Good instinct to notice that — you're right that insertion sort feels fast sometimes, and that's the key to Big-O. Quick question first: when it felt fast, was the list already mostly sorted, or shuffled randomly?

That's exactly it. Big-O describes the worst-case growth rate, not every run. On a nearly-sorted list insertion sort barely shifts anything — that's close to its best case, O(n). On a reverse-sorted list every element shifts past all the previous ones — the O(n²) worst case. It's a ceiling on how bad it gets, not a promise. This is CSC 203 Module 1, Week 2 (Asymptotic Notation). Want a merge-sort example that doesn't have that best/worst swing?