I've just read the book, and I found it a good read. However, Walker has neglected an important factor in his models, at least as far as appetite control goes: time delay.
Imagine the house analogy from the book. Only, this time, we have the heater/air conditioner (let's call it a heat pump, with the ability to pump both ways) in one room, the room you want to control the heat in, and the thermometer that affects the feedback in another. Let us examine how this system works now.
Let us first assume that the heat is uniformely spread throughout the house. Both rooms will after a certain amount of time during which the heat pump is not working, be at the same temperature, and fall and sink at the same rate. This is the state we have our system in when we start off.
Now, the temperature drops below our minimum. The thermometer registers this, and sends a message to the heat pump to start negative feedback. The room with the heat pump is quickly brought back up above he minimum temperature. The room with the thermometer, however, is still colder. So the heat pump continues heating. And then, finally, the thermometer measures a temperature above minimum, and turns the heat pump off.
By now, however, the room with the heat pump is far above the preferred temperature.
The same happens in our bodies. We eat, and the food is processed. The body, once it has processed enough of it to see that it has gotten the energy it wants, tells you that you are full, that you don't need to eat anymore. Of course, by the time this happens, you've already eaten an amount of food that hasn't been processed yet. Depending on how fast you eat, this amount can vary from negligible to obesity.
So many of us don't actually have an Eat Watch that is wired wrong, but rather one that is set too slow. See now why it's actually a good idea to eat slowly?
It's also probably worth mentioning that it seems like one can actually rewire one's Eat Watch. Though that's something I won't go into.