For the past few years the lab has leaned on Complex Adaptive Systems, or CAS, as the frame that holds its three lenses together. This page makes a different case. CAS is not the newest form of systems design; it is the third era in a much longer story. The lab's real unit of analysis, the configuration, already belongs to a fourth era that CAS cannot quite name.
What follows maps those four eras, places CAS firmly inside the third, and proposes a name for the fourth: Entangled Systems Design. It is the wave-4 successor to the systems-design lineage that runs from Cross, Buchanan, and Owen through Suchman, Barad, Forlano, and Frauenberger. CAS is not discarded. It stays as the bridge that makes the framework legible to organizations, while Entangled Systems Design names the practice the lab already performs. This is ongoing work, offered as a proposal rather than a settled position.
Systems design has gone through three major reorganizations since the 1940s. Each reorganization redefined what kind of thing the system is, where the observer stands in relation to it, and what design is supposed to produce. Era 4 is the current rearrangement — the one the lab works inside but has not yet named in its own framework.
Read each row to see how a single dimension evolves; read each column to see how the era's pieces hold together. The shaded column — Entangled Systems Design — is where the lab actually works, and what its current framework does not yet name as its own.
| Dimension | Era 01 Hard Systems c. 1948 — c. 1970 |
Era 02 Soft Systems c. 1970 — c. 1992 |
Era 03 CAS c. 1992 — c. 2015 |
Era 04 Entangled Systems Design c. 2015 — present |
|---|---|---|---|---|
| A · Ontological commitment | ||||
| Unit of analysisWhat you study | Components & feedback loops Bounded subsystems with measurable inputs, outputs, and control variables. |
Stakeholder worldviews Multiple perspectives on a wicked problem, made commensurable through structured inquiry. |
Agents & their interactions Heterogeneous agents following local rules; the system's behavior is the emergent pattern. |
The configuration A phenomenon from which humans, technologies, and contexts emerge through intra-action. |
| Observer positionWhere the analyst stands | Outside, objective The system is observed without being disturbed; design is a view from nowhere. |
Inside the system Von Foerster's second-order move: the observer is part of what is described. |
A participant in adaptation The analyst is one more agent in the network, with limited foresight. |
Inside the cut The observer is one of the entities produced by the agential cut — not prior to it. |
| Entities are…The ontological question | Pre-given & bounded Classical objects with stable properties. |
Pre-given but interpreted The same entity holds different meanings for different stakeholders. |
Pre-given & interacting Agents are separable; what emerges is their interaction pattern. |
Produced through relation Nothing pre-exists its enactment. Boundaries are continuously negotiated. |
| B · Anchors | ||||
| Theoretical anchorsThe literature each era stands on |
Bertalanffy's General System Theory · Wiener on cybernetics · Forrester on system dynamics · Ackoff.
Bertalanffy 1968 · Wiener 1948 · Forrester 1961
|
Rittel & Webber's wicked problems · Checkland's Soft Systems Methodology · Von Foerster's 2nd-order cybernetics · Maturana & Varela on autopoiesis · Schön's reflective practitioner. Design's own systems lineage formalizes here.
Rittel 1973 · Checkland 1981 · Von Foerster 1974 · Schön 1983 · Cross · Buchanan · Owen
|
Holland's Hidden Order · Arthur on increasing returns · Kauffman on self-organization · Mitchell on complexity · the Santa Fe Institute. Brought into design by Teixeira & Forlano 2016.
Holland 1995 · Arthur 1999 · Kauffman 1993 · Mitchell 2009 · Teixeira & Forlano 2016
|
Suchman's Human–Machine Reconfigurations · Barad's agential realism · Forlano's posthuman design · Frauenberger's Entanglement HCI · Escobar's pluriverse · Wakkary's things-we-could-design.
Suchman 2007 · Barad 2007 · Forlano 2017 · Frauenberger 2019 · Escobar 2018 · Wakkary 2021
|
| C · What design does in this era | ||||
| Design's jobThe act of designing, named | Optimize the machine Reduce variance, minimize error, control output. The engineer's stance. |
Structure inquiry Surface assumptions, hold competing stakeholder logics, reflect-in-action. |
Intervene in the network Find leverage points, design for emergence, work near the edge of chaos. |
Produce diagnostic visibility X-ray the configuration; name what is being enacted; re-X-ray as the configuration changes. |
| What design producesThe deliverable | An optimized system Specifications, control loops, performance benchmarks. |
A workable accommodation Maps of stakeholder positions; agreed-on framings that hold for now. |
A resilient architecture Loosely-coupled, modular, capable of adaptation; "edge-of-chaos" design. |
An ephemeral X-ray A snapshot of an ongoing materialization, valid for the moment of its taking. |
| D · How the era breaks | ||||
| What the era assumedThe hidden commitment | That the system can be modeled without its modeler. |
That stakeholders can be enumerated, and that worldviews are the deepest layer. |
That agents are separable entities whose interactions produce the system. |
(we are still inside this era — its core assumptions are still being articulated) |
| What broke itThe threshold to the next era | Rittel's "Dilemmas in a General Theory of Planning" (1973) — wicked problems cannot be solved by hard systems. |
Nonlinear dynamics revealed system properties invisible to any stakeholder. Emergence does not appear on a SSM diagram. |
AI configurations dissolved the agent boundary. The "human in the loop" was never a stable agent — it was a configuration with a momentary cut. |
— (the era is open) |
The clearest way to see the difference: CAS still rests on a residual Newtonian backbone. It accepts that agents exist as separable units, then asks how their interactions produce emergent patterns. Era 4 — drawing on Barad's agential realism, Suchman's reconfigurations, and Forlano's posthuman design — refuses that separation at the ontological level.
A CAS reading and a configurational reading of the same AI-integrated team will describe overlapping phenomena — but they make incompatible ontological commitments about what is real.
Designers, AI, and managers are three kinds of agent. They follow local rules, interact across crosscutting levels, and produce emergent patterns no individual agent intended.
The agents are given. The system is what they make together.
Useful for: organizational vocabulary, leverage-point analysis, "no global controller" framing.
"Designer" and "AI" and "manager" are positions enacted through specific configurations. A designer who has spent a year working with Claude is not a designer-plus-tool; she is a different kind of entity, continuously re-cut by the configuration.
The agents emerge. The configuration is what produces them.
Useful for: identity-level analysis, the lab's actual unit, what wave-4 HCI and Barad's STS already commit to.
If CAS belongs to Era 3 and the lab works in Era 4, its framework carries a small but real tension. The proposal resolves it without throwing anything away. CAS stays on as the accessible systems vocabulary the lab shares with organizations. Entangled Systems Design names the wave-4 practice that the three lenses already demand. Neither one replaces the other. What the lab is really arguing lives in the relationship between them.
This is not only a story about systems theory. The lab's Configurational Probes method has its own lineage in four moves: think-aloud protocols, then contextual inquiry, then cultural probes, and now configurational probes. Set that sequence beside the four eras and the probes start to look like the methodological face of the same trajectory. The fourth wave of probes belongs to Entangled Systems Design. Method, paradigm, and integrating practice end up pointing in one direction, and that convergence is itself part of the argument.