* Posts by Half a metric tonne

3 posts • joined 14 May 2012

Lies, damn pies and obesity statistics: We're NOT a nation of fatties

Half a metric tonne

I'll weigh in - excuse the pun

wiggers says: "You lose weight in the kitchen and gain health in the gym. It's 80% diet, 20% exercise."

Couldn't be any more true in my opinion. There are a few things worth saying though.

Firstly BMI makes no sense at all. If a taller person were a proportionally scaled up version of a smaller person, you would expect that their mass (which is proportional to volume of course) would increase with the cube of their height. The BMI adjusts mass by the square of the height. Given that the population is getting taller, I am not in the least surprised that BMIs are getting higher. One thing that is very interesting is that if use BMI as a measure of obesity, we find that tall people are generally more obese. If, on the other hand we assume that tall people and short people should, cross-sectionally, have an equal amount of obesity; and we perform a statistical fit to compute the relationship between height and weight, we find that a more appropriate measure of BMI would be weight / height ^ 2.5 (ish).

Secondly, carbs are the real issue in our diets. The evidence for this is mounting. Even if we correct the deeply flawed BMI measure, we find that the populations of many western countries (UK and USA for example) are getting much more obese. This occurs during a period when consumption of carbohydrate is actually reducing. If we go back about 30-40 years, the population of the USA got about 30% of its calories from fat, now they only get about 20% of calories from fat. But the calories from carbohydrates has gone up 10 percentage points during the same period. Why do carbs make you fat? As many people routinely point out, the only reason you get fat is if you take on (and digest) more calories than you expend in exercise (and base metabolism etc.). Why should calories from carbs be so much worse than calories from protein, or even fat? The answer is simple. When we eat a carb heavy meal, the body produces insulin to control the sugar level in the blood. When that insulin starts to dissipate an hour or 2 later, we suffer an insulin crash which makes us hungry again. Some people have no problem controlling that hunger, but many people end up snacking. That increases caloric intake. Furthermore, the body has many mechanisms to shut down appetite so we don't overeat. Certain sugars, especially fructose (but remember that sucrose is no more than a fructose and a glucose joined with a particularly weak bond), don't shut down our appetite through any metabolic pathway. This contributes to overeating.

I learnt about a lot of this because many years ago I got a medical condition which resulted in me putting on a lot of weight. I had a BMI of 44. Once the medical condition was under control I decided to start losing weight. I did a lot of gym time (this has the nice side effect of increasing your base metabolism), but it didn't result in any meaningful weight loss. As a final throw of the dice I went to see a doctor specialising in obesity. The doctor recommended a low-carb diet. Not an eat as much as you can, no carb diet (as some people seem to think the Atkins diet is). Over the course of less than a year my BMI came down to 30. Funnily enough, according to BMI I'm still obese. But no-one who saw me would agree.


Leaked memo: Apple's iMessage crypto has DEA outfoxed

Half a metric tonne

Re: The Code Book, by Simon Singh

"All ciphers are crackable, the pertinent question is how long it takes to crack the cipher."

Wrong! A one-time pad with a key from truly random data that is never re-used is impossible to crack. This is provable in a very simple way. A particular cipher-text can decrypt into any message of the given length. So even if you can cycle through every key and decrypt the message. You can't tell whether a message you get out is the right message. For example, if you encrypted a message, "The enemy is descending on our position" with one key, you could decrypt it with a different key and get, "The enemy is running away in disarray!!" There is no way, unless you actually have the real key, to know that you have the real message.

Even symmetric keys have limits to them. If you place a Dyson sphere around the sun to capture all energy it produces for the remainder of its life. And you also create a perfect counting state machine that uses a single quanta of energy to change state. There will be insufficient energy to count through all the keys of a 256 bit symmetric key algorithm. That's just counting the keys and doesn't include the problem of actually decrypting the message to test each time. So whilst theoretically true that "symmetric key ciphers can be cracked with enough time" the reality is, enough time often does not exist.


China begins work on world-beating MEGA power cables

Half a metric tonne

Re: It doesn't bode well...




Sometimes it is, sometimes not.



Biting the hand that feeds IT © 1998–2017