As I've said elsewhere, calorie estimates on packages are close enough that it doesn't really matter what you eat as long as you estimate consistently.

That still doesn't make sense. One example cited in the article says that for almonds, 129 calories were absorbed out of the 170 listed on the label. That's a 25% difference. The article unfortunately doesn't give a whole lot more info on how this tracks across various foods under various methods of preparation, but it would seem to me that if this difference is highly variable, it basically defeats the idea of fungibility of "label calories", which seems to me to be the entire point of listing them in the first place. Granted, if you're basically eating the same stuff, but just increasing or decreasing helpings, that's not such an issue.

People might disagree with my reasoning here, but I'm not really seeing how I deserve downvotes.

That 25% difference is for a single serving of food and only translates into a ~2.5% error in your daily food total. People eat a varied enough diet that a 25% difference in a single food item really won't make a huge difference.

Lets take a man who needs ~2500kCal per day to maintain his current weight. A nice steady loss of 1lb per week would require this guy to eat at roughly a 500kCal/day deficit (a total deficit of 3500kCal translates to a fat loss of 1 lb). Even underestimating his total caloric intake by 10% (he eats 2200kCal instead of 2000kCal) every single day will still see him losing weight at .6lb per week.

After 4 weeks of this he can see that he's only lost 2.4 lbs and can therefore reduce his intake by another 200kCal/day to achieve the desired 1lb/week loss. Tracking your weight/calories like this makes it fairly trivial to lose weight. It really doesn't matter if all the individual food items have perfectly assessed caloric content...as I say, they're close enough.