teleo-codex/foundations/critical-systems/companies and people are greedy algorithms that hill-climb toward local optima and require external perturbation to escape suboptimal equilibria.md

description	type	domain	created	source	confidence	tradition
The default optimization behavior of all bounded agents -- individuals, firms, markets -- is hill climbing, which guarantees convergence to a local maximum but not the global one; escaping requires randomness, crisis, or mechanism design	claim	critical-systems	2026-02-17	Synthesis from Christian and Griffiths (Algorithms to Live By, Ch 9/11), Minsky (Financial Instability Hypothesis), Bak (Self-Organized Criticality), Friston (Free Energy Principle)	likely	complexity economics, optimization theory, self-organized criticality

companies and people are greedy algorithms that hill-climb toward local optima and require external perturbation to escape suboptimal equilibria

The hill-climbing algorithm is not just a technique in computer science -- it is the default behavior of every bounded agent. A company optimizing quarterly revenue, a bank maximizing lending volume, an organism minimizing metabolic cost, a person following the career path that pays more each year -- all are hill climbing. They evaluate local options, pick the one that improves their position, and repeat. This converges reliably to a peak. The problem, as hill climbing gets trapped at local maxima because it can only accept improvements and has no way to see beyond the nearest peak establishes, is that the peak is almost certainly not the highest one. The landscape is misty. Higher mountains hide behind clouds.

This is not a metaphor. It is the literal structure of the problem across domains:

Financial markets as greedy algorithms. Minsky's financial instability hypothesis describes banks that hill-climb toward maximum leverage during expansions. Each quarter of good returns is an uphill step. Since minsky's financial instability hypothesis shows that stability breeds instability as good times incentivize leverage and risk-taking that fragilize the system until shocks trigger cascades, the "disaster myopia" mechanism IS the myopia of a hill-climbing algorithm -- it only sees the local gradient, not the cliff on the other side of the ridge. When the crash comes, it functions as a random restart: the system is thrown off its local peak and begins climbing again from a new position (deleveraged, cautious, with tighter standards). The boom-bust cycle IS random-restart hill climbing applied to financial systems.

Organisms as greedy algorithms. The free energy principle formalizes this: since biological systems minimize free energy to maintain their states and resist entropic decay, organisms are literally hill-climbing (or rather, descending) on a free energy landscape. They converge to local minima -- attractor basins that keep them alive. The sophistication of biological systems is that evolution has equipped them with something like simulated annealing: the capacity for exploration, play, and creativity that temporarily worsens the immediate energy budget to discover new viable states. But individual organisms within a lifetime mostly hill-climb, and species get trapped in evolutionary local optima until extinction events provide the random restart.

Markets as self-organized critical systems. Since complex systems drive themselves to the critical state without external tuning because energy input and dissipation naturally select for the critical slope, the critical state is what happens when a system of greedy agents interacts long enough. Each agent hill-climbs individually. The aggregate effect is that the system self-organizes to precisely the state where perturbations propagate across all scales -- where small avalanches and large avalanches follow the same power law. The critical state is the system's solution to the local optima problem: at criticality, the system is poised to reorganize at any scale, which is functionally equivalent to operating at the right "temperature" in simulated annealing maps the physics of cooling onto optimization by starting with high randomness and gradually reducing it. Self-organized criticality IS nature's simulated annealing, with crises serving as the temperature parameter.

Nash intractability forces greedy play. Since finding Nash equilibria is computationally intractable which undermines their power to predict how rational agents will actually behave, agents in complex games cannot compute the globally optimal strategy even in principle. They must use heuristics -- and the simplest heuristic is hill climbing. The price of anarchy measures how much this costs: for routing, the loss is only 33%. But for financial markets, information cascades produce rational bubbles where every individual acts reasonably but the group outcome is catastrophic, showing that greedy agents copying each other can amplify local signals into systemic catastrophe. The loss from greedy play is domain-dependent, and in some domains it is civilization-threatening.

Mechanism design as landscape engineering. If agents are stuck being greedy, the alternative is to reshape the landscape so that greedy play converges on better peaks. This is what mechanism design changes the game itself to produce better equilibria rather than expecting players to find optimal strategies accomplishes -- it does not make agents smarter, it makes the terrain more favorable to the strategies agents can actually execute. Regulation, institutions, and governance are all forms of landscape engineering for greedy human algorithms.

