Why we do not publish verdicts

Software that reads the news for you is tempting to build as an answer machine. Feed in the reporting, return a conclusion: true, false, likely, escalating. It demonstrates well. It is also the wrong shape for people who answer for their decisions.

The reader owns the conclusion

An analyst who briefs a decision-maker is accountable for the judgement, including its reasoning. A tool that hands over a verdict cannot be questioned in a meeting, because there is nothing behind it to point to. The analyst is left either repeating a conclusion they cannot defend or redoing the work to find out whether it holds.

Verdicts hide uncertainty

Most geopolitical claims are neither true nor false in a clean sense. They are partly supported, supported only by one origin, accurate in fact but misleading in framing, or too early to call. Collapsing that range into a label throws away exactly the information a careful reader needs. A claim that is “likely” on thin evidence and a claim that is “likely” on strong evidence look identical once they have been rounded.

What we do instead

• Describe the state of the evidence in plain terms: what is reported, by whom, with what support.
• Show the inputs behind every score so that a reader can weigh them differently.
• Separate what is established from what is attributed and from what is inferred.
• Leave a clear place for the analyst’s own assessment, kept distinct from ours.

A calm voice is part of this

Declining to issue verdicts has a consequence for tone. The writing stays level. There are no alarms where the evidence is thin and no reassurance where it is not. Where something is uncertain, we say what would settle it.

We describe. You decide.

This is slower than a one-line answer. It is also what lets a team trust the record enough to build on it.