The Epistemological Problem
Intelligence analysis has an accountability problem. When a consultancy publishes a report claiming a company is "high risk," there is no mechanism to test that claim against future events. When a sell-side analyst upgrades a stock, there is no formal statement of what would prove the upgrade wrong. When a geopolitical risk monitor warns of "elevated tensions," there is no definition of what "elevated" means or what would lower the assessment.
The result is an industry full of unfalsifiable claims—statements that sound analytical but cannot be tested, verified, or disproven. Unfalsifiable claims are not analysis. They are rhetoric with a data appendix.
This matters because decision-makers act on analysis. If the analysis cannot be tested, the decision-maker cannot calibrate their confidence in it. They cannot distinguish between an assessment that has survived rigorous testing and one that was plausible-sounding but never examined. They are flying blind with instruments that have never been calibrated.
What Falsification Conditions Are
A falsification condition is a specific, observable, and testable statement of what would make a claim false. It is not a hedge or a disclaimer. It is a contractual commitment to the evidence.
Every claim Forreast publishes carries a falsification condition. The condition specifies:
- The observable event or data point that would invalidate the claim
- The timeframe within which the falsification would be observable
- The source from which the falsifying evidence would come
For example, if we assess that a company's supply chain is vulnerable to disruption from export controls on rare earth elements, the falsification condition might read:
This assessment would be falsified if the company's primary supplier maintains uninterrupted shipments through Q4 2026 despite the export control regime taking effect, as verified by shipping manifest data and port customs records from [specified ports].
This is not a vague "things could change." It is a specific, checkable claim: if shipments continue uninterrupted through a defined period, the assessment was wrong, and we will say so.
How the Methodology Works
Falsification conditions are not added after the fact. They are constructed as part of the analytical process, before the claim is published. This forces analytical discipline at the point of assessment—because the analyst knows their claim will be tested, they construct it differently from the start.
Step 1: Construct the Claim
The analyst identifies the specific assertion they want to make. Not "the company faces regulatory risk"—that is unfalsifiable—but "the company is likely to face antitrust enforcement action within 12 months based on its market concentration in [defined market] and the enforcement posture of [specific regulator]."
Step 2: Define the Falsification Condition
The analyst specifies what would prove the claim false. In this case: "This assessment would be falsified if no enforcement action is initiated by [regulator] within 12 months, or if the company's market share in [defined market] is confirmed below the threshold that triggers regulatory review."
Step 3: Specify the Evidence Source
The falsification condition must reference a specific, accessible evidence source. Regulatory enforcement databases, shipping records, corporate registry filings, sanctions lists—sources that are publicly verifiable or accessible through the WorldGraph. Internal Forreast assessments do not count as evidence sources.
Step 4: Set the Timeframe
Every falsification condition has a deadline. "Eventually" is not a timeframe. If the claim cannot be falsified within a defined period, the claim is too vague to publish.
Step 5: Publish
The claim and its falsification condition are published together. The reader sees both: what we believe, and what would prove us wrong.
Why Most Analysis Fails This Test
Apply this standard to the average intelligence report and watch it collapse. "Elevated geopolitical risk in the region"—falsified by what? "Growing regulatory pressure on the sector"—what specific pressure, measured how, falsified when? "The company's financial position is deteriorating"—deteriorating relative to what benchmark, falsified by what metric?
Most published analysis is unfalsifiable because falsifiability is hard. It requires the analyst to commit to a specific claim, define their terms, and accept the possibility of being wrong. It is much easier to write "the outlook is uncertain" than to write "we assess a 70% probability that [specific event] occurs by [specific date], falsified if [specific condition] is met."
But the easy path produces analysis that cannot be used. A decision-maker who reads "elevated risk" cannot act—they can only worry. A decision-maker who reads "we assess a 70% probability of regulatory action within 12 months, falsified if no action is initiated by [date]" can make a decision: hedge, divest, monitor, or accept the risk with eyes open.
Falsification in Practice: A Case Example
Consider a Forreast assessment from Q2 2026. We assessed that a logistics company's primary shipping corridor was at elevated risk of disruption due to pending regulatory changes in a transit jurisdiction. The assessment included:
The claim: The [corridor] faces a 60-80% probability of operational disruption within 6 months due to [specific regulatory change].
The falsification condition: This assessment would be falsified if throughput on [corridor] remains within 10% of Q1 2026 baseline through [date], as measured by port authority customs data.
The outcome: The regulatory change was implemented on schedule. Throughput dropped 34% within the assessment window. The claim held. The falsification condition was tested, not triggered.
Had throughput remained stable, we would have published a retraction and examined why the assessment failed. This is the discipline: we are accountable to the evidence, not to our prior conclusions.
The Meta-Point
Falsification conditions are not a feature. They are a philosophy. They encode the belief that intelligence analysis should be held to the same standard as any empirical claim: it should be testable, it should be accountable, and it should be honest about its own limitations.
The intelligence industry does not currently operate this way. Most published analysis is unfalsifiable, untested, and unaccountable. Decision-makers who rely on it are making bets with uncalibrated odds.
Forreast's commitment is simple: every claim we publish can be tested. If it fails the test, we say so. If it survives, your confidence in it should increase—not because we said so, but because the evidence said so.
This is the difference between analysis and assertion. This is why falsification conditions are on every claim we publish.
Forreast Intelligence publishes every assessment with explicit falsification conditions. Subscribe to the newsletter for monthly intelligence briefings, or request a capability briefing to see how falsification-condition methodology applies to your decision-making.