This Tuesday's puzzle by Alan Arbesfeld reminds me of a trend I've been noticing as I've been examining NYT puzzles over the past several years. They're getting better.
In looking now at puzzles from the late 1990s, there is clearly a lot of ingenuity but what tends to be missing from the older crosswords is a sense that the constructor just nailed everything. So often I have the feeling that it wasn't quite right. The theme clues weren't completely symmetrical or they came in the wrong order or one only fit in via a particularly clumsy forced answer. You see that less and less now.
Mr. Arbesfeld's puzzle, a straight-ahead early-week relatively easy affair, did it all right. The five long answers started with WHO, WHAT, WHERE, WHEN, and WHY in that order. Everything fit together very naturally.
I speculate that various factors contribute. Mr. Shortz improves with age. Constructors learn from each other. I bet computers help too but it's hard to guess how much that contributes. I expect the influence of computers manifests itself most in the general increase in the quality of the clues and that great construction remains a mostly human endeavor despite the continued improvements in programs like Crossword Compiler, but WHO knows?
