I used to write about crosswords for The New York Times at a blog called Wordplay. That job ended today so, of course, it’s not too soon for my first tell-all exposé that will blow the lid off that whole operation. Here’s the key question:
How early did I get access to the puzzles?
NYT Crosswords are typically published at 10pm Eastern time and Wordplay posts go live moments later. How is that possible? Some of the more clever commenters have suggested that I must get early access and I’m often asked, '”How early?” The full explanation is more bizarre than you can imagine. The surprising answer, and I swear this is true, is that the posts are written two to three months ahead of time, well before the crosswords themselves are constructed. Let me explain.
Times at the Times were particularly tough in 2008 when Will Shortz got hauled in front of NYT’s Supreme Budget Committee to hear the bad news. The lavish “puzzle lifestyle” was over. Costs had to be slashed immediately and the first cuts were to be in the admittedly well-padded construction budget. One of the junior accountants, a kid named Bumfry, had downloaded a demo version of the Crossword Compiler software with the Grid Generator plug-in. He noticed a button called “Auto-fill” and asked the obvious question: “Why can’t we just use that?” Shortz was taken off guard. He tried to argue using vague concepts like “quality” and “innovation” but made little progress until he blurted out, “What if I could come up with a solution that would slash costs but actually improve the puzzles? I have a radical notion but I need two weeks to pull it together.” He was given three days. The race was on.
The free-market solution
Most people know that Shortz did graduate work at Indiana University on Machiavelli and Adam Smith, but few also know he co-owns three key eBay patents and one more that underpins Google’s auction-based adwords business. If only there were some way to apply that same thinking to crosswords.
Then he got an idea. An awful idea. The Shortz got a wonderful, awful idea! My phone rang immediately. Was I in? The concept was so wild, so crazy, so utterly insane, that I immediately agreed.
Here’s how it worked. I wrote my posts describing each crossword and those descriptions were electronically distributed to constructors through a secret network code-named CRUCIVERB-L. Each post was attached to an interactive web page where constructors bid on how cheaply they could produce a puzzle that exactly fit those specs. It was brilliant.
So, for example, when budgets were particularly tight, I would write something along these lines: “This Tuesday puzzle has only three theme answers and displays little cleverness in the surrounding fill.” Veteran constructors would pass right over that one, knowing that competitive bids would drive the price down to just a few dollars. It was a godsend for newbies, though, who could try their hand at finally getting published without having to compete with Patrick Berry on the same day.
That saved enough money that, from time to time, I could be a little more creative. Sometimes, I admit, I was even cruel, using phrases like: “seven theme answers” or “every answer in this grid contains at least one B” or, in one particularly nasty case, “quadruple pangram.” It was great fun watching constructors struggle with how low they could bid on that one.
Reaction from puzzle authors
Immediate reaction from the constructor community was negative. Hostile, even. When I emailed Will Shortz to tell him the constructors were revolting, his quick response was, “Aren’t they, though!” I should have seen that one coming, I suppose. Many constructors dropped out in protest.
Those who stayed on came to appreciate the new system. Art thrives within constraints and crosswords had never been so thoroughly constrained. Although the average payment was substantially lower than before, pay for some of the more ambitious grids actually increased.
I sometimes got accused of playing favorites, particularly with visual elements in the grid. Descriptions like “connecting the circles in alphabetical order generates a reasonably recognizable portrait of Beethoven” became known as “Gorski Gimmees” because nobody but she even bothered to compete. Liz made out like a bandit on those ones. Quadruple stacks and record-low block counts weren’t nearly as lucrative, since Kevin Der and Joe Krozel kept outbidding each other. In other words, the system worked!
Puzzle of the year
There’s a particularly funny story about the crossword that I eventually named my 2010 Puzzle of the Year (Across Lite, XWord Info solution, Wordplay.) Francis Heaney apparently misread my blog description and, thinking it was a more straightforward normal rebus, opened the bidding at a ridiculously low $1,000. Of course, nobody else would touch it and he was on the hook to somehow put it all together.
The fact that it worked and ended up being such a wonderful puzzle completely validates the Shortz Strategy, as it has come to be known. Once again, free-market capitalism proves to be the answer to every question.