13 Weeks Extra: In Which We Do Arithmetic
Is it true that "calories are calories"? My experiment suggests not.
November 21, 2012 - 7:00 am
I’m doing a low carb Gary-Taubes-like diet and adding high intensity training for 13 weeks to see how it works. This 13 Weeks series is my diary of the experiment; you can also follow me day to day at my Facebook page.
Yeah, yeah, I know it’s not Sunday, But I noticed something so interesting I decided to do an extra post; I’m still planning to talk about high intensity training this Sunday. I’ve been doing a food diary at LoseIt!, and tracking my weight in a spreadsheet and at Physics Diet. So I’ve got a pretty solid diary of what I’m eating and its nutritional contents. Now, Physics Diet is pretty solidly devoted to the traditional thermodynamic “a calorie is a calorie” model of weight loss. When you enter your weights, it computes some interesting statistics and charts them; it also computes how many calories you have been under (or over) your needs based on the rate of change in your weight. So, without further adieu, here are some charts.
First, a chart from Excel showing my weight and fasting blood sugar, both taken immediately after awakening every day. (Click to enlarge the charts.)
Notice that both trend lines are going down quite nicely.
Now, here are my charts from Physics Diet. First, here’s a chart for the whole time since I started watching carbs on 19 October.
Now comes the arithmetic. Nominally, a pound of weight is 3500 kcal; you have to cut out 3500 kcal to lose a pound, and if you eat 3500 kcal too much, you gain a pound. As of today, I weighed 278.6; I’ve lost roughly 23 lbs since 19 October, when I weighed 301.5. That means by the “calories are calories” model, I had to have cut 80,500 kcal over that month and a day, or about 15,000 kcal a week — 2515 kcal a day — under my metabolic needs in order to get an average weight loss of 4.25 pounds per week. Honestly, that seemed unlikely.
But then if we look at the chart for just the time I’ve been really running the experiment, it gets even more interesting.
Now, this is “only” 3.6 pounds a week, but that’s still a 12,600 kcal/week deficit. And here the interesting thing: I go back to LoseIt!, and I find that I’ve been running something more like a 4200 kcal/week deficit. In other words, I seem to be losing weight at something like 3 times the rate that would be predicted by simply looking at the calorie deficit. Oh, and that calorie deficit from LoseIt! accounts for calories expended in exercise — the working out has been included.
In other words, the “calories are calories” model doesn’t seem to account for the real results. The usual explanation for that is “oh, that’s just water weight.” But over the month and a bit in which I lost 23 pounds, if I was averaging that 4200 kcal/week deficit — or 600 kcal/day — for all 32 days, that would predict a loss of about 5.5 pounds. So, we need to account for 23-5.5=17.5 lbs of water. A gallon of water weighs 8.35 lbs, so that is 2.1 gallons of water. For the water explanation to hold, I would need to have lost, and sustained the loss of, over two gallons of water.
That’s a lot. (And I can tell you, I’m exhibiting no clinical signs of dehydration.)
Now, another possible objection is that I must have not been keeping the food diary accurately; all I can say is I’ve been religious about it. But consider, I’ve been eating less than 2500 kcal/day. For it to be the food intake, we have to account for another 2500 kcal/day. I’m a big boy, but I promise you I hadn’t been eating 5000+ kcal a day before I started this; it seems very unlikely that I could have been much under 2500 kcal and still feel like I’ve been eating plenty, as I do. Hell, my big splurge day, 3 November, just before I started the experiment, including 2 Double Whoppers with Cheese and a large fries, as well as a big sweet tea. Even on that I only managed to eat a little under 3200 kcal. And I was uncomfortably stuffed for much of the afternoon.
It really looks like this “calories are calories” model isn’t fitting my experimental results.