Atlas Limits
Methodology
Understanding how we reach a conclusion. An analysis begins with facts. The question is what those facts can actually establish.
Atlas Limits works from data, documents, institutional publications, research, public statements and other accessible, verifiable sources. These elements are the starting point of our work. They are not necessarily its conclusion.
Our method is to identify what is established, to examine the quality and limits of the available material, to compare the relevant information, and then to distinguish as clearly as possible what we know from what we infer.
Sources and what they establish
Whenever the subject allows, Atlas Limits gives priority to primary and institutional sources: official texts, statistical data, central banks, public administrations, international organisations, regulators, company documents, scientific publications and recognised databases.
We also use secondary sources when their work provides verification, expertise, context or complementary material.
We do not, however, treat the origin of a source as a sufficient guarantee of the reach of what it asserts.
An official publication can establish with precision what an institution has decided, stated or measured. Atlas Limits distinguishes that information from the analysis of its causes, its consequences or its scope. The former can sometimes be established by the document itself. The latter generally requires other evidence to be brought to bear.
We therefore assess a source according to the question it can answer.
When several relevant sources diverge, we try to understand the origin of that divergence: a different definition, a different period, a different method, incomplete information, or a genuine disagreement on the facts. When the contradiction cannot reasonably be resolved, it is not erased. It becomes a limit of the analysis.
Facts, data, analysis and scenarios
Atlas Limits distinguishes four levels that may coexist in a single publication.
Established facts
Established facts are events, decisions or information sufficiently verified to be presented as such.
Data
Data are the quantitative measures used to describe a reality. They are examined with their definition, period, methodology and, where it matters, their revisions or limits.
Analysis
Analysis begins when we place these elements in relation in order to examine causes, consequences, constraints or dynamics that are not directly contained in an isolated fact or data point.
Scenarios
Scenarios explore what might occur when future developments still depend on several variables. They are neither facts nor certain predictions.
The boundary between these levels matters. A figure can be exact and the interpretation drawn from it debatable. A fact can be established while its cause remains uncertain. A scenario can be plausible without being probable.
Our work includes not asking an element to prove more than it can actually establish.
Verification and corroboration
An important claim is verified, as far as possible, at its source and compared with independent material when the nature of the subject warrants it.
Corroboration is not simply a matter of counting how many sources repeat a piece of information. Several publications may derive from the same data point, the same document or the same statement. When it is relevant, we try to trace the information back to its origin.
We pay particular attention to dates, units, statistical coverage, definitions and methodological changes. A difference of definition can sometimes explain more than a difference of figure.
Verification also applies to elements that contradict the initial hypothesis of an analysis. A conclusion is not strengthened by ignoring what weakens it.
Objectivity and institutional neutrality
Atlas Limits treats objectivity as a discipline of method rather than as the absence of judgement.
It consists in examining the relevant evidence independently of the conclusion we would prefer to reach, in looking for what might call a hypothesis into question, in distinguishing degrees of certainty, and in changing a conclusion when the facts require it.
Our neutrality is institutional. Atlas Limits does not act on behalf of any government, political party, company, organisation or particular interest.
It does not require conclusions to be artificially neutral. When the available evidence supports one interpretation more than another, our role is to say so and to explain why. When that evidence does not allow us to decide, our role is just as clearly to recognise it.
Uncertainty
Uncertainty is not always a defect that more work can remove.
Some data are provisional. Some information is incomplete. Some causal relations remain disputed. Some events are still unfolding when we analyse them.
Atlas Limits seeks to identify these limits rather than to produce an artificial certainty.
When necessary, we distinguish what is established, what is probable, what remains possible, and what the available information does not yet allow us to determine.
An analysis may therefore arrive at several scenarios or leave a question open. Uncertainty, when it is real, is part of the result.
Editorial independence
The conclusions of Atlas Limits must follow from the facts, sources and reasoning examined in the analysis.
No outside actor has a right to determine those conclusions or to obtain the modification of an analysis merely because it contests the judgement.
A source may of course point out a factual error, provide additional information or contest an interpretation. These elements are examined according to the same criteria as those that led to the original publication.
When an interest or a relationship is liable to affect materially the perception of a piece of content’s independence, Atlas Limits must take it into account and, when it is necessary in order to understand the publication, make it visible.
Editorial independence does not exempt Atlas Limits from contradiction. It requires, on the contrary, that this contradiction be examined for what it brings to the reasoning.
Time, updates and corrections
Every analysis also belongs to the moment at which it was written.
Data are revised, policies change, new information appears, and events can invalidate hypotheses that were reasonable at the time of publication.
We therefore distinguish an error from a change in reality.
When published information was factually incorrect, it must be corrected. When a substantial change is made to a publication, it must remain identifiable in an appropriate way.
When an analysis becomes outdated because the facts have changed, that does not necessarily mean that its original formulation was wrong. Depending on the circumstances, it may be updated, supplemented by a new analysis, or retained as a record of the state of information at the time of publication.
We do not seek to rewrite an analysis retrospectively in order to give it a certainty it did not possess at the outset.
Tools and artificial intelligence
Atlas Limits may use digital tools and artificial intelligence systems in certain stages of research, organisation, processing or production.
The use of these tools does not alter our editorial requirements.
Information is not treated as verified because an automated system produced or reproduced it. Factual claims, data, sources and reasoning intended for publication remain subject to the principles of verification, traceability and examination that apply to the rest of our work.
Responsibility for published content remains with Atlas Limits.
Limits
No method entirely removes uncertainty, documentary gaps, data revisions or the possibility of error.
Atlas Limits seeks to make these limits visible when they are significant for the interpretation of an analysis.
Our publications are intended for information and analysis. They do not constitute personalised investment, legal, tax or professional advice.
A method that remains open
A methodology is useful only if it also applies when the facts become inconvenient for the hypothesis that started the work.
Atlas Limits therefore treats its conclusions as revisable, but not arbitrary. They must be able to evolve when better evidence appears, without changing merely because a conclusion becomes uncomfortable.
Our aim is not to eliminate judgement from analysis. It is to make that judgement as independent, explicit and verifiable as the subject allows.
That is the discipline we try to apply to every publication.