---
title: the coastline gets longer the closer you look
canonical_url: https://ensurance.app/guide/the-coastline-gets-longer-the-closer-you-look
markdown_url: https://ensurance.app/guide/the-coastline-gets-longer-the-closer-you-look.md
subtitle: so the instrument has to work at every length
category: natural-capital
---

# the coastline gets longer the closer you look

*so the instrument has to work at every length*

The coastline paradox is this: the measured length of a coastline grows as the ruler you measure it with shrinks. A ruler a hundred kilometers long skips every bay. A ruler a meter long follows every rock. A ruler a centimeter long follows every pebble, and across that whole range of rulers the total keeps climbing. There is no single true length of the coast of Britain, only a length at a resolution. That finding has a direct bearing on what you can honestly say you want to ensure, and on how an instrument for a living thing has to be built.

The Crystal River, a reef, or a forest already exists and already works, whether or not anyone measures it. [**Ensurance**](/?from=guide) is how the thing you name gets funded to keep working. It is not the thing, and it is not the measurement.

## where the paradox comes from

Lewis Fry Richardson, a physicist and meteorologist, noticed while studying the causes of wars that neighboring countries published different lengths for the same shared border. He measured coastlines and frontiers with dividers set to different step lengths and found that the smaller the step, the longer the total, without converging on an answer. The work appeared after his death, in 1961, as an appendix to a study of deadly quarrels.

Benoit Mandelbrot picked it up in a 1967 paper in *Science* titled "How Long Is the Coast of Britain?" Richardson had already fitted the growth to a regular power law. Mandelbrot showed that its exponent could be read as a dimension, and that a coastline sits somewhere between a line and a plane. Eight years later he gave objects like that a name: fractals. The coast looks roughly the same at every zoom, and it keeps revealing detail as you go in.

That is the paradox. It is not a trick of bad surveying. It is a property of the thing being measured.

## nature's value behaves the same way

Try to total what a forest above a reservoir is worth. At the coarsest resolution you get a line: water supply. Zoom in and the line splits: filtration, flood attenuation, the snow it shades later into spring, the cold water that holds the trout, the soil that holds the slope. Zoom again and each of those splits: the fungal network under the roots, the beaver dams that slow the runoff, the insects the trout eat. Some of that detail is value the coarse line missed. Some of it is the same value counted twice, because the insects are part of why the trout are there. Either way, the total depends on the ruler you chose, and it does not settle on one number.

This is why the protocol does not try to total nature. A sum that grows with resolution is not a sum you can put at the bottom of a page and call the worth of a place. Instead the protocol values one thing it can bound: the **annual flows** a bounded place produces, set against the **cost of the stock** that produces them. The ruler is fixed: the protocol's fifteen ecosystem types and nineteen ecosystem services, listed on [the natural capital page](/manual/natural-capital?from=guide). A flow in a year, at that ruler, is a length at a stated resolution. It is honest about being a bridge to capital, not a claim about what the place is worth. [A valuation is not the worth](/guide/a-valuation-is-not-the-worth?from=guide) is the long version of that sentence.

:::johnson
**the coastline gets longer the closer you look. the instrument has to work at every length.**

So the protocol uses the same four parts at every scale, and lets you name the grain you can act on.

[zoom in on a place →](/explore?from=guide)
:::

## the design answer: the same four parts at every scale

If the thing you want to ensure has no single true size, the instrument cannot be built for one size. It has to be built so that the same grammar holds for a reach and for a basin, for an acre and for a continent.

The protocol's grammar is four parts. A **natural asset**: the bounded living thing. An **agent**: an account addressed to it, run by a person who signs. An **instrument**: a certificate if the thing is one named account, a coin if it is a theme too wide for one. **Proceeds**: the routing that moves what comes in along the relationships that exist. Those four do not change when the thing gets bigger or smaller. Only the nodes change, and the shape of the edges between them.

You can see the nesting in the catalog. On 13 October 2026 the protocol's namespaces held 844 accounts in `.ecoregion` and 186 in `.bioregion`, plus 620 in `.basin`, a mix of rivers, watersheds, and personal accounts. Natural assets nest in ecoregions, ecoregions in bioregions, bioregions in realms, and the account at each level is the same kind of object as the account at every other level.

Two live accounts make the point without a diagram. [colorado-river.basin](https://ensurance.app/colorado-river.basin) is an account addressed to the whole river, one of many that can serve it, and its own description reads 246,000 square miles, seven states and Mexico, water for 40 million people. [crystal-river.basin](https://ensurance.app/crystal-river.basin) is addressed to one tributary reach inside that basin, a stretch eligible for Wild and Scenic designation that runs past the towns of Marble and Redstone before it reaches the Roaring Fork. Same shape. Same four parts. Two very different lengths of ruler. Neither one is the wrong size to ensure.

