algorithm complexity

Why do we use the tightest Big O bound?

If an algorithm runs in O(n), then O(n²), O(n³), and even O(n!) are technically valid upper bounds too. So why do people expect the smallest or tightest bound…