@McTag,
The thing is Mac that the clue and the answer constitute another clue. It is to the state of mind of the compiler. Which I glimpse in myself when I go through the list of horses names before the start of the race looking for spoonerisms. I came up gold the other week with a horse called Breaking the Bank which spoonered very wittily I thought to Beaking the Brank.
But it doesn't last long. And only a name spooners satisfactorily occasionally.
The compiler must be at it in his sleep as well as all day long. He fancies an 11 letter word to chuck the clever readers under the virtual chin. Right? He Googles 11 letter words. One beginning with "a" to save time. He then Googles anagrams and gets up a long list of words contained in the word. He then looks at the letters left out. 4 in this case. If they make a word --wolf say, he's scored. Animal in theatre. Any category which fits the word he made from the laft out letters.
With astute and practiced arrangement of the black squares he can fit in the intersecting words with relative ease.
When Frayn's compiler died in TTEOTM there were the next ten years of puzzles in his desk. It is an excellent piece of malarky.
Influence from beyond the grave?
If there is a Google dictionary of crossword clues for all the words in the language it might be possible to produce puzzles untouched by human hand. To be marketed from Alpha Plus to Epsilon Minus. And having all the scientific hallmarks of artificial intelligence.