Happy Tuesday, Transformation Friends. Another week, another opportunity to go Beyond the Status Quo.
A few issues ago, we looked at the Auditor General’s five spring reports side by side. We found a consistent theme: the visible parts of the work, the parts that were easy to count and show, got completed (mostly). The less visible parts, the ones that really determined whether the intended goals and outcomes were achieved, didn't.
In that issue, we touched on two concepts that might explain why this happens. Last time, we looked at Robert Merton’s goal displacement: the familiar way that a measure becomes the goal it was meant to represent.
Today, we’re looking at the second idea, which helps explain why organizations are vulnerable to that shift.
The idea is called legibility, and it’s central to James Scott’s 1998 book Seeing Like a State (Scott, 1998). It’s fairly intuitive and goes like this: government needs simplified versions of reality that it can record, compare, and act on. We see this every day in forms, surveys, categories, maps, dashboards and standard processes that reduce complex systems to something government can understand and manage.
This simplification is absolutely essential. First, a large public service could not function if it tried to understand every minute detail of a system; it’s just too complex. Second, it cannot tailor every benefit and service to every person’s exact circumstances, so it relies on shared categories and standards to provide comparable treatment across the country. Someone in Fredericton should be able to expect the same basic treatment from a federal program as someone in Victoria.
But, of course, there’s a risk to oversimplification. If we forget about the real world beneath it and mistake the simplified picture for reality, we lose all local knowledge and useful variation.
Scott begins with a real case about a forest oversimplified. A literal example of missing the forest for the trees. It’s an excellent illustration of the problem, so we’ll start there today.
Grab your morning coffee, and let’s get started.
How a forest became an administrative model
Scott starts with a story about German scientific forestry. It goes like this. From about 1765 to 1800, foresters in Prussia and Saxony developed methods to calculate the volume and revenue of timber in forests. To do this, they used a Normalbaum, or standard tree. Then, progressively throughout the nineteenth century, actual forest management increasingly made forests resemble the Normalbaum so they would more closely match the calculations. This meant fewer species, same-age stands, cleared underbrush, and perfectly straight rows.
From the state’s perspective, this worked quite well. Officials could count the trees, better predict the timber yield and issue standards that worked really well. All of this made harvesting easier and revenues more predictable. In essence, the forest on the ground began to match the forest in the books.
Although early results looked impressive, as time went on, officials began to see things the books had missed; things the simplifications smoothed out.
The thing is, the old forest depended on a number of key factors: nutrient cycles, decaying material, varied tree species, and symbiotic relationships among fungi, insects, plants, and animals. The standardization, which involved clearing and monoculture, completely disrupted all of this. The result was catastrophic. Growth declined, and the soil quality degraded. In addition, disease, pests and storms did more damage.
All of this did not go unnoticed by foresters. The effect was so prevalent that a new term was born to define it: Waldsterben, or forest death. There were significant efforts to reverse these effects, and another concept emerged: “forest hygiene.” This involved methods to reintroduce insects, birds, and other animals into the ecosystem.
Scott’s analysis here builds his entire thesis. The state has simplified its view to what it could measure, the Normalbaum, then reshaped the forest around that view. Everything outside that view was expendable until feedback from the forest revealed its true worth.
Scott opens with this forestry case as an illustration of legibility. He also puts this story inside a larger argument about high modernism: a deep belief in expert design, scientific progress, and the ability to order society from above. He goes on to look at cases of collectivization, Brasília, and Ujamaa villages. From these, he draws similarities and infers four conditions that must be present for these “planning disasters” to manifest: administrative simplification, high-modernist confidence, an authoritarian state willing to impose the plan, and a civil society unable to resist it.
Although contemporary public sector administration operates in a very different setting from 18th-century Germany, many of these factors persist today, and the main lesson remains valid: our “official” representation of a system or a group can begin reshaping the reality it was meant to describe.
Let’s pull on this a little more.
The knowledge that the system cannot fully record
Scott gives a name to what these simplifications miss. He calls it mētis and describes it this way. The practical skills and acquired intelligence gained through experience in changing circumstances. He notes that you can describe pieces of that knowledge, but its full value only appears in the moment someone uses it.
This reminds me of the contrast between a lesson identified and a lesson learned. It’s one thing to note something didn’t work and have an idea of how to improve, but it’s a whole other thing to try that idea and see the result.
Scott contrasts mētis with what he calls techne: the formal knowledge that can be stated as rules or principles, and verified and taught in standard form. Whereas mētis develops through experience in a particular setting, like repeated contact with a place, group, or process.
Most of our work in the public sector requires both. Formal knowledge (techne) provides consistency, transparency, and a basis for accountability. Practical knowledge (mētis) helps us apply those formal rules intelligently.
