Galatine Contact us

galatine / glossary

Glossary

The terms the lab uses in its notes, publications and tools. Each has a short definition and a link to where it is explained in full.

01

The lab's work

Horizon

The span of time a piece of research covers: near (up to two years), strategic (two to five) or far (five to fifteen).

In full →
Weak signal

Early, imprecise information about a change that may happen. Signals from every horizon go into one register.

In full →
Indicator

A sign whose appearance will confirm or refute an assumption. It is named in advance and has a date.

In full →
Task passport

The question of a piece of work, its deadline, the owner of the result and the completion criterion.

In full →
Project team

A temporary team for one question, with a task passport, a deadline and an owner of the result. It is disbanded after the release.

In full →
Expert network

The outside community of experts and partners around the lab. Its members join project teams according to their status.

In full →
Gate

A step of the cycle without which the work does not go on: the legal and ethical check before the work and the independent review before the release.

In full →
Analytical note

The form of the lab's result: nine blocks in a fixed order, from the key conclusion to the sources and the confidence.

In full →
Alternative analysis (red team)

The work of a separate group that looks for other explanations, hidden assumptions and signs of groupthink in strategic assessments and public conclusions.

In full →

02

Grades and checks

Reliability of a source

A letter from A to F: what the source's past says about how often it is right. F means there is no history to judge by.

In full →
Credibility of a message

A digit from 1 to 6: how well the message agrees with what is established. A 1 needs independent sources.

In full →
Probability scale

Seven words, from “almost no chance” to “almost certain”, each with its own range. The lab's conclusions use no other words for probability.

In full →
Confidence

The quality of the grounds for an estimate: high, medium or low. It is stated apart from probability.

In full →
Analysis of competing hypotheses

A method that weighs every explanation against the data at once and counts the evidence against each.

In full →
Forecast register

The record of every forecast with its wording, probability, horizon, author and date of the check, and its outcome after that date.

In full →
Calibration

How well the words of the probability scale match what happens, seen from the accumulated outcomes of the register.

In full →

03

Investigation

Node

A record in the graph of an investigation: a person in a public role, an organisation, a place, a document or a source.

In full →
Gap

What an investigation lacks: a relation without evidence, a place without coordinates, a conclusion the chain does not reach.

In full →
Verification task

The work a gap turns into. A closed task returns its result to the record.

In full →
Verification

Four marks under a record: the source is named, a document is attached, the verification task is closed, a second person has confirmed. The marks say how far the check went and say nothing about whether the record is true.

In full →
Material

The text that comes out of an investigation, with footnotes leading to the records it rests on.

In full →