Slack and Civilizational Fitness

Imagine a planet full of eyeless animals. Evolving eyes requires three parts, in order, each requiring a rare mutation. Each part is metabolically expensive with no intermediate payoff — you need all three or they're useless. Under intense competition, Eye Part 1 is a handicap: the animal that wastes energy on it dies. Under zero competition, nobody evolves anything useful. Eyes only emerge in the middle — where there's enough selection pressure that having eyes eventually reaches fixation, but enough slack that a temporarily costly intermediate step doesn't get you killed.¹

This parable, from Scott Alexander's "Studies on Slack," captures something genuinely profound about optimization. The deepest pits in the fitness landscape are often separated from you by hills. Perfect competition keeps you in local optima. You need slack — the paradoxical art of winning competitions by not trying too hard at them — to reach the global optimum.

Slack as Anti-Moloch

In Alexander's framing, slack is the complement of Moloch. Moloch is total competition: every individual actor optimizing ruthlessly, producing collective ruin. Slack is the breathing room that lets systems invest in long-term improvements. Neither is inherently better. Too much Moloch and you never evolve eyes. Too much Slack and you drift without direction. The trick is finding the right balance, and then — crucially — evolving the meta-system that maintains that balance.¹

This is a group selection problem. A species that chose to give itself slack would out-evolve one that didn't. But no individual animal can unilaterally choose to compete less — it would die. So one-layer evolution fails at producing slack for the same reason it fails at all group selection problems. The solution, when it works, involves two-layer systems: an outer layer of competition that evolves inner-layer rules constraining individual competition.

Multicellular life is the paradigm case. Individual cells "want" to replicate as fast as possible — we call this cancer. The organism evolves rules (shared genome, immune system, apoptosis, telomere shortening) that prevent unconstrained cell competition. These rules exist because organisms with better anti-cancer mechanisms outcompete organisms without them. The outer layer (organism competition) evolves the inner layer (cell rules) that produces coordination.

The same structure appears in human institutions. Countries constrain individual competition through laws. Companies constrain employee competition through policies. Universities constrain researcher competition through tenure. In each case, the outer layer (competition between countries, companies, or universities) selects for inner-layer rulesets that produce the right amount of slack.

Seven Examples

Alexander's essay walks through seven concrete cases where the slack/competition balance determines outcomes.¹

Monopolies exploit low-slack environments. Between you and "own a giant airplane factory" lies an enormous hill of investment with no intermediate payoff. Boeing only survives because nobody has the slack to climb that hill. Peter Thiel inverts the metaphor and calls the hill a "moat," but he's describing the same thing: monopoly power is the ability to ensure that potential competitors never have enough slack to reach the deep fitness pit of genuine competition.

Tariffs are deliberate slack injections. Japan protected its postwar auto industry with tariffs, giving Toyota enough breathing room to develop its revolutionary production methods without being crushed by temporarily superior American competitors. But Japan didn't seal itself off completely — once Toyota was strong enough, it competed on the global stage and eventually forced American automakers to adopt its innovations. Too much protection and the innovation never faces reality. Too little and the innovator dies in the cradle.

Bell Labs and Xerox PARC were slack factories. Freed from quarterly earnings pressure by monopoly profits, they could invest in research that wouldn't pay off for decades. Xerox PARC invented the personal computer, the mouse, Ethernet, and laser printing. Bell Labs invented the transistor, the laser, Unix, and information theory. But Xerox couldn't capitalize on its own inventions — it was too insulated from competition to feel the urgency. The slack that enabled the innovation also prevented its exploitation. This is the core tension: you need slack to invent and competition to deploy.

The Civilization strategy game makes the dynamic explicit. On a map with two civilizations close together and one isolated, the isolated civilization wins every time. The close neighbors spend the Stone Age hacking each other with axes (no slack, no long-term investment). The isolated civilization spends it building infrastructure, philosophy, and advanced technology. By the time the isolated civilization discovers the others, it crushes them. But note: the isolated civilization doesn't just have slack. It also has enough competition (against the environment, at least) to keep it innovating.

Eddie Lampert's Sears is the cautionary tale of slack removal. Lampert, reasoning that free-market economies outcompete planned ones, reorganized Sears as a set of internally competing departments that traded with each other at market prices. It was catastrophic. The departments hoarded resources, sabotaged each other, and the company imploded. The lesson isn't that internal competition is always bad — it's that the right amount of internal competition for a company is very different from the right amount for an economy. Lampert removed all the slack, and the system ate itself.

Gall's Laws and System Failure

John Gall's "Laws of Systemantics" provides a complementary perspective: complex systems almost always fail, and the ones that work invariably evolved from simple ones that worked.² You can't design a complex system from scratch — you have to start simple and add complexity incrementally, keeping the system working at each step. This is, in a different vocabulary, the same point about fitness landscapes: you can't jump to a distant optimum. You have to walk there through working intermediates.

Some of Gall's laws are eerily resonant with the slack framework. "Loose systems last longer and work better" — slack. "Efficient systems are dangerous to themselves and to others" — insufficient slack. "Systems develop goals of their own the instant they come into being" — Moloch, the misalignment between the system's evolved objectives and its intended ones. "Great advances are not produced by systems designed to produce great advances" — because systems designed for greatness optimize too hard, squeezing out the slack needed for genuine surprise.²

The deepest law might be: "A complex system that works is invariably found to have evolved from a simple system that works." This is Chesterton's Fence applied to entire institutions. Before you optimize a working-but-slack-seeming system, make sure you understand why the slack was there. It might be the thing that's keeping the system alive.

The Long-Term Stock Exchange

Wall Street itself is a slack machine — a "giant multi-trillion dollar time machine funneling future profits back into the past," giving inventors the slack to build things that wouldn't otherwise exist.¹ The Long-Term Stock Exchange, approved by the SEC in 2019, is an experiment in giving companies more slack than the standard market provides: investors get extra voting power for holding stocks longer, executives are incentivized toward far-future performance instead of quarterly earnings.

Whether it works is an empirical question. But the logic is clear: if the standard stock exchange creates companies optimized for short-term metrics — the corporate equivalent of the animal that never evolves Eye Part 1 because the metabolic cost is visible on the quarterly balance sheet — then an exchange with more slack might produce companies that reach deeper fitness pits. The outer layer (competition between exchanges) will eventually reveal which ruleset produces better companies.

This is perhaps the most optimistic implication of the slack framework. The balance between competition and slack isn't fixed. It's itself subject to evolution. If the right amount of slack is context-dependent, and if meta-systems can learn what context they're in, then the answer to Moloch isn't to abolish competition or to resign to it — it's to build two-layer systems that discover and maintain the optimal balance. Some of those systems already exist. We call them institutions, norms, and culture. The question is whether they can evolve fast enough to keep up with the new forms of competition that technology creates.