doug lerner wrote:

http://elliptics.com/dieters/mod/gro...3&group_guid=5

Calories, fat grams, protein grams, carb grams, fiber deductions from
carb grams, claims of other types of deductions from carb grams - They
all have error levels that are pretty bad. Folks try counting to the
nearest half gram or five calories or whatever end up with counts that
appear to have far more digits of accuracy than they actually do.

Yet counting works anyways. Why? Because dieting processes don't need
anywhere near the accuracy that counters target.

Not that I'm excusing grossly wrong numbers or deliberately wrong
numbers, just that paying attention to results corrects the errors.