Agency Stratification of Strategy

The strategy and product lenses people accumulate as a toolbox are better organised by the kind of agency each one concerns than by their subject matter: some describe conditions you can only observe and adapt to (forces - given), some describe choices you make in response (positions - chosen), and some describe constraints on the order in which action is even possible (sequences - permitted). The claim worth naming is not the particular three-layer cut of any one framework but the more general principle behind it - that mistaking one agency-type for another is a distinct and recurring class of strategic error, and that this misclassification is diagnosable: a failure can be traced back to the agency-layer that was treated as the wrong kind of thing (a force fought as if it were a choice; an ordering constraint overridden as if it were optional; a choice abdicated as if it were a fixed force).

Forces - Positions - Sequences

This is stable enough to name for two reasons. First, it has independent diagnostic traction: the anti-pattern → skipped-layer mapping (Anti patterns reveal the skipped layer) holds across every layer of the worked example without forcing, which is what an organising principle should do. Second, it is not parochial to product strategy - the given/chosen/permitted distinction is a general statement about how agents relate to a system, which is why it sits naturally beside Experimental adaptation to complexity: that idea says how to act under irreducible uncertainty (designed, survivable experiments), while this one says how to classify the things you are acting on before choosing the posture. The two share the same realism about limited control - agency stratification draws the line between what is yours to choose and what is not, and experimental adaptation governs action on the part that is.

The idea inherits one honest qualification from its provenance. It was abstracted largely from a single synthesis (Strategy lenses stratify by agency, primary, high-confidence) rather than from many independent sources, so it rests on the Abstraction rule’s second branch (one primary, independently stable claim) rather than the first (convergent multi-source). It is also bounded by Strategy layers coevolve not cascade: the layers are distinct in kind of agency but not causally independent - chosen positions co-evolve the given forces - so the stratification is a sound diagnostic lens but not a predictive partition. Its current home is the diagnostic spine of Evolution stack; whether it earns stable depends on it proving useful on decisions outside the product-evolution context it was born in.