What might the future of agriculture look like?

A while back I posted some speculation on the future of grocery stores, looking at how increasing technology might make for a much different grocery-shopping experience than found today. That piece briefly mentioned robots being deployed in places like vineyards, yet was primarily focused on issues of grocery warehouses, energy efficiency and the role of self-driving vehicles in delivery. Yet, as I think more about things there's a lot that seems to remain to be said regarding possible future innovation on the agriculture side of things. As I think more about it I wonder if the future means both an end to small farms and farming becoming both more organic and possibly a task that humans are no longer able to do efficiently.

At the moment, though a lot of people don't realize this, small farms currently seem to be more efficient than large-scale farms. To quote, e.g., George Monbiot:

an unexpected discovery ... was first made in 1962 by the Nobel economist Amartya Sen, and has since been confirmed by dozens of further studies. There is an inverse relationship between the size of farms and the amount of crops they produce per hectare. The smaller they are, the greater the yield.

Why might this be? To again quote George Monbiot:

The most plausible explanation is that small farmers use more labour per hectare than big farmers. ... With more labour, farmers can cultivate their land more intensively: they spend more time terracing and building irrigation systems; they sow again immediately after the harvest; they might grow several different crops in the same field. In the early days of the Green Revolution, this relationship seemed to go into reverse: the bigger farms, with access to credit, were able to invest in new varieties and boost their yields. But as the new varieties have spread to smaller farmers, the inverse relationship has reasserted itself.

Somehow I suspect that as agricultural robotics become more advanced this relationship is likely to reverse itself with the large farms becoming more efficient than their smaller counterparts again. My quick justification for this answer involves a mix of factory farming with modern warehousing, advanced logistics, advanced sensing, and advanced robotics.

Where I think modern warehousing practices become relevant is in the adoption of systems like chaotic storage. Chaotic storage is used by Amazon to manage its warehouse spaces and removes any inherent relationship between the type of object and its location in a warehouse (amongst a few other things). A TV might be stowed between a book and tins of black beans. Where systems able to support this come into play in the area of agriculture seems to me likely to be in the manner of reducing or eliminating the current monoculture approach in factory farming where large areas are dedicated to producing a single, specific crop. Robotics aren't quite ready for this yet, particularly in the area of agriculture but it seems to me that agricultural robots able to function outside of a monoculture are coming in the not-too-distant future.

If robots can function in a polyculture, what this seems to do is enable factory farms to adopt practices of sustainable agriculture - e.g. "multi-cropping, intercropping, companion planting, beneficial weeds, and alley cropping." Such practices can improve yields and reduce the need for the use of fertilizers. It seems to be the polyculture that gives small farms their boost, but large farms adopting such practices would seem to eliminate that small-farm advantage. Sensor networks allowing for micro-monitoring of soil conditions would seem to allow large farms to better select areas in which to plant particular crops. And then there's the issue of actually keeping tracking of everything, which is where it starts to become difficult to keep humans involved in running everything. If you've got, e.g., a total of 5 acres of rutabagas strewn in small plots or inidividually across 10,000 acres of farmland managed in a chaotic fashion, using sensors to determine the ideal spot to plant each, actually finding those rutabagas would seem to be something that a human worker simply can no longer handle. i.e. You need computers and - why not? - robots. Such robots could be scheduled in a fashion to allow them to maneuver through fields more efficiently - similar to how UPS estimated having cut down on fuel use by an estimated 10 million gallons by increasing route efficiency through means such as reducing the number of left turns taken.

At the moment monoculture-based farming seems to involve heavy use of fertilizer, but the resulting fertilizer runoff is messing up the oceans (and contributing to the jellyfish takeover of the seas as jellyfish function comparatively well in the resulting low-oxygen environments). Canada banned phosphates from home-use dishwasher detergents (but not commercial ones) though waste-water treatment plants already filter out the vast majority of the stuff and the stuff makes dishwasher operate more effectively.

Agricultural runoff of nitrogen-based fertilizers doesn't go through the same filtering process and, I'd argue is a much more appropriate issue for laws to address. Introduce pigovian taxation - which roughly translates to English in this instance as you-pollute-you-pay. This would seem to incentivize a move from heavy use of fertilizer. Polyculture-based agricultural practices would seem to help here. Ditto micro-irrigation. However there'd seem to be an incentive to try to build robots able to handle weeding as well as planting and harvesting (amongst the tasks robots are already being used for). Hiring people - even poorly-paid ones - to do this doesn't seem likely be to very cost-effective compared to the future of robotics.

In other words, you might wind up with large and largely organic farms operating without any humans. The makers of the LettuceBot claim that a single machine can replace about two dozen workers and improve production. The same folks are also working on robots capable of handling those weeding tasks. With even sweatshops finding robots cheaper than workers future wide-scale adoption of these things in agriculture seems more or less inevitable to me. Workers account for about 30% of greenhouse operating costs in Canada - might this mean slightly cheaper produce?

Mix in self-driving vehicles and the largely-automated grocery stores of the future that I was speculating about before and might, in the not-too-distant future, the first time a human touches a fruit or vegetable be when it's been delivered to your house and you're ready to eat it?