A score with no published method is decoration. Ours is an editorial judgment expressed on a fixed scale, and the scale is defined here so that readers can disagree with a specific number rather than with the idea of numbers.

The four dimensions

Each story is assessed on four dimensions, scored 0–10, then weighted. The weights are fixed and do not vary by section.

• Reach (30%), how many people or organizations are materially affected, not how many will read about it.
• Durability (30%), whether the effect persists past one news cycle. Announcements score low; shipped changes score high.
• Irreversibility (25%), how hard the change is to undo. Regulation and infrastructure score high; pricing scores low.
• Evidence (15%), the quality of what the claim rests on. Independent replication scores high; a vendor benchmark scores low.

The bands

The band printed above the number is a plain-language reading of it, not a separate judgment.

• High impact, 8.5 and above. Consequential now.
• Developing, 7.0 to 8.4. Consequential if it holds.
• Watch, 5.5 to 6.9. Worth knowing, not yet worth acting on.

Anything below 5.5 does not enter the edition. It may still appear on the wire, which is unranked by design.

What the number is not

It is not a prediction, a confidence interval, or a claim to precision. The Evidence dimension in particular means a widely reported story resting on a single vendor benchmark will score below a quieter story with independent verification. That is the intended behavior, and the most common source of reader disagreement.

Scores are set before publication by the editor responsible for the story and reviewed by one other editor. They are not revised after publication except to correct an error, and any revision is noted on the article.

Disagreeing with a score

Tell us which dimension you think is wrong and why. That is a more useful argument than the total, and it is the one we can actually act on.Get in touch.