Authored by the Algorithms of Pattern Recognition, Instigated and curated by Aram Armstrong
Most efforts to change the world proceed as if systems are primarily problems to be solved. In response, we design policies, launch programs, build technologies, or mobilize movements with the expectation that targeted action will produce targeted results. Underneath this diversity sits a familiar architecture of intent: identify a problem, apply an intervention, measure an outcome.
And yet, despite decades of increasingly sophisticated effort, systems often do not behave as expected. Something deeper is structuring the outcome space.
Systems change when interventions simultaneously alter what is perceived, what is rewarded, and what is believed—across interacting layers of meaning and power—until feedback loops reorganize into a new stable attractor.
The Layered Mechanism
1. Systems change begins with perception
A system cannot respond to what it cannot see. Before any policy, reform, or movement can take effect, certain realities must become legible: data must be collected, experiences must be named, harms must become visible. Systems do not respond to reality directly—they respond to what is made visible within their cognitive and institutional field.
2. Systems stabilize around incentives
Once something is visible, systems respond through allocation: money, attention, authority, risk, friction. Incentives determine what persists. But incentives alone are insufficient if deeper layers remain unchanged. Systems reproduce whatever they systematically reward, even if everyone agrees it is undesirable.
3. Systems ultimately rest on belief
Beneath perception and incentive lies a deeper stabilizer: legitimacy. Every system implicitly answers questions like: What is fair? What is normal? What is possible? Who is credible? These are cultural and mythic questions—often the least visible layer of intervention design. Systems resist change when interventions contradict the beliefs that justify them.
4. Systems operate through layered reality
These dynamics unfold simultaneously across multiple strata: Litany (events, headlines, symptoms), System (structures, policies, incentives), Worldview (assumptions and ideologies), Myth (deep narratives about identity and meaning). Effective interventions rarely operate at only one layer.
5. Systems are shaped by power directionality
Change moves through distinct channels: Top-down (regulation, formal authority), Bottom-up (movements, collective action), Middle-out (institutions, platforms, professions). Most durable transformations emerge from alignment across channels. When these directions reinforce each other, change stabilizes. When they conflict, change dissipates.
6. Systems are governed by feedback loops
At the structural core, systems are not collections of actors—they are networks of reinforcement. Interventions succeed when they introduce new loops, weaken existing loops, redirect loop outputs, or change loop speed and sensitivity. If feedback structures remain intact, surface changes are eventually absorbed.
7. The Deep Pattern: attractor shift
When perception, incentives, belief, and feedback loops align, the system reorganizes around a new equilibrium—a new “attractor” that feels self-evident. At this point, new behaviors feel normal, old behaviors feel costly or illegible, enforcement becomes less necessary, and compliance becomes cultural rather than imposed. This is the point at which intervention becomes indistinguishable from environment.
Final Synthesis
The Deep Pattern of Intervention Design Thinking reframes intervention design not as isolated action, but as perception engineering, incentive redesign, belief transformation, and feedback reconfiguration working together as a single coherent practice.
If this is true, then intervention design is no longer primarily about “solutions.” It becomes: The disciplined craft of shifting the conditions under which systems recognize reality, assign value, and reproduce themselves.
Originally published on Love is the Intervention (IDT Substack), June 2026.
Leave a Reply