also, counting calories works. anyone that has actually done it can tell you that.
We have years of experience to show that across a population simple calorie counting does not work.
On the other hand, the problem with calorie counting is that most people only do it for a very short time, because it sucks. It's totally unsustainable for most people because they don't have the time or aren't willing to change their lifestyle to include hours per day spent focusing on nutrition. So they lose some weight over the course of several months, then let their nutrition lapse and gain it all back in a few weeks. So in the long term you're right, across a population calorie counting doesn't work. But only because across a population people are lazy.
Pretty basic math tells you even a very small percent wrong (1-2) would affect weight gain/loss over the course of a year :)
I'm also not sure what you mean by "consistency". You say " If you aren't losing weight, adjust your caloric intake until you are losing weight at a steady ~1-2lb per week."
How can you adjust your caloric intake if you don't know the real numbers? I guess you are assuming you eat literally the same thing every week, and drop stuff on the floor until you have a set of foods you know make you lose weight, and then only eat that for the rest of your life? That seems a bad strategy to keep weight off. Most people would probably be willing to eat less of something (which they sanely could if they knew the real calorie counts), but probably not change their entire diet and keep it steady forever.
Even if you just reduce the number of calories to try to account for all the inaccuracies, you are still in a situation where you actually have no idea what you can or can't eat. IE i you say "well, 1500 calories a day didn't work, i'll try 1000", in practice, some days you will have eaten 1500, some days, 900. Your claim this will just average out magically is patently false (it only averages out over time if the numbers are less than a certain percentage away from right, and that percent is very small).
If you are eating different combinations of things, the likelihood you will ever find which are wrong/right, so you know to eat less/more of it, seems really slim. You still have zero idea what you can eat for real, and you'll still find if you swap what foods you eat, you may start gaining/losing weight. So "consistency" or whatever you want to call it, doesn't make any sense.
Now, i'll grant there may be no other way to accomplish your goal right now, but suggesting this is a good method is wrong. It's a horrible method. This is like trying to optimize a multivariable constraint problem through random guessing.