The space of inefficiency
A weak system has countless ways to fail: use a poor model, miss information, waste energy or find the right answer too late. The space of all possible systems is huge, but the space near limiting efficiency may be much narrower.
Different intelligences may converge not because they copy one another, but because they independently approach the same limits of reality.
Decision, cognition and architecture
The weakest convergence is the same outcome. A stronger form is the recurrence of causal modelling, decomposition, prediction, uncertainty estimation and self-correction.
Near performance limits the physical implementation also becomes an object of optimisation: time, energy, memory, reliability and scalability become part of intelligence quality.
The optimum remains subjective
Complete knowledge of the world does not determine a choice on its own. Goals, needs, values and priorities of the particular subject still matter.
The attractor therefore does not imply one universal morality. It describes a reduction in the gap between what a system would choose with full understanding and what it can actually compute and realise.