The red line is the smoothed trend line, while the grey line is the raw daily weight values.
My argument is that losing and keeping weight off is entirely about the balance of calories consumed and calories burned, and the fat cells are little batteries that store excess calories when there’s an surplus (you gain weight) and (slowly) convert fat into energy when there’s a deficit. The entire system is complicated, nonlinear, and full of delays between changes of inputs and measured results. But what we’ve learned from control theory, dating to the 19th century, is that such a system can be controlled quite effectively by measuring well-chosen metrics of its state and adjusting mutable parameters accordingly. The Hacker’s Diet chooses a smoothed trend of daily weight measurements as the metric and the calories contained in the food one eats as the controllable input. If one ignores the signal from the trend, it won’t work, just as if you disconnect the sensor from a thermostat it won’t do its job.
I have observed nothing in more than thirty years of my own experience that causes me to doubt these observations, and extensive correspondence with people who have used the tools described in the book indicates that re-gaining lost weight only occurs when people set them aside and try to “wing it” without the feedback from the trend line.