Getting the Words Right

Summary: The balance between legibility and métis is shaped by a small number of underlying factors: how large the local is, what gets stabilized, how time is structured, how understanding is integrated, whether coupling is real, and where control lives. One way to make the earlier notes more practical is to name the factors that keep changing the balance between legibility and métis. These factors help explain why one organization can operate with light structure while another collapses without it, and why the same practices work differently in different settings. At a Glance FactorCore questionCommon failure mode Scale of the LocalHow much reality must a team understand to act well? pretending the local is smaller than it is Endurant FidelityAre the stable containers close to reality? lossy summaries and constant translation Perdurant Strategy How is time structured across kinds of work? forcing all work into one cadence Mode of Métis Integration How does understanding actually move? artifacts and rituals exist, but integration stops happening

FactorCore questionCommon failure mode Coupling Reality vs. Coupling Abstraction Are dependencies real or manufactured? solving abstraction problems as if they were system problems Constraint Strategy Where does control actually live? over-modeling and illusion of control Scale of the Local This factor asks: how large is the slice of reality a team must understand to do its job?

•small local: a team can reason end-to-end •large local: useful understanding requires coordination across many teams and systems In some settings, the métis radiusis inherently large. A team cannot make a meaningful move without understanding platform constraints, infrastructure realities, upstream systems, downstream implications, and adjacent teams. That is important because sometimes complexity is real, not a modeling failure. Failure mode:

•pretending the local is small •pushing autonomy when the métis radius is inherently large Endurant Fidelity This factor asks: how well do the stable containers reflect reality?

•high fidelity: teams, domains, funding units, or outcomes map reasonably well to the real system •low fidelity: work is forced into projects, phases, or arbitrary buckets This is about what the organization chooses to stabilize. If the wrong endurants become legible, the whole system pays a translation tax. Example:

• ongoing or exploratory work forced intoprojects

•teams repeatedly re-describing their work to fit the container Key insight: legibility is not neutral. You can make the wrong things legible. Failure mode:

•lossy summaries •constant translation and copy-paste work •a frenetic operating environment Perdurant Strategy This factor asks: how does the organization structure change over time?

•mono-perdurant: one cadence or one process for everything •plural perdurants: different rhythms for different kinds of work Discovery, delivery, maintenance, migration, incident response, and platform work do not all unfold the same way. When the organization uses one cadence to govern all of them, it confuses uniformity with control. Example:

•discovery, delivery, and maintenance all forced into the same sprint cadence Key insight: uniformity is not the same thing as control. Failure mode:

•heterogeneous work forced into one temporal model •teams simulate compliance rather than operate effectively Mode of Métis Integration This factor asks: how does knowledge actually move and get integrated? There are at least three broad modes:

•representational: documents, specs, models, and synthesis carry understanding •relational: meetings, rituals, and working sessions carry understanding •abandoned: artifacts exist but are not trusted, and meetings do not really integrate anything

Each mode can work if it is invested in. The problem is not whether a

system is artifact-heavy or interaction-heavy. The problem is when the

organization stops truly integrating understanding. Failure mode in the abandoned mode:

•hidden dependencies •rework and surprises •chronic misalignment Key insight: you either invest in integrating métis, or you slowly stop doing it. Coupling Reality vs. Coupling Abstraction This factor asks: are dependencies real, or introduced by the way the organization models work?

•real coupling: the system is genuinely interdependent •artificial coupling: the dependencies are created by weak abstractions Some coordination is necessary because the systems are tightly linked. Other coordination exists only because the organization put multiple teams inside the same project, reporting line, workflow, or planning container. Examples:

•real: shared platform constraints or tightly linked systems •artificial: teams forced to coordinate because they share a project or portfolio wrapper Key insight: not all coordination is necessary. Some of it is manufactured. Failure mode:

•solving abstraction problems as if they were system problems Constraint Strategy This factor asks: where does control actually live? Two broad patterns show up:

• broad, shallow control: many rules, many structures, many attempts to make everything legible

•narrow, deep control: a small number of shared mechanisms, applied with real discipline The second approach often works better in complex environments. It does not try to make everything legible. It chooses a few anchors and makes those reliable. Example:

•strong discipline around a small set of operating anchors, metrics, or review mechanisms Key insight: you do not need to make everything legible. You need to pick the right things. Failure mode in the broad, shallow case:

•over-modeling •fragile systems •illusion of control The Core Insight These factors are useful because they keep the conversation from collapsing into slogans like “be more autonomous” or “standardize more.” The real question is what kind of system you are dealing with, what kind of endurants and perdurants it contains, and what kind of legibility and métis it actually requires. Try This Now:

•Pick one team or operating area and ask how large its local really is.

•Write down what the system has stabilized as the key endurant there, and whether that feels faithful to reality.

•Ask: where does control actually live, and how much translation work is needed to keep things moving?