of a dilemma.

But a word of warning. If you ever see a Friday or Saturday puzzle of mine that appears to have no theme, look again. There's never any telling when I'll manage to slip one by.