Resilience, Failure and Redundancy

A farm is resilient in the season it is tested and not before. Every system looks sound in a good year, which is why the only useful conversation about resilience is a list of things that actually broke and what it cost. This lesson is that list, and then the price of the backups that would have caught each item.

In Meadows' terms, resilience is a property of a system's structure: redundancy, diversity, feedback, and buffers. It is not a yield figure. The farm's own synthesis puts it the same way: resilience emerges from structure, and a farm optimized for peak yield has traded that structure away, the position the global food system is in with four crops carrying 60 percent of its calories.

What actually breaks

Water, first. Pelman left his farm for a little over two months to teach in China, his first departure in seventeen years. The irrigation failed mid-summer and the vegetable garden dried out. It recovered, but the episode is the whole lesson: a one-person system with one water line has no redundancy, and the failure arrived on the first day the person was not there.

Weather and animals. In his worst year, wild boar took 70 percent of Pelman's crops. He absorbed it, and the logistics note names why that mattered: he lives in a market economy and could buy his way out. A household or a community feeding itself has no such fallback, which is why the loss must be planned for before it happens.

Pests in the store. The weevil is the second harvest. Columella warns against the instinct to winnow infested grain in the bin, because the insects are mixed through the whole mass instead of staying where they bred. The store you built to survive the winter is a food supply you can lose in a warm week.

Animals. Disease, predators, and a failed hay year all take the same animal. Livestock that converts forage into food is also a second set of ways for the farm to lose.

Health. The farmer's body is the farm's only engine, and it ages. A system that needs a strong back at the top of a hill has a scheduled failure date, so the design puts the work where a seventy-year-old can still reach it.

The catastrophic single loss. The store burns, the cellar floods, the roof goes. These are rare and they end the household's food year in a night, which is why the farm's own store chapter sets two years of grain, beans, and oil as the goal. A two-year store is not paranoia. It turns a bad year into a hard year.

Diversification is real and it is measurable

The model engine ran the comparison. A wheat monoculture on one hectare returns a base of 2.86 people supported, an expected 2.37 under the failure scenarios, and a worst case of 2.00. The diversified mix, the same one from the fat lesson, returns a base of 3.70, an expected 3.26, and a worst case of 2.93. The diversified system is better in the good year and better in the bad year, and the gap between its worst case and the monoculture's worst case is the value of diversity stated as a number.

Two honest limits belong with that result. The 30 percent loss and the diversification bonus are illustrative parameters, and the model does not count a single drought that hits every crop at once. Correlated failure is the case diversity does not answer, and it is the reason the water and the store matter as much as the crop mix.

Redundancy that costs something

The redundancy worth having shows up in the account. A second water source, dug before the first fails. A second way to preserve a crop that needs no electricity, so the harvest has somewhere to go when the grid does not. Seed kept out of the food store, so a bad winter cannot eat next spring. The two-year grain reserve. A neighbour who knows where the key is. The farm's resilience note ranks community first for good reason: one person inside a network of twenty households is more resilient than a family of four alone, because community is the redundancy no household can stock by itself.

Two ploughs, oxen picked for calm, a driver picked for steadiness. Hesiod is buying the redundancy that keeps the ploughing inside the autumn window, and paying for it in timber rather than in a lost sowing. Redundancy costs something real, and it is cheaper than the failure it prevents, bought before you need it.