A good illustration of the difference between these two is the concept of work-to-rule. The French term grève du zèle, quite literally, “strike of zeal,” is much more colourful, especially when we consider the French definition of zeal: eagerness and dedication in serving someone or an idea; keen commitment to doing one’s work extremely well.
In work-to-rule, employees strictly follow written rules (techne) and withdraw things like judgment, adaptation, and cooperation (mētis). The effect is ironic: strictly following the rules as written is extremely detrimental. This tactic works because the “rulebook” doesn’t contain the whole job. It requires everything outside of it, too, to keep the work moving.
This directly connects to what we discussed last time about keeping rules and measures tied to purpose to mitigate goal displacement. When a rule handles a case poorly, people need the room to serve the purpose behind it. Mētis explains why this works because discretion is how practical knowledge reaches the formal system.
Of course, local knowledge also deserves scrutiny. Informal practices preserve bias, inconsistency and outdated ways of working. It’s the “we’ve always done it this way” monster rearing its ugly head. This is why we need both mētis and techne, with a meaningful connection between the two to enable them to correct each other: common rules protect rights and consistency, while feedback and discretion expose where those rules don’t work well.
A sharper picture can still be incomplete
I think Scott’s argument becomes more useful as government acquires better data. The collection and centralization of data through data lakes, real-time access and visualization, along with AI, can find patterns, process, analyze, and apply decisions at rates unthinkable to our 18th-century forester-administrators.
But they also make the official picture feel more complete than it actually is.
The OECD’s review of AI in government is a good place to see both sides. It looked at 200 real-world examples across 11 core government functions. It found that automation and modern tools support tailoring of services, better decision-making and forecasting, fraud and anomaly detection, and better job quality for public servants. However, they also create risks related to bias, transparency, explainability, and public accountability, especially when governments rely on poor data or weak human oversight (OECD, 2025).
In my view, the present-day extension of Scott’s argument is about confidence, not the technology itself. It’s the old saying from computer science: garbage in, garbage out. Bad data can be exposed with both mētis and techne. However, a polished dashboard built on bad data hides its blemishes and gives the impression that the data is good.
The dashboard is the Normalbaum. It’s a simplification that’s easier to manage than the underlying reality.
We need to ask ourselves, what is the forest telling us that the dashboard isn’t?
Uneasiness with this question should be the cue to step away from your screen to test the picture against the real world.
Keeping the picture open to correction
Back to Scott’s point about legibility: simplified versions of reality that can be recorded, compared, and acted on are necessary for effective governance. But discipline is important: we need to treat every simplification as a partial truth. When the feeling that it’s complete creeps in, this should act as a trigger to look deeper.
Three habits can help.
1. Name what is being left out
Start with the view that every simplification, standardization, or proxy leaves something out. Take time to think about what is being left out and the implications. For each, rate the impact and justify why leaving it out is acceptable.
Then use feedback to iterate and do this at effective intervals, such as right after key reporting milestones. Ask, “With my current simplifications, how good is my knowledge of what’s really going on, and is it still helping me to achieve the desired outcomes?”
Adapt and continue.
2. Build correction into the system
A simplification gets safer when the people with mētis can correct and improve it. Make sure the channel exists and assign someone to be accountable for it.
Test this by asking, “When was the last time someone on the frontline provided an improvement to our official picture?”
3. Go to the ground when the data looks perfect
This one follows from my philosophy on project dashboards. If it’s all green, it simply means you’re not looking hard enough for risks and issues; go back and look harder.
The more comprehensive the dashboard, the harder you should look at what it’s flattening and what you’re not seeing. It should prompt the question, “What are we missing?”
These habits are also well documented in the Government of Canada’s Digital Standards, which call for ongoing user research, frequent iteration, staff empowerment, and ethical service design. I encourage you to read them and put them into practice.
Wrap up
The techne gives government the common language it needs to be fair and accountable, by setting rules and standards. That common language and the simplifications it introduces will always describe a narrower world than the one people actually live in.
Our job is also to ensure that mētis, the practical knowledge techne leaves out, keeps the official picture open and honest.
Three questions to carry into your week:
For a simplification like a report or dashboard, what mētis is it failing to capture?
Where might a simplified picture be reshaping the work?
Who can tell you that the simplification is wrong, and who can correct it?
Until next time, stay curious and I’ll see you Beyond the Status Quo.
References
OECD (2025) Governing with Artificial Intelligence: The State of Play and Way Forward in Core Government Functions. Paris: OECD Publishing. Available at: https://doi.org/10.1787/795de142-en.
Scott, J.C. (1998) Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed. New Haven: Yale University Press.