Horses make bad spreadsheets

Working with horses is a useful correction to anyone who likes clean operating models. You can measure feed, weight, workload, weather, turnout, water, and schedule. You should. The numbers matter.

And then the horse changes.

A feed that worked for months stops being right. Teeth change what can be eaten. Weather changes water intake. A small behavior becomes information about pain, stress, boredom, or nothing at all. Two horses in the same barn can require different answers.

Feedback is part of the system

Machines are not actually simple, but we often model them as though the input-output relationship is stable. Living systems are more obviously recursive. The thing receiving the intervention responds to it, and that response changes what should happen next.

You do not feed the plan. You feed the horse in front of you.

Standardization has a boundary

Standards are still useful. Feeding routines, safety practices, records, and shared expectations reduce preventable variation.

But the standard has to leave room for observation. A system that treats every deviation as noncompliance can miss the information the deviation contains.

Good care is not endless customization. It is the capacity to notice when the default no longer fits and respond before the mismatch becomes a crisis.

That is true in barns. It is also true in organizations. People, customers, markets, and communities change while the process is running.

The question is not whether to standardize. It is whether the system can remain attentive to the living thing the standard was built to serve.

Environment builders

I have described solutions architects as environment builders. That distinction matters to me. You cannot take one tool or application and bolt on solutions through optimism and force. A system demands an understanding of the environment in which the parts have to live.

The same is true of organisms. There is no formulaic way to build one without starting with essentials and relationships: what feeds it, what constrains it, what signals move through it, what it can repair, what changes when one part changes.

This is why I keep borrowing from biocentric design and autopoiesis when I think about organizations. The useful question is not only whether a system performs today. It is whether it can sense, respond, reorganize, and preserve enough structural integrity to remain itself while conditions change.

Optimization changes the thing being optimized

This is true in organizations too. Change a quota and people change behavior. Add a required field and people learn how to satisfy the field. Remove slack and the team compensates somewhere else. The intervention becomes part of the environment, and the environment changes the response.

That is why I am cautious with operating models that assume people are passive inputs. The model may be useful. It is not the organism. The useful question is what the system does after we touch it.

Observe before correcting

With horses, the fastest way to make a small problem larger is to decide too quickly what the behavior means. I have learned to look again: teeth, feed, weather, pain, routine, another horse, a new environment. Sometimes the answer is obvious. Sometimes it is not. The discipline is staying curious long enough to notice the difference.