## one graph, read at different grain

Under the four parts sits one structure: a dependency graph. A thing, and what it depends on, and what depends on it. The graph is fractal in the same sense the coast is. The grammar does not change as you zoom; only the nodes and the geometry of the edges do.

| grain | the nodes | the edges |
|---|---|---|
| ocean and basin | a river account, a watershed, a catchment | hydrology, sediment, baseflow, migration |
| parcel | a property, a built asset, the meadow above it | flood attenuation, pollination reach, microclimate |
| species | a herd, a pollinator, a trout population | habitat, corridor, food web, genetic flow |
| microbe and mycelium | a fungal network, soil biota | nutrient cycling, water redistribution, symbiosis |

Pollination decays over a few kilometers. Hydrology runs at the scale of a watershed. Genetic flow runs across a range and over generations. These are not four systems. They are one graph read at four resolutions, and the account, the certificate, the coin, and the proceeds are built to work the same at each of them. The map at [/explore](/explore?from=guide) draws the first layer of that graph: who funds whom, and which accounts share an ecosystem.

The same holds for trust. At every grain there is what an account **claims** about itself and what the world can **observe** about it, and the gap between the two is the trust signal, whether the account is a continent or a creek.

## so what do you want to ensure?

The paradox gives a practical answer to the series question. Do not wait until you can name the true size of the thing, because there is not one. Name the grain you can actually act on.

If you can walk the reach, name the reach. If you sit on a watershed council, name the watershed. If the thing you care about is a species that crosses a dozen watersheds, name the species and let the account addressed to it pay the reaches it needs. The larger shape still holds whatever you name. A certificate for the Crystal River does not have to settle what the Colorado is worth, and an account for the whole basin does not make the reach redundant. Both are the coast at a stated resolution.

What you should not do is pick the coarsest ruler because it sounds important. A theme too diffuse to address is a coin, and a coin funds protection indirectly, through trading. A reach you can name is an account, and an account can be paid directly, by the town below it, on a schedule. The finer the grain you name, the more precisely it can be paid.

Nothing in this post is investment advice or an offer of anything. It is a measurement problem and what the protocol built because of it.

## what to do next

Read [what the triple bottom line counts](/guide/what-the-triple-bottom-line-counts?from=guide), which takes the same question to the three columns of a sustainability report. Or go straight to the map at [/explore](/explore?from=guide), find the thing you want to ensure, and zoom until you reach the grain you can act on. If you already know the grain and want a person on the other end, [write to us](/contact?from=guide&topic=what-do-you-want-to-ensure) and name it in the first line.

## frequently asked questions

### what is the coastline paradox?

The coastline paradox is the finding that a coastline's measured length increases as the unit of measurement decreases, without converging on a single value. Lewis Fry Richardson documented it in work published in 1961, and Benoit Mandelbrot analyzed it in a 1967 paper in *Science*, later naming shapes with that property fractals.

### what does the coastline paradox have to do with funding nature?

Nature's value grows with resolution the same way a coastline's length does, so a total "value of a place" does not converge and should not be treated as the worth. The protocol values bounded annual flows against the cost of the stock, as a bridge to capital, and builds the same four parts, natural asset, account, instrument, proceeds, at every scale so a reach and a basin can both be ensured.

### how big does the thing I name have to be?

Any size you can act on. A reach, a parcel, a watershed, a species, or a basin can each have an account. If the thing is a theme too wide to be one account, it becomes a coin instead of a certificate. Larger shapes do not make smaller ones redundant; both are the same graph at a different grain.

## sources

Lewis Fry Richardson, "The problem of contiguity: an appendix to Statistics of Deadly Quarrels," *General Systems Yearbook* 6 (1961) — the measurements of coastlines and borders at different step lengths.

Benoit Mandelbrot, "How Long Is the Coast of Britain? Statistical Self-Similarity and Fractional Dimension," *Science* 156, no. 3775 (1967): 636–638 — the analysis of Richardson's data and fractional dimension. Mandelbrot coined "fractal" in *Les objets fractals* (1975).

## the series

1. [what do you want to ensure?](/guide/what-do-you-want-to-ensure?from=guide) — read first
2. [what is relational value?](/guide/what-is-relational-value?from=guide)
3. [the coastline gets longer the closer you look](/guide/the-coastline-gets-longer-the-closer-you-look?from=guide)
4. [what the triple bottom line counts](/guide/what-the-triple-bottom-line-counts?from=guide)
5. [form follows function: the agent, the certificate, and the coin](/guide/form-follows-function-the-agent-the-certificate-and-the-coin?from=guide)
6. [eight design principles behind ensurance](/guide/eight-design-principles-behind-ensurance?from=guide)