Slate on achieving substantial similarity without copying

This is a really interesting story about how two crossword puzzles could end up looking extremely similar without copying, through the interaction of a similar idea and the scenes a faire of crossword construction. A little reminder that almost all genres are more complicated and convention-bound than they might look to an outsider.

Here's author Matt Gaffney's grid:

Here's the predecessor's (Mike Shenk):

Why are those long entries nearly identical, placed in the same way, and with nearly the same black-space pattern? As Gaffney notes, the crossword community is small enough that people are likely to detect copying--and he heard from a number of people about the similarities here. The explanation begins with the popularity of themed puzzles:

[M]y raven puzzle isn't an isolated case—crossword constructors duplicate one another's themes and grids all the time. The all-time most overused theme might be this list of breakfast foods each beginning with a European adjectival: ENGLISH MUFFIN, FRENCH TOAST, DANISH PASTRY, and SPANISH OMELET. (These puzzles are invariably titled "Continental Breakfast.") Another one that I'd rather lose an eye than see again: ERNEST HEMINGWAY, THE SUN ALSO RISES, and A FAREWELL TO ARMS, all conveniently 15 letters long.

Gaffney argues that four factors explain the similarities here:

Why was the theme exactly alike? There are perhaps two-dozen types of crossword themes, and puzzlers like me have done them all repeatedly. One popular example is adding or subtracting a letter to well-known phrases to get humorous new nonsense phrases. (A puzzle titled "C-Minus" might include the entries ORLANDO MAGI and ASH FOR CLUNKERS.) Embedding a word (like RAVEN) in longer entries is another popular convention. Since we're all essentially hunting the same wordplay quarry, it makes sense that two crossword constructors would hit upon the same bright idea. Unfortunately, Shenk beat me to the punch on this one.

Why were the theme entries almost exactly the same? Within those couple of dozen theme types, constructors look for specific criteria when selecting their theme words. This leads into a finer point of crossword design: When embedding a keyword in longer entries, it's considered elegant to break up that word as many different ways as possible.

... In the case of the RAVEN puzzles, Mike Shenk and I were both looking for lots of keyword-splitting variation. We each found four different splits: RAVE/N (BRAVE NEW WORLD), RAVEN (CONTRAVENE, INTRAVENOUS DRIP), RA/VEN (COBRA VENOM), and R/AVEN (VENTNOR/ST. CLAIR AVENUE). ...

Why were the theme entries in the exact same places in the grid? Primarily because American crosswords exhibit something called "180-degree rotational symmetry." In other words, if you turn the grid upside down, the pattern of black squares will look the same as it does right-side up. This requires puzzle designers to offset each theme entry with a same-length entry, a constraint that largely locks in the shape of your grid once you've got your longer clues down. ...

This constraint explains further why both Shenk and I chose our five entries: The 10-13-15-13-10 pattern is beautifully symmetrical. An otherwise good nine-letter entry like film director WES CRAVEN (extra points since he's a horror director and this is a creepily themed puzzle) was unfortunately left on the cutting-room floor. I simply couldn't find another nine-letter entry to offset it.

Why, then, were the five entries placed the same? INTRAVENOUS DRIP, as the only 15-letter entry, had to go in the center. Putting the 10-letter entries in the third and 13^th rows was also a no-brainer: nine-, 10-, and 11-letter entries fit nicely on the third and 13^th rows of a standard 15-by-15 grid, as they allow the top two rows to be broken into chunks of four-, four-, and five-letter words. (More on this later.) A 13-letter entry doesn't fit in row three—it would require awkward clusters of black squares—so it's relegated to the center of the grid.

Why did Shenk and I both place CONTRAVENE on top and COBRA VENOM on the bottom? ... People tend to solve crosswords from the top to the bottom, so we both chose to lead off with the dullish CONTRAVENE (a semi-boring word that semi-boringly embeds the keyword completely inside) and finish with the awesome COBRA VENOM (snakes are very cool creatures, plus the keyword is divided in an unexpected way). ...

Why is the black-square pattern so similar? Because a series of crossword rules makes it likely. American crosswords disallow two-letter words, meaning a 15-by-15 grid is likely to be filled with many three-, four-, and five-letter entries. This is especially true of grids with five theme entries, on the highish end for a 15-by-15 grid. ...

Gaffney then asked a third crossword-maker who hadn't seen either grid to construct a RAVEN-themed crossword. The result was not quite as strikingly similar, but there was still an overlap of four long clues and a similar, though mirror-imaged, layout of black space. Gaffney concludes that, in a form like the crossword, events of this sort are inevitable, though you'll still be able to distinguish copying from shared inspiration: "even puzzles with common themes won't be identical—the shorter words (the 'fill,' in crossword terminology) will certainly be different, as will the clues. Crosswords are like snowflakes, you see—even the ones that look a lot alike are still unique."