Skip to content

Ranking Semantics Legend

Artifact — instantiates Salience-Significance Decoupling

A published key that states what a ranking's order actually means — the sort key behind it — so 'at the top' is never quietly read as 'most important.'

Ranking Semantics Legend is a documentation artifact that travels with a ranked list, leaderboard, or feed and decodes what its ordering means. It works at the level of the whole list, not the individual item: it names the sort key, maps the salience channel that produces the order, and disclaims what the ranking does not mean. Left undocumented, a ranked list carries a silent implicature — "these are ordered, so the top is the most important" — and the legend's job is to replace that with an explicit statement: ranked by X; X is a salience channel, not a significance measure.

Example

A streaming service shows a "Top 10 Today" rail, and viewers read it as the best, or the most important, shows. The Ranking Semantics Legend states the actual semantics beside it: ranked by hours streamed in the past 24 hours among logged-in accounts in your region — an ordering that structurally over-weights brand-new releases riding a marketing push and titles set to autoplay. It maps the salience channel (recency + promotion + autoplay) and carries a plain disclaimer: "Trending is not 'recommended for you,' and not 'highest rated.'" The outcome is that "Top 10" stops masquerading as a quality ranking; a viewer can now read the position for what it is.

How it works

What distinguishes it from a per-item badge is that it explains the order, as a persistent artifact bound to the list:

  • Publish the sort key in plain terms — the actual variable the ranking is ordered by.
  • Map the salience channel — name what drives an item up the order (recency, spend, engagement, autoplay), i.e. why prominence accrues where it does.
  • Disclaim the non-meanings — state explicitly what the rank does not certify (quality, personal fit, significance), and place that where the list is actually read.

Tuning parameters

  • Disclosure depth — a headline sort key versus the full weighting formula. Deeper is more honest but heavier and easier to game.
  • Placement and prominence — inline beside the list versus a buried footnote. A legend nobody sees is no legend; placement is half the mechanism.
  • Channel granularity — name only the dominant driver, or enumerate every salience input feeding the order.
  • Update coupling — static text versus a legend regenerated whenever the ranking algorithm changes, so the semantics never drift from the sort.

When it helps, and when it misleads

Its strength is pre-empting the reflexive leap from rank to importance before it happens, at the point of display. It is a standing guard against Goodhart's law:[1] once a ranking's position becomes the thing everyone optimises and cites, it drifts from whatever it once proxied, and the legend keeps the proxy's true basis visible.

Its failure mode is inertness — a legend nobody reads changes no behaviour, so disclosure without prominence is mere theatre. The classic misuse is writing the legend to reassure rather than to inform: vague copy like "curated by our editors" that drapes a pay-to-play or engagement-maximising order in the language of quality. The discipline that guards against this is to state the actual sort key in words a lay reader understands and to place it where the ranking is consumed, not in a policy page nobody opens.

How it implements the components

Ranking Semantics Legend fills the label-and-disclosure side of the archetype — the components that make a ranking's basis legible:

  • dashboard_priority_key — it is the key that decodes what the priority ordering means and by what variable it is sorted.
  • salience_channel_map — names the channel (recency, promotion, engagement) that pushes items up the order.
  • salience_disclaimer_or_label — carries the "top ≠ most important" disclaimer on the list itself.

It does not recalibrate or re-rank the list to fix it (reweighting_and_revision_rule — that's Dashboard Salience Calibration), define significance criteria (significance_criteria_register — that's Salience-Significance Matrix), or gate any downstream action (decision_weight_gate — that's Notification Priority Review). It explains the order; it does not change it.

Notes

Keep it distinct from Display Reason Label, its closest neighbour: the label explains why this one item is in front of you ("Sponsored," "Because you watched X"), while the legend explains what the whole ordering means. And it is a documentation artifact, not a fix — where the ranking itself needs correcting rather than explaining, that is Dashboard Salience Calibration's job.

References

[1] Goodhart's law — "when a measure becomes a target, it ceases to be a good measure." A ranking position is exactly such a measure; publishing its semantics keeps the basis of the order visible even as it comes under optimisation pressure. ↩