Systems That Eat Themselves

John Gall's Systemantics is a book of laws about complex systems, written in the style of Murphy's Law but aimed at something deeper than pessimism. Where Murphy says "anything that can go wrong will go wrong," Gall says something more specific and more useful: systems develop goals of their own the instant they come into being, and those goals rarely align with the goals of the people who created the system.¹

The key laws, distilled:

A complex system that works is invariably found to have evolved from a simple system that works. A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over, beginning with a working simple system. This is Gall's law, and it's one of the most reliable heuristics in engineering and institutional design. The temptation is always to design the complete solution upfront. The reality is that the complete solution can only be reached by iterating from something simpler.¹

Systems tend to oppose their own proper function. Le Chatelier's principle, borrowed from chemistry: complex systems resist being pushed out of equilibrium. Apply this to institutions and you get a familiar picture — reform efforts that are absorbed, diluted, and neutralised by the very systems they're meant to reform. The system doesn't need to conspire against change; it just needs to keep doing what it was already doing, and the inertia is enough.

The system itself does not do what it says it is doing. This is the operational fallacy, and it's devastating. The educational system doesn't produce learning; it produces credentials. The healthcare system doesn't produce health; it produces treatments. The criminal justice system doesn't produce justice; it produces processing. In each case, the system's actual function has diverged from its stated function, and the people inside the system often can't see the gap because they're evaluated on the system's internal metrics rather than its external effects.

Why Systems Develop Their Own Goals

The most interesting Gall law is #19: "Systems develop goals of their own the instant they come into being." This isn't mysticism — it follows from the fact that systems are made of people, and people have their own goals. A committee formed to solve problem X immediately acquires the secondary goals of: continuing to exist, expanding its budget, maintaining its members' status, and avoiding blame. These secondary goals don't require anyone to be selfish or corrupt. They emerge from the ordinary fact that the people staffing the system have lives and careers that depend on the system's continuation.

This is closely related to Eliezer Yudkowsky's analysis of inadequate equilibria — the observation that institutions can be stuck in bad states that nobody individually has the incentive or ability to fix. But Gall is making a stronger claim: it's not just that systems get stuck in bad states. Systems actively develop goals that conflict with their purpose. The immune system of the institution attacks reformers as threats, because from the system's perspective, they are.

The Failure Mode Theorem

"Complex systems usually operate in failure mode." This is Gall's most counterintuitive claim, and probably his most important one. Most of the time, complex systems are not working as designed. They're working around some set of failures, compensating for broken subsystems, and producing outputs that are "good enough" through mechanisms that nobody fully understands. The system's apparent success is actually the result of accumulated workarounds, not of the original design functioning correctly.

This has a practical implication: when you try to optimise a system that's operating in failure mode, you may accidentally remove one of the workarounds that was keeping it functioning. The system crashes, and everyone is baffled because you only made a "small improvement." This is why large-scale reforms so often make things worse before they make things better (if they make things better at all).

Simon Sarris's "In Praise of the Gods" makes a complementary argument about what happens when rationalistic thinking is applied too aggressively to systems that work through mechanisms we don't understand. The Le Corbusier plan to demolish medieval Paris and replace it with cruciform towers was perfectly rational by the planning criteria of its era. It would have destroyed one of the most beloved urban environments in history. The medieval city worked because of centuries of accumulated, illegible wisdom encoded in its streets and buildings — wisdom that no planner could have specified in advance.²

Loose Systems Last Longer

"Loose systems last longer and work better. Efficient systems are dangerous to themselves and to others." This is the one that should keep engineers up at night. The instinct to optimise, to remove slack, to make every component maximally efficient is exactly the instinct that makes systems fragile. Slack isn't waste — it's the system's ability to absorb shocks, adapt to unexpected conditions, and survive the inevitable failures of individual components.

Gall's laws read like comedy, but they describe something real about the gap between how we design systems and how systems actually behave. The core insight is humility: complex systems are not legible to their designers, and the attempt to make them fully legible — to understand and control every interaction — is itself a source of failure.