What One Long Question Costs You

The test runs on a fixed clock with a fixed number of questions, so time is a shared pool. Spend four minutes on one stubborn math question instead of the average pace, and that extra time doesn't vanish quietly — it comes directly out of the minutes meant for later questions. Because sections are linear, not adaptive, those later questions are sitting there waiting, often easier than the one eating your clock. Overspend early and the damage shows up at the end as blanks or rushed guesses, not as a dramatic moment in the middle.

Here's the trap: once you've sunk two minutes into a hard question, finishing it feels like it 'pays back' that time. It doesn't. The two minutes are already gone, and the question was worth exactly one point whether it took 20 seconds or 200. The only real decision left is what your NEXT minute is worth — and a flagged guess now, followed by two fresh questions later, almost always beats three more minutes of grinding on one problem.

This is why a personal time cutoff matters more than raw knowledge on borderline questions. Flagging a question and moving on at a set limit, say 90 seconds, protects every question behind it. It connects directly to how the whole section's pacing works: since every question counts the same, the smartest use of a scarce, non-adaptive clock is to guarantee you reach every question at least once before you let any single one hold the rest hostage.