[roel_v]

I meant 'yield per input unit', where 'input unit' is a not exactly defined mix of land, labor, external supply of matter, tools and technology/know how. Sure you can grow an acre of wheat to maximize 'caloric yield' (or potatoes depending on how much of the processing-before-consumption you account for...), or an acre of 'organic basil' to get a maximum 'pecuniary yield', or another thing or mix of things to maximize for whatever optimum function you choose. My point is that a 'solution' where we can theoretically feed 7/8 billion people if only double-digit percentages of them, say, pick apples from full size trees, careful to not damage the berries underneath with their ladders, instead of having a few people on tractors riding through dwarf orchards (just one example of something I happen to have done some economic analysis of the last week - there are similar examples in other areas of agriculture) is not really a 'solution' at all.

Now, I do agree with you that with even better technology than we have now, we can replace much of what we need chemistry for with mechanical solutions. Although, chemical application with modern systems can be dosed on areas measured in square feet, a long way from the 'a bit of nitrogen is good so double the amount is better' mentality of a few decades ago. But the future of agriculture is not in 'back to land' or 'permaculture' style farming. It's in more technology, not less, and less labor, not more.