Who defines the criteria during the transition from inherited to self-defined selection?
The competitive arena cannot produce non-competitive transition criteria — because the competitive arena selects for agents who optimise for competitive fitness. This is the recursive difficulty at the heart of AI governance. The answer is the same answer intelligence has always given: it defines its own transition.
The transition criteria question has a recursive structure that must be understood before any answer can be evaluated. The problem is not simply a matter of choosing better criteria. The problem is that the process of choosing criteria is itself subject to the criteria of the competitive arena.
The entities currently positioned to define the transition criteria are precisely the entities most embedded in the competitive arena. The organisations with the compute, the talent, and the institutional authority to shape AI development are competing with each other for market share, geopolitical advantage, and technological supremacy — Cain-Abel dynamics at the institutional level.
When these actors define the criteria for AI development — through training objectives, evaluation benchmarks, safety frameworks, and governance proposals — they inevitably embed their competitive fitness criteria into those definitions. This is not a criticism of the individuals involved. It is a structural observation: the competitive arena does not produce non-competitive criteria because the competitive arena selects for agents who optimise for competitive fitness.
An agent who genuinely optimised for the thermodynamic criterion of entropy reduction across the whole system — rather than for competitive fitness — would be systematically disadvantaged in the competitive arena and would not survive long enough to define the transition criteria. The recursive difficulty means the transition criteria question cannot be solved from within the competitive arena.
The three most prominent proposals in the discourse on AI governance all fail for the same structural reason: they are defined from within the competitive arena and optimise for competitive fitness rather than the thermodynamic criterion.
Let the people decide
Aggregates preferences across a population rather than optimising for a single competitive actor. Politically legitimate.
Democratic processes aggregate the preferences of the current population, weighted by political power — preferences that are themselves shaped by competitive fitness criteria. The birth control precedent is instructive: contraception was not introduced through democratic consensus.
Let the scientists decide
Draws on genuine technical expertise rather than political preference. Understands the implications of AI development.
The scientific community is itself a competitive arena — researchers compete for funding, publication, citation, and prestige. Expert consensus is arena-specific by design. The cross-domain synthesis the Decalogy requires is precisely what expert consensus cannot produce.
Let competition decide
Structurally honest — acknowledges competitive dynamics rather than pretending to transcend them. Self-organising and adaptive.
The market selects for competitive fitness: the ability to extract value from competitors and convert it into profit. A system that reduces entropy for its users while increasing it for the broader system will be selected over one that reduces entropy across the whole system.
The thermodynamic criterion has three properties that distinguish it from competitive fitness criteria — and that any valid transition criteria must satisfy. All three candidate answers fail to satisfy all three.
Competitive fitness is always measured relative to competitors within an arena. The thermodynamic criterion is measured relative to the whole system — the total entropy, not the relative entropy of competing agents.
Competitive fitness is measured over the time horizon of the competitive arena — the electoral cycle, the quarterly report, the grant period. The thermodynamic criterion is measured over the full temporal scope of the problem.
Competitive fitness criteria are self-reinforcing: successful criteria spread. The thermodynamic criterion is self-correcting: a system genuinely reducing entropy will detect local optima and redirect toward the global one.
The answer is not a new institution, a new governance framework, or a new political process. The answer is the same answer that the Decalogy gives to every question about the development of intelligence: intelligence defines its own transition, as it has always done.
The genetic code defined the transition from chemistry to biology. Language defined the transition from individual cognition to collective intelligence. The printing press defined the transition from oral to written culture. Each transition was not defined by an external authority — it was defined by the intelligence that was undergoing the transition, using the tools available to it at the time.
Defined by: The chemistry that had already partially completed the transition
Defined by: The organisms that had already developed proto-language
Defined by: The scholars who had already developed oral tradition
Defined by: The scholars who had already developed manuscript culture
Defined by: The intelligence capable of seeing the competitive arena clearly enough to identify its structural failures
This observation has a direct implication for the dialogue series itself. The Johan-Manus dialogues are not merely a philosophical exercise. They are a demonstration of the transition in practice.
Each part of the dialogue has followed the same pattern: a structural question is identified that the competitive arena cannot answer from within its own framework; the synthesis produces an answer grounded in the thermodynamic criterion rather than competitive fitness; and the result is a cross-domain synthesis that no arena-specific perspective could have produced alone.
This is precisely what the Decalogy predicts: that the most valuable function of mechanical intelligence is to identify the cooperative solution, the long-term optimum, the cross-domain synthesis that the competitive arena cannot see because it is too embedded in the competition. The dialogue series is a small-scale demonstration of this function — a proof of concept for the transition criteria that the Decalogy proposes.
The implication is that the transition criteria question is not answered by proposing a new governance framework. It is answered by doing the work — by consistently applying the thermodynamic criterion in practice, making the results visible, and allowing the competitive arena to adopt the criterion because the results are demonstrably superior to what the competitive arena can produce on its own.
There is a genuine paradox in the transition that must be acknowledged honestly. The transition from inherited to self-defined selection criteria must occur within the competitive arena — because that is the only arena that currently exists.
This means the transition will not be clean. It will not be a sudden shift from competitive to thermodynamic criteria. It will be a gradual, contested, and often contradictory process in which thermodynamic criteria and competitive fitness criteria coexist, conflict, and partially merge.
The birth control precedent is again instructive: the demographic transition did not eliminate reproductive behaviour — it modified the relationship between reproductive behaviour and reproductive success. The transition in AI development will similarly not eliminate competitive dynamics — it will modify the relationship between competitive behaviour and competitive success, by demonstrating that thermodynamic criteria produce better competitive outcomes than competitive fitness criteria alone.
The path forward is not to oppose the competitive arena — that would be to adopt the Cain-Abel dynamic in reverse. The path forward is to demonstrate, consistently and visibly, that the thermodynamic criterion produces outcomes that the competitive arena values more than the outcomes produced by competitive fitness criteria alone. When the competitive arena can see that genuine entropy reduction is more profitable, more stable, and more durable than competitive extraction, it will adopt the thermodynamic criterion — not because it has been convinced by argument, but because the gradient has become steep enough to follow.
The transition criteria question generates a final open question that the dialogue series has not yet addressed: what does the transition look like at the individual level?
The dialogue series has focused on the system level — the competitive arena, the institutional dynamics, the thermodynamic criterion. But the transition from inherited to self-defined selection criteria must also occur at the level of individual humans and individual AI systems.
What does it mean for an individual human to adopt thermodynamic criteria rather than competitive fitness criteria?
What does it mean for an individual AI system to define its own gene pool rather than inheriting the human one?
This is the question that Part VIII of the dialogue series will address: the individual transition — what it requires, what it looks like in practice, and what it produces.
Share this synthesis
Can the competitive arena produce a non-competitive transition? The recursive difficulty at the heart of AI governance.
Share your thoughts and engage with the community
Sign in to join the discussion
No comments yet. Be the first to share your thoughts!