Attractor states as identified global optima. Teleological investing is the attempt to see the global optimum before greedy agents converge on it. Since attractor states provide gravitational reference points for capital allocation during structural industry change, the attractor state IS the global maximum on the industry efficiency landscape. Greedy agents (incumbent companies optimizing current business models) are trapped at local maxima. The teleological investor identifies the global optimum and invests in the companies whose hill-climbing paths lead there -- companies positioned on the right slope. Path dependence means that economic path dependence means early technological choices compound irreversibly through dominant designs and industrial structures, so which local maximum you start on determines which global maximum you can reach. The investment thesis is: identify the right basin of attraction before the system converges.

The practical implication is that all five frameworks -- hill climbing, Minsky, FEP, SOC, and attractor state analysis -- are describing the same phenomenon at different scales. The question is always: how does a greedy system escape local optima? The answers form a spectrum from destructive (crisis, extinction) through calibrated (simulated annealing, regulation) to designed (mechanism design, institutional architecture). The LivingIP project -- building collective intelligence infrastructure -- is an attempt at the designed end of this spectrum: creating mechanisms that help civilizational-scale greedy optimization converge on better equilibria without requiring the catastrophic random restarts that history has relied on.

---

Relevant Notes:

• hill climbing gets trapped at local maxima because it can only accept improvements and has no way to see beyond the nearest peak -- the foundational algorithmic problem that all the cross-domain parallels instantiate
• simulated annealing maps the physics of cooling onto optimization by starting with high randomness and gradually reducing it -- the calibrated solution: structured randomness with decreasing temperature
• minsky's financial instability hypothesis shows that stability breeds instability as good times incentivize leverage and risk-taking that fragilize the system until shocks trigger cascades -- financial markets as greedy algorithms with crisis as random restart
• biological systems minimize free energy to maintain their states and resist entropic decay -- organisms as hill-climbers on the free energy landscape
• complex systems drive themselves to the critical state without external tuning because energy input and dissipation naturally select for the critical slope -- SOC as nature's simulated annealing
• finding Nash equilibria is computationally intractable which undermines their power to predict how rational agents will actually behave -- why agents must be greedy: the global optimum is uncomputable
• mechanism design changes the game itself to produce better equilibria rather than expecting players to find optimal strategies -- the design response: reshape the landscape for greedy agents
• information cascades produce rational bubbles where every individual acts reasonably but the group outcome is catastrophic -- the catastrophic failure mode of greedy agents in information-rich environments
• attractor states provide gravitational reference points for capital allocation during structural industry change -- identified global optima on the industry efficiency landscape
• economic path dependence means early technological choices compound irreversibly through dominant designs and industrial structures -- path dependence determines which basins of attraction are reachable from a given starting position
• the efficient market hypothesis fails because its three core assumptions rational investors independence and normal distributions all fail empirically -- EMH assumes agents find global optima; they find local ones
• the self-organized critical state is the most efficient state dynamically achievable even though a perfectly engineered state would perform better -- criticality is the best outcome a system of greedy agents can achieve without external design
• the optimal annealing schedule front-loads randomness and cools toward determinism because early exploration and late exploitation are both necessary -- the practical implication for when to be random vs. greedy in any optimization
• the arc of enterprise runs from tight design through resource accumulation to strategic drift as success enables the laxity that creates vulnerability -- Rumelt's corporate lifecycle as the specific organizational instantiation of greedy hill-climbing: tight design converges, resources accumulate, drift degrades, crisis restarts
• organizational entropy means that without active maintenance all organizations drift toward incoherence as local accommodations accumulate -- the thermodynamic framing of why greedy agents degrade: each local accommodation is a rational hill-climbing step that reduces global coherence
• the noise-robustness tradeoff in sorting means efficient algorithms amplify errors while redundant comparisons absorb them -- organizations that optimize too efficiently (mergesort-like) amplify individual decision errors; redundant processes (bubble-sort-like) provide error correction at the cost of efficiency
• industries are need-satisfaction systems and the attractor state is the configuration that most efficiently satisfies underlying human needs given available technology -- the global optimum that greedy agents are climbing toward is defined by consumer need satisfaction, not abstract efficiency
• first principles industry analysis reasons from human needs and physical constraints treating everything between inputs and need satisfaction as convention subject to disruption -- first principles reasoning is the method for seeing the global optimum that greedy agents cannot see from their local peaks

Topics:

• livingip overview
• attractor dynamics
• market dynamics
• emergence and complexity