---
title: a claim is not a reading
canonical_url: https://ensurance.app/guide/a-claim-is-not-a-reading
markdown_url: https://ensurance.app/guide/a-claim-is-not-a-reading.md
subtitle: the sentence says what the place is for. the reading says what the place did. the gap between them is the useful part
category: ensurance
---

# a claim is not a reading

*the sentence says what the place is for. the reading says what the place did. the gap between them is the useful part*

A reach of creek rose, held, and fell last spring. Someone also wrote a sentence about that reach. A **claim** is what someone asserts — this reach is the buffer, we watch water level, the canopy improved. **Evidence** is what can be pointed at on the same ground: a boundary, a coordinate, or a reading with a time, a unit, and a method. A boundary and a coordinate show where. Only a reading shows what the place did.

Those two things get filed together and read as one thing. They are not one thing. A program officer, a disclosure lead, and an underwriter all want to know which of the two they are holding before anyone relies on it.

A creek, a soil profile, a canopy, and a storm already produce a signal whether or not a network stores it. A probe, a lab assay, a camera, a radar pass, and an indicator sheet are readings of that living system. [Ensurance](/?from=guide) keeps the reading beside what the place claims, so funding can cite the condition. It does not replace anyone's sensors, and it is not the living system.

:::johnson
**the gap is a budget, not a grade.**

The distance between the sentence and the reading is the most useful line in the file, because it shows where the next observation buys certainty instead of more language.

[the place keeps the record →](/guide/the-place-keeps-the-record?from=guide)
:::

## a claim names what is being relied on

A claim is a sentence about a place and what that place does. Three versions, from three desks that rarely sit at the same table.

A program officer reads a grantee report: the riparian buffer along two miles of the north reach was restored, and it is holding. A disclosure team writes its own line: this basin supplies the plant, and we monitor water level. An underwriting file carries a third: the wetland upstream lowers the loss expectancy on this book.

None of those is dishonest. Each is an assertion with an author, and that is most of its value: a claim tells a reviewer what is being relied on and who was willing to say so, which is more than many files state out loud. The trouble starts only when a claim is introduced in the register of a measurement, as though someone had stood in the water.

So keep the word. A sentence nobody can point at is still worth keeping, under its own name. The claim is where reliance gets declared, and labeling it as a claim is not a demotion.

## a reading is what the place did

A reading is one dated result about one place. It carries six things: the name the device or the lab already uses for the measure, the result, the unit, the time it happened, the method, and the place it belongs to.

A technician who walks to a staff plate at 7am and writes down 1.42 feet has produced a reading. So has a sonde on a logger, a lab sheet from a soil sample, a canopy figure from a satellite pass, and an indicator a steward has tracked on paper for eleven seasons. The instrument does not have to be new, networked, or ours. Bring the one you already read — the creek is not waiting on a different gauge.

What is not a reading: a number by itself. A spreadsheet cell that says 24 is not evidence of anything — 24 what, measured how, where, and when? A number with no method, no unit, or no place can be kept as a note. It does not count as evidence, and dressing it up as evidence costs the file its credibility with the one reader who checks.

## the gap is the signal

Put the sentence and the reading on the same ground. One of three things is true.

**Aligned.** The sentence and the reading are about the same place, the same property, and the same window, and the reading supports the sentence. Confirming that is the reviewer's work.

**Open.** There is a claim and no reading yet. This is the usual state in a nature file, and it is not a scandal. It is a priced question: what would it take to observe this, and who would do the observing?

**Contested.** There is a claim and a reading, or two readings, and they do not agree. A staff plate says one thing and a retrieval from a satellite image says another. A steward's season of observation disagrees with a lab sheet.

A gap is not a failure score. There is no blended trust number here, no automatic 1 through 10 that a probe writes, and no price a sensor sets. Two sources that disagree both stay visible, each with its method, and the person who has to sign decides. Collapsing a disagreement into one tidy figure destroys the only information it contained.

Read the three states as a budget, not a grade. For a foundation, the open rows are where a modest grant turns a sentence into something a later reviewer can check. For a company, they are the lines a disclosure team is carrying on narrative alone. For an insurer, they are where the file says *wetland* and nothing has been observed on that wetland in years.

