The challenge is using efficiency across just a single game as if it's meaningful. That runs into the same problem as other single-game metrics, which is that the numbers are too small. It was 64 possessions. Some people are even taking this down to specific 5-minute spans in a game!
In a given game, any number of factors can result in a lot of variance. The player who gets hot and hits a ton of threes in the face of stiff defense. The game we bungle several wide open layups. The tight whistle game where an opponent was hot from the FT line, or where we went ice cold. The game where their star player was out or missed most of the game due to health/foul trouble. Etc.
Over the course of a season, the variance smooths itself out.
Looking only at a single game, defensive efficiency doesn't necessarily speak to how good or bad the defense was - you can give up a lot of points in a game where you've played great defense if the other team gets hot, and you can hold a team to fewer points than their average in a game where you've played poor defense if the other team is cold. Over 30 games in a season, those factors smooth out.
As Pike says, sometimes the shots fall, and sometimes they don't. If you take more good shots, you'd expect more of them to fall - but that's not always the case in the microcosm of a single game. And if you force more bad shots, you'd expect more of them to miss - but that's not always the case in a single game,e ither.