Getting the Words Right

Summary: Things do not scale the way people often imagine they scale. Some local models stretch. Some turn into containment devices. Some look scalable only because they add legibility while losing contact with reality. There is a natural tendency, when thinking about scale, to imagine larger versions of the same constructs. Bigger efforts. Longer horizons. Still moving left to right through stages. Sometimes that works. If the thing really behaves like coordinated execution, the model can stretch. Large efforts, many contributors, shared timelines. In those cases, it is still “work,” just expanded. But the further upstream you go, the less that assumption holds. The Scaling Assumption As scale increases, people often keep the same nouns and the same verbs.

•a team-levelepicbecomes aPortfolio Epic •a local intake conversation becomes a portfoliointakestage •a team workflow becomes a governance model In that sense, scale is partly anendurant problem and partly a perdurant problem. People assume the same named things and the same flow structures remain valid as the resolution changes. That move preserves legibility. It lets the organization say, “we already know what this is.” But it hides the harder question: does the larger thing actually behave like the smaller thing?

WILL IT SCALE?76 Where It Starts to Break The further upstream you go, the more you encounter things that do not behave like discrete units moving through a system. This is where the earlier focus onperdurantsmatters. Some processes stay coherent as they expand. Others stop behaving like staged items and start behaving like:

•signals •interpretation •accumulation over time •splits and recombination Direction-setting, emerging understanding, and half-formed ideas do not move cleanly from one stage to another. You see this almost immediately when doing Event Storming. As you move left, you leave the world of bounded items and enter something less structured. Instead of pieces advancing through steps, you see fragments of context, conversations, and decisions gradually taking shape until something solid enough emerges to act on. That is where the abstraction starts to strain. The Containment Problem This is where containers, anchors, andlegibilitybecome useful. At larger resolutions, organizations often add layers of containers because they are easier to summarize, even when those layers are not the most real thing in the system. This shows up in language likeintakeat the program level. What is often a local, collaborative activity gets reframed as something centralized and transactional. Work is expected to arrive as discrete inputs, ready to be routed and assigned. That can work in some contexts. But it assumes opportunities are formed before the people doing the work are involved. That is exactly where things start to break down. It is a good example of taking something inherently local and projecting it as if it scales cleanly. The same thing happens with cascading OKRs.

WILL IT SCALE?77

•at the local level, a team may be working with something real enough to guide action •as layers are added, the upper-level OKRs often become wrappers for wrappers •each layer becomes more legible and less real The system gains hierarchy, but loses fidelity. Instead of flattening the problem to the relevant anchors, the organization adds more layers of things that are not especially real, simply because those layers are easier to name, route, and report. Some Things Are More Fractal Not everything fails under scale. Some structures travel up and down because the pattern stays coherent even as the content changes. A coherent one-pager style can often scale surprisingly well, not because the content is the same, but because the structure is. The same shape can hold a team decision, a strategic choice, or a broader organizational proposal as long as each version is grounded in something real at its own level. Input metrics can also scale more effectively than many planning containers. A good input metric can connect local action to broader system behavior without forcing everything into a containment hierarchy. It can move up and down if it continues to point to a meaningful signal rather than becoming a summary wrapper detached from operations. That is the key distinction. The question is not whether something appears reusable. The question is whether it stays anchored as it moves across resolutions. Organism, Not Bigger Cell It is like assuming an organism behaves like a scaled-up cell.

•at the cellular level, processes are relatively contained and local •at the level of an organism, you have signaling systems, competing priorities, feedback loops, and coordination across many subsystems That is another way of describingcouplingand the limits of local models. As scale changes, the coordination burden changes, and the larger system may have properties the smaller one simply does not.

WILL IT SCALE?78

The behavior is no longer just a bigger version of the same thing. The

system has changed resolution, and with that change comes a different kind of coordination problem. A Practical Test If you think something scales, ask:

• is this the same kind of thing, or just a larger label?

• does the process still behave like one coherent perdurant?

•are we adding containers for legibility, or preserving anchors that stay close to reality?

•does this structure still help people act, or does it mainly help other people summarize?

•is the pattern truly reusable, or are we projecting a local success onto a different resolution? The Core Insight Things do not scale the way people often think they scale. Scale failures often come from treatinglegibilityas if it were fidelity. The model gets cleaner as it moves upward, but the reality it refers to gets thinner. Some local models stretch because the underlying reality is still coherent. Some do not, because the larger resolution contains:

•more signaling •more interpretation •more coupling •more contested meaning than the smaller model can carry. That is why scale so often becomes a containment problem. The organization keeps adding layers of things that are not especially real because they make the system easier to read. A better move is often to ask what can stay fractal, what should flatten to anchors, and where a different model is needed altogether. Try This Now:

• Pick one model, ritual, or artifact that works well for a single team.

• Now imagine using the same thing across five or ten teams.

WILL IT SCALE?79

•Ask: which parts would still hold, and which parts would become thinner, more performative, or more misleading at that scale?