## four things arrive looking alike

| what arrived | what to call it | what a reviewer can do with it |
| --- | --- | --- |
| a sentence about the place | a claim | ask who asserted it, and what they are relying on |
| a dated result with a unit, a method, and a place | a reading | go to the same ground and take it again |
| a result computed from other inputs | a model | check the label, then the number |
| a number with no unit, method, or place | a note | file it, and do not count it |

That is the whole discipline. Most confusion in ecological reporting is one of those four wearing another one's clothes.

## a model keeps its label

A model can be the right tool and still be a model. A soil-moisture figure retrieved from radar, a flood extent classified from a satellite image, a device-side call that something living was detected, a carbon number from a methodology — all useful, all computed from other inputs rather than observed at the place.

The label travels with the number. A model that arrives labeled can carry real weight, because the reader knows what kind of confidence to extend. A model that lost its label gets read as though someone stood in the creek with a tape, and the decision that follows is more confident than the evidence deserves.

Labeling costs nothing, which makes it the cheapest integrity control available. It is also why a satellite pass does not retire the plate in the channel. Two sources about one reach, and both are allowed to exist.

## what is checkable today, without a feed

Stated plainly, so nobody goes looking for a login that does not exist. Today a place can be named and held — an account in the place's own name, where its record lives — and a sentence can sit with that hold as a claim. The shape of a reading is decided — one dated result, with the device's own name, a unit, a method, and a place. The public writer is not switched on. There is no URL to post a feed to, no observation log on a public page, and no continuous refresh behind a password.

Holding a place does not prove the condition of that place, either. The hold is where the record lives. It is not a result. The location-data series already told the corporate reader this, and the shorter version is there: [the place keeps the record](/guide/the-place-keeps-the-record?from=guide).

## three questions that sort any file

1. **What was asserted, and who signed it?** That is the claim, and it should be legible as a claim.
2. **What can be pointed at on the same ground?** A boundary, a coordinate, or a reading with a time, a unit, and a method. That is the evidence.
3. **Which of those is a model, and does it still have its label?** If the label came off somewhere between the instrument and the slide, put it back.

Asked in that order, those three sort almost any nature file an auditor, a grantmaker, or an underwriter is handed. None of them needs a new sensor. All of them need the words to stay honest about which kind of thing they name.

## frequently asked questions

### what is the difference between a claim and evidence?

A claim is what someone asserts about a place: this reach is the buffer, the canopy improved, we monitor water level. Evidence is what can be pointed at on the same ground — a boundary, a coordinate, or a reading that carries a time, a unit, and a method. A claim has an author. Evidence has a place.

### what if the sensor and the claim disagree?

Both stay. The claim keeps its author and the reading keeps its method, and the disagreement itself is recorded rather than resolved by arithmetic. Two readings that conflict are also both kept. A conflict is information about the place, and about the instruments watching it.

### is there a trust score?

No. There is no blended number, no grade, and no automatic rating produced by a probe. A reader can see what was claimed, what was observed, and where the two do not meet. Judgment stays with the person who has to sign, which is where accountability already sits.

### can a model count as evidence?

A model can carry weight in a decision as long as it arrives labeled as a model. A retrieval, a classification, or a formula result is kept as what it is. What is not acceptable is a model briefed as a direct observation. The label is the condition of admission, not a disclaimer at the bottom of a page.

## what to do with the gap

Go find the gap in the file in front of you. Mark the sentences that are sentences. Mark the readings that carry a time, a unit, a method, and a place. Mark the models, and check that their labels survived the trip into the deck. Whatever is left over is a note.

That exercise costs an afternoon and needs nothing from us. If you want the record kept with the place rather than in a folder on a drive, start from a named place: browse the [natural assets](/natural-assets?from=guide) and find the watershed, the stand, or the floodplain your claim is actually about.

Read next: [what mrv actually is](/guide/what-mrv-actually-is?from=guide) — measurement, reporting, and verification are three uses of one reading, not three products to buy.

## the series

- [what your sensors already measure](/guide/what-your-sensors-already-measure?from=guide)
- [a claim is not a reading](/guide/a-claim-is-not-a-reading?from=guide)
- [what mrv actually is](/guide/what-mrv-actually-is?from=guide)
- [the radio stays outside](/guide/the-radio-stays-outside?from=guide)
- [what radar can see](/guide/what-radar-can-see?from=guide)
- [how to plug a sensor stack in](/guide/how-to-plug-a-sensor-stack-in?from=guide)
7. [what lidar can see](/guide/what-lidar-can-see?from=guide) — a laser times a return; height is that distance