Agency, Error, and Reality
Humans are agents* - entities that can act. Agency is the capacity to act through the formation and execution of intentions.
Because our understanding of reality is always incomplete, errors are inevitable. An error occurs when we try to do something and it fails because our understanding of reality is wrong or insufficient.
Preserving and expanding agency therefore depends on learning from failure, on systems that detect errors, allow criticism, and revise ideas and practices in light of what does and does not work. This is how knowledge grows.
*How agency emerged, and how human agency became explanatory, is not fully known. The following is a possible explanation.
The Sun supplies a persistent low-entropy energy flux, driving matter far from equilibrium. Energy gradients dissipate, and the laws of physics restrict how matter can move, forcing dissipation to occur through specific interactions rather than arbitrarily. Transient ordered structures can arise as a consequence of these restrictions (Schrödinger, What Is Life?).
Most such structures arise only briefly. Some molecular processes, however, use energy flow to repeatedly copy their own structure. Information-encoding molecules enable repeatable, energy-driven copying. Replication arises from this repeatability, not from purpose or preference.
Copying with variation enables evolution. In constructor-theoretic terms, self-reproduction and Darwinian evolution are compatible with ordinary, no-design laws of physics when certain informational transformations are possible (Marletto, The Constructor Theory of Life).
Once replication with variation exists, informational structures that are copied more frequently become increasingly represented over time. Genes can be understood as such replicators, with organisms serving as transient vehicles for their propagation (Dawkins, The Selfish Gene).
Some systems evolve mechanisms in which internal chemical states trigger physical responses to environmental conditions, for example chemotaxis toward nutrients. Such mechanisms increase replication frequency by coupling sensed conditions to action. This is minimal agency: the control of action in relation to internal states and external conditions.
Brains evolved as increasingly flexible extensions of this capacity. They allow organisms to coordinate perception, memory, prediction, internal state, and action. This creates learning agents: agents that can revise behavior within a lifetime rather than only through genetic evolution.
In humans, this flexibility crossed a further threshold. Social learning, imitation, shared attention, and language allowed internal models to become public. Once ideas could be communicated, they could be copied, varied, criticized, recombined, and preserved across generations.
This created cultural evolution: a second process of variation, transmission, and selection operating on ideas, practices, tools, norms, and explanations rather than only on genes. When joined with criticism, this process becomes a powerful form of error correction.
Human beings are therefore explanatory agents. They can turn failure into problems, problems into conjectures, and conjectures into criticizable explanations.
A four-term Greek formulation of critical rationalism: certainty is denied; inquiry begins from conjectural belief; belief is subjected to criticism, testing, or refutation; and error is corrected through revision. Its Greek core is derived from Xenophanes of Colophon’s fragments B34–B35, Xenophanes lived c. 570–c. 478 BCE, roughly 2,500 years ago, possibly among the earliest surviving expressions of the fallibilist idea: humans do not possess τὸ σαφές, the clear and certain truth, but proceed through δόκος, opinion or conjecture, which may resemble truth without becoming certainty. The final terms, ἔλεγχος and διόρθωσις, extend the sequence in the spirit of Karl Popper’s critical rationalism, from conjecture through criticism/refutation to correction.