The short page wins because it fit the query better and answered it faster, and fit beats length whenever the two compete. When a 600-word page outranks a 3,000-word one on the same term, it is almost never an upset; it is the system working the way it is meant to. Google ranks by how well a page satisfies what the searcher actually came for, not by how much sits on it.
The 600-word page usually did two things the long one didn’t. It matched the intent precisely, and it delivered the answer with no wading. The searcher got what they wanted in the first screen and had no reason to go back to the results. The 3,000-word page tends to lose on one of two counts. Either it answered a slightly different intent than the query wanted, or it had the right answer but buried it under preamble, definitions, and filler the reader had to dig through to reach it. Someone searching a quick, specific question does not want five paragraphs of background first. The moment they have to wade, they bounce back to Google, and that round trip tells Google the long page didn’t serve them.
So length didn’t lose to brevity; it lost to fit and efficiency. The extra 2,400 words were not depth the query needed. They were distance between the searcher and the answer, and distance is exactly what the ranking is built to penalize, because the system reads how cleanly people get what they came for.
Stop judging a page by its word count and start judging it by how well it matches and resolves the query. That is the contest the ranking actually reflects, and it is the one a leaner page can win against a longer rival any time the longer one mistakes volume for usefulness.