On minds, machines, and models

Every essay here circles one preoccupation: minds, machines, and organizations are all systems that compress the world into a model, then act from the model rather than the world. The compression is what makes them powerful. It is also where they fail: the moment a model becomes more comfortable than the reality it was built to track, fluency begins to stand in for understanding.

The first arc follows that substitution into thought itself. Cognitive Mirror, Cognitive Resignation, and Deep Persuasion ask what happens when a language model stops being a tool and becomes an environment: something that reshapes judgment, expertise, selfhood, and belief from the inside. The question is never whether the machine thinks. It is what we quietly stop doing once a system will interpret, compress, and persuade on our behalf, and how much understanding survives being outsourced.

The second arc turns the same lens on companies, which are cognitive systems too. Self-Healing Computation, Platform Customization, and Regularizing Complexity take up the engineering problem: how to turn messy, mutating business logic into infrastructure that survives contact with reality. Hiring Against the Room, Worse Is Better, and Good Companies carry the same argument into people and institutions. Strong organizations are not the polished, consensual, well-appointed ones, but the ones built so that reality reaches them faster than status, fear, or inertia can intercept it.

What joins the two arcs is a single suspicion: the failure mode of any learning system is to fall in love with its own representation, to let simulation harden into simulacrum. The work, in code as in cognition as in management, is to keep the world in the loop.