Calories in, Calories Out (CICO) really stuck for me after experimentation. I managed to verify the model on myself, and on dozens of people over the years. The easy way was controlling people's diets and exercise. That let me see that I could easily have them control their weight. But then, I discovered a simpler, much less involved, and far more scientific way to do this in a blogpost on the topic. The blogpost is entitled "Calories in, calories out" and it shows a simple method you can use to prove to yourself that CICO is real, true, and a great description of reality. To get started, the post says to take your weight and to correctly record reported calories in the food you eat, plus to get down your running mileage for the day. You plug these into a recurrency relation and compute a bunch of predicted weights. When you do this, you eventually get predictions that are right on the mark: Simple enough, no? So, I did this for myself. I used measured my calorie intake, activity, etc., and I started predicting my weight changes with a shocking degree of accuracy. I tried and failed a few times at the outset, but eventually, after I got all my measurements dialed in, it all just worked, and I could even plan my days ahead of time and still get accurate weights predictions. That was proof enough to me that CICO worked. After I found this worked for me, I decided to evangelize. I told people about this, and I applied it to my girlfriend at the time. Turned out, it did not work for her. Odd, I thought, so I decided to monitor her activity and diet more carefully, and I found that what she reported to me was wrong. When I catalogued everything, we had some discrepancies, and when I addressed them, suddenly this worked for her, too! But now I had become acutely aware of the recording issue. If you're unfamiliar, when we actually measure caloric intake using doubly-labeled water, whole-room calorimeters, and other methods like those, there's evidence of systematic misreporting. For example, two recent datasets both showed that fatter people tended to underestimate their caloric intake:* So, I started leading people this by asking them to record their weights on a water fast. That seemed to be a one-shot trick to make the predictions line up really quickly. If you want to get more accurate, the equation you get your predictions from has a burn value per pound of weight. For men, this value tends to be higher than it is for women, in large part because men have a higher proportion of muscle and that tends to burn more calories than fat. You can easily obtain this burn value by measuring your caloric intake and weight and then figuring out the least-squares best-fitting value for the burn rate. The author of the blogposts did this (and provided a means to do it), and it seemed to take him a few weeks to get to a stable value, but eventually he did. With this value personally estimated, you can improve your predictions even further. You can also get to this value faster through leading in with a fast, as that takes out noise from misestimating your caloric intake. Neat, huh? The next question is: Does this hold when you're not losing weight? The answer is, broadly, yes! The author kept recording and eventually got to the point where they went on a trip. They ate at buffets, ate out at restaurants, and were unable to get accurate calorie counts. They also ate enough to regain weight. As you can see, the predictions went out of whack a bit, but then return to normal. They were all fine (good and consistent measurements), all messed up (bad and sparse measurements), and then back to normal (good and consistent returns!). Part of this is the poor measurement, and another part is something that I've managed to experimentally verify: water weight! When you lose weight, you deplete your glycogen stores and weight loss is initially fast as a lot of it that sloughs off you is water. When you regain weight, you rapidly regain glycogen, and thus body weight due to added water. This comes off just as easily, but it throws the model for a loop. If you start consistently gaining weight, you get back to accuracy. Or if you return to losing weight, you get back to accuracy. But in the transitionary period, when there are big swings, you lose substantial—but far from total—accuracy. And then when you're maintaining weight, you stay accurate the whole time. Discovering this felt incredible. Everything really just works, and you can reliably predict your weight over time. You can even add complexity, like playing around with information on your cycle if you're a woman.** You don't really get that much benefit from adding complexity besides, maybe, tightening up those transitionary swings, so either way, this is fine. The really nice thing here is that this obviates a lot of issues with interindividual variability. A lot of people object to CICO because of the effects of things like differences in resting hormone or activity levels. And that's fine, but it's not relevant here, since we're measuring one person, and they tend to be consistent enough that this just works. This is inherently self-controlled, so after your calibration, all of those individual-varying factors get wiped out, except in as much as they evolve with time as your weight changes, for whatever reasons. If you want to do this yourself, I highly recommend reading the articles and doing just that. If you do, you too can figure out how to trivially start being a true believer in the power of CICO, because it just works, experimentally. Links: * The girlfriend I mentioned was very thin, so this systematic bias wasn't the issue. Notice the individual variability around points at all levels. I consider that a more likely issue. ** A few women I know of gathered data from different parts of their cycles and found burn rate differences. They were able to measurably improve prediction accuracy with that information.

Jeder Fast-Food-Laden versucht sein Glück mit einem Geburtstagsgeschenk für einen ganz besonderen Schwein.