Beyond Scaling Laws: Why "Thinking Longer" Is a Systems Problem, Not a Prompting Trick

For five years, the answer to "how do we make the model better" was always the same: bigger model, more data, wait for the loss curve to bend. Scaling laws made AI progress feel almost boring — predic...