Founder's research notes — what Donella Meadows makes me ask about measurement, read alongside the agenda we have just set down.
A system cannot be understood by examining its parts independently.
The book — Donella H. Meadows, Thinking in Systems: A Primer
Edited by Diana Wright · Chelsea Green, 2008
The EVOLVRS Research series · A founder's reading alongside The Measurement Gap and the essays that followed
Why this book, now
I read to find language for things I already half-know. This quarter the book that did that was Donella Meadows's Thinking in Systems, and the line I kept circling was the plainest one in it: a system cannot be understood by examining its parts independently.
I have known that for a long time in the way you know something you cannot yet use. Meadows gave me the words, and once I had the words I began to see the same wall everywhere in our own work. Everything we have written this year — the position paper, the essays after it — has been, underneath, an argument about wholes and parts without quite naming itself that way. Meadows names it. She also explains why the difficulty is not a failure of tooling but a property of how systems are.
I came to this work through information inside human systems — first in healthcare, then through the plainer question of how people actually decide. Systems thinking is the older discipline standing behind both, and reading her felt less like learning something new than like being handed the grammar of a language I had been speaking badly for years.
A note on parts
Meadows builds her primer out of a few plain pieces — stocks, the things that accumulate; flows, the rates that fill and drain them; and feedback, the loops through which a system responds to itself.
The lesson underneath the vocabulary is that a system's behavior comes from how those pieces connect, not from the pieces themselves. You can know every element and still be surprised by the whole, because the surprise lives in the interactions. She is generous with examples, and every one of them is really the same example: the relationship, not the component, is the system.
I have started to think our whole measurement inheritance is organized against this. We measure departments, roles, individuals, quarters — parts, cleanly separated so they can be counted. The separations are what make measurement tractable, and they are also what make it miss. The thing Meadows points at — what happens between a decision and the work it lands on, between a technology and the people who must absorb it, between one function and the next — is exactly the part no single instrument was ever aimed at.
What unsettles me is that this is not a gap you close by measuring the same parts more precisely. More resolution on a part tells you more about the part; it tells you nothing new about the relationship. If Meadows is right that the relationship is the system — and in our work she keeps being right — then a measurement built for parts will only ever measure the wrong thing more accurately.
A note on blame
The idea in the book I find hardest to hold onto is that a system, to a large degree, produces its own behavior.
Put well-meaning people into the same structure and you tend to get the same results, which means much of what we read as individual performance is the structure expressing itself through whoever happens to be standing there. Meadows is patient about this. She keeps turning the reader away from who did this and toward what arrangement made this the likely outcome — not to excuse anyone, but because blame aimed at a person leaves the structure that produced the behavior fully intact.
This is uncomfortable for anyone who measures people, and we measure people. It suggests that a good deal of what organizations attribute to talent, or to failure, is the shape of the system showing through the individual. I don't take that as determinism — people are not merely outputs of their structure — but as a caution about attribution. When we saw, in the field this year, that where a leader sat in the organization explained more about their experience than their rank did, it did not surprise me the way it might have a year earlier. Meadows had already told me to expect the structure to be louder than the person.
The question it leaves on the desk is not a comfortable one: how much of what we measure as a person is really a reading of the system they are standing inside — and are we honest, when we report it, about which of the two we are actually describing?
A note on leverage
Near the end, Meadows offers her list of leverage points — the places you can push on a system, ranked by how much they actually change it.
The counterintuitive part, the part that stayed with me, is that the levers everyone reaches for first — the numbers, the targets, the parameters — are the weakest. The ones that move a system more sit deeper: its rules, its goals, and the flow of information about how the system is behaving to the people living inside it. She treats a missing information flow as one of the most common reasons systems misbehave, and restoring one as among the cheapest powerful interventions there is. Return a signal people could not previously see, and behavior shifts, without rebuilding anything physical.
That one stopped me, because it is us. It means measurement is not the neutral observation I sometimes describe it as. To measure something and hand that reading back to the people inside the system is to intervene in the system. We are not standing outside taking a photograph of an organization; we are changing it by letting it see itself.
If building an instrument is an intervention and not only an observation, then the responsibility is larger than accuracy.
It includes what a measurement does to the people it describes once they can see it — how it lands, what it invites, what it quietly makes harder. That is a design question, and an ethical one, not only a methodological one. I am still sitting with what it obligates. I suspect the sitting is the point.
What I'm keeping
It does not settle them. Thinking in Systems did not hand me answers about measurement; it made my questions better, and harder.
I am keeping three of them where I can see them. If a system is more than its parts, what would it take to measure a relationship as carefully as we currently measure a person? If structure produces behavior, how do we report on people without mistaking the system's shape for their character? And if measurement is an intervention, what do we owe the people we make visible to themselves?
None of these has a clean answer yet, which is why they are on the desk and not in the file. Meadows closes her own book not with control but with humility — with watching a system long enough to move with it rather than against it.
That is roughly where this leaves me: less sure that measurement is something we do to organizations, more convinced it is something we do with them. The agenda we have just set down is, I think, an attempt to earn that preposition.
Donella H. Meadows. Thinking in Systems: A Primer. Edited by Diana Wright. Chelsea Green Publishing, 2008.