Moloch and the sandpile catastrophe

It often feels good to slide down efficiency gradients, but they can have tragedies at the bottom. Scott Alexander taught us to name this problem: he called it Moloch.

There is war in the Ukraine. The world’s largest wheat exporter, Russia, is fighting the world’s fourth-largest wheat exporter, Ukraine. Russia’s ability to get paid for its exports is under threat; Ukraine’s production has been so badly hammered that it will likely be a net wheat importer for years. As a result, there is a very strong near-term possibility that hundreds of millions of people in the Middle East, Africa, and Asia will starve. Even where starvation does not kill directly, political unrest and food wars may do it.

Understanding how we got here is important. Bear with me, I will get to a rationality heuristic, but there’s a story to tell first.

Pre-industrial societies were chronically vulnerable to famine because all staple food production was local and could be disrupted locally. Elites might import spices from the Indies but transport costs and risks were too high to allow long-distance food dependency to develop. This only began to change in the 1700s with the mass importation of sugar, tea, and coffee to Europe. Even though these were luxury goods that could have been foregone, one of the consequences of the trade was the first global war—the Seven Years’ War of 1756-1763.

The globalization of food production took its next major step in the mid-19th century when the developed world became dependent on phosphate fertilizers to perk up tired soils. Minor wars were fought over literal birdshit—guano islands were a critical phosphate source. But the phosphate flowed; none of those conflicts seriously disrupted it. The life-critical consequences of phosphate-supply disruption got consigned to the bin marked “That Will Never Happen”.

Then came the post-1945 Pax Americana, with the U.S. Navy guaranteeing global free trade. National economies went into a frenzy of optimization by seeking comparative advantages. Places where food production was expensive outsourced it to places where it was cheap. A population boom followed. Peaceful, steady global trade became life-critical to a large fraction of humanity in a way it had never before been in all of history. And nobody noticed this!

Nobody noticed this this because the Pax Americana was an actual pax—it successfully prevented major wars involving food exporters for 77 years. (The closest we came to an exception before 2022 was several brushfire wars between India and Pakistan.) Americans would have had trouble noticing it anyway since the U.S. is effectively food self-sufficient—we only import staple foods as a price-taker, not because we don’t have plenty of domestic capacity to produce them.

But the Russo-Ukraine war has changed everything. It can and will screw up life-critical international supply chains—Russia is the world’s largest phosphate exporter, too—but the U.S. can’t stomp on the problem because Russia has nukes. Awkward...

I said I’d get to a rationality heuristic. Might look like we’re far from one right now, but let’s look more closely at what could have been done if anyone had seen this coming.

There’s a answer pushed by various nationalist and populist types that says we should deglobalize, or should never have globalized in the first place. The problem with this prescription is twofold:

  1. In much of the world, it’s now impossible unless you’re willing for your population to die back to pre-1945 levels.

  2. Even in places like the U.S. it’s politically unstable. The cost of not outsourcing your food production whenever that reduces prices is that your people pay more for food. Especially, your poor people pay more for food. That’s a grievance that can never remain unexploited in a democracy

In both cases, the cost of being autarchic against shocks to the international food-supply chain is prohibitive. And even if you’re forced through that knothole by something like a major war, once peace and free trade are restored, the slide down the efficiency gradient leads right back to globalization—and fragility.

This model predicts that the future will consist of episodes of stable trade and falling food prices punctuated by food-supply-chain shocks causing massive loss of life. If we’re lucky, the stable episodes will be long and the catastrophes infrequent. We might not get lucky.

In the language of systems theory, complex adaptive systems want to be supercritical. Efficiency-seeking drives them to the sandpile catastrophe. This is not a happy thing to know, but it is an important thing to know.

If everybody knew this, and everybody knew that everybody else knew it, the way people plan for the future might change. It’s probably beyond hope to expect democratic political systems to maintain more than token buffer stocks, but individual resilience could be valued and pursued more. Mormons keeping a year’s supply of shelf-stable food in their basements might not be considered odd anymore, just normal.

It bears thinking on. Never forget that Moloch always lurks at the bottom of the efficiency gradient, pitiless, with “WHO SAW THAT COMING?” lettered around him in runes of fire and blood.