Alex Tabarrok likes this Nick Szabo post "about the interaction between historical agricultural productivity and security."

...two large islands which have been largely or entirely protected from invasion for hundreds of years, Japan and Britain, also had among the highest agricultural productivities per acre during that period as well as the greatest cultivation of even marginal arable lands.... Contrariwise, this theory predicts agricultural productivity will be lowest in unprotected continental regions. Indeed, interior continental regions easily reached by horse tended to be given over to much less productive nomadic grazing. Security constraints were probably what prevented any sort of crop from being grown.

I don't think so. There were several prior constraints, not least the fact that they aren't very good places to do farming.

Besides, the premise is false. Britain was repeatedly invaded and conquered. That's one of the main aspects of British history. The Norman invasion, the Vikings, the Dutch, the French and a long tradition of raiding across the channel from various and sundry entrepreneurs as well as smugglers and misfits, in both directions, was continuous until the industrial revolution. Unlike the fairly homogeneous Japanese there is quite a lot of mixing in Britain, and not just Welsh, Scot and English - there is also Irish, French, every sort of Nordic and Germanic tribe, and Spanish genes in the mix. Many of them were invaders in the first place, driving the indigenous tribes inland and uphill. Japan and Britain aren't really much alike.

It also isn't true that agricultural productivity is lowest in unprotected continental regions. French farmers will doubtless dispute that, and might have done so for many hundreds of years if they hadn't been so busy being oppressed, with armies marching across them to get at one another from every direction.

It depends on the continent. In N. America, presumably a fine example of Szabo's thesis, climate is the determinant. Like many mid-continental areas variation can be quite extreme, with cyclical droughts that last decades followed by decades long periods of less intense variation. American settlers have nice documentary evidence of this since the 1860s when the nomadic horse and bison cultures were exterminated and the land opened to sod busters. At first they did pretty good but then droughts, fires, hail storms or something would drive many of them out even though they had the advantage of industrial technologies. Looking at the land ownership records of abandoned farms in Kansas often shows that as many as a dozen fools tried their hand at making it work, each going bust when the hard times came.

Some researchers, such as Wes Jackson, claim that the land actually supported more indigenous people who did not farm much, they gardened a little, than it supports now even with high technology and huge energy expenditures to pump ground water and transport inputs and outputs long distances.

There are some vivid examples of the climate cycles of the region if dry statistics aren't compelling. Robert Smithson's "Spiral Jetty" earthwork sculpture from the Great Salt Lake in Utah, built in 1970 near the end of a long drought period, soon submerged and remained that way for almost 30 years. Another drought period began in 1999 and has now lowered the lake level enough to see the art work again. The drought affected much of the inter-mountain west, not just the deserts of Utah, and seems to be related to long cycle variations in N. Pacific sea surface temperatures and jet stream tracks. When certain combinations of cycles happen to coincide they reinforce one another, and the climate makes large amplitude excursions from average conditions and stays that way for long periods. The "dust bowl" drought of the 1930s is another example, thought to be affected by Atlantic ocean cycles too.

There are some compelling arguments that these regions should be restored, that their net productivity is greater as grasslands, and that the early pioneers who had to cross them to get to the west coast were perhaps more correct in their judgement that this was the great American desert and no place for farmers. The diversion of rivers through major irrigation projects, the pumping of fossil water from underground aquifers that are being emptied at a rate far in excess of the recharge rate, the loss of top soil, the degradation of the rivers that have affects for hundreds of miles downstream including nutrient pollution as well as reduced flows of water and valuable sediment loads that maintain deltas, all speak against the idea that these lands should be farmed even though the soil is truly marvelous. The perennial native grasses were the ideal crop for such soils.

Similar stories can be told of other continental interiors. Szabo dismisses such considerations as incorrect though this is merely asserted not argued. It's not a convincing assertion. There are many contradictory examples. The preponderance of evidence is clearly contrary to his thesis.

Another interesting example comes from Asia where militant, nomadic horse cultures such as the Avars maintained high security for centuries, but didn't farm. Horses weren't just transportation and war machines - they were food, drink, clothing and shelter as well. They were wealth. And when the climate took a bad turn after a volcanic eruption cancelled summer for a couple of years (it is thought), they knocked on Rome's agricultural door and demanded tribute. Rome paid though it didn't save them in the end.