Skip to main content
SiftingSignal logo SiftingSignal.com
THE SIFT · weekly cartography of discourse
→ All viz components · live.html
The Sift·Topic Landscape·Media
Week of May 24, 2026·32 topics·1,247 signals
The Sift · Media · week of May 24

Every topic, plotted in discourse-space.

This week's contested questions, mapped against the evidence behind them. On the X axis: how contested a topic is — from settled (consensus across all five tiers of source) to highly contested (top experts and popular voices in open conflict). On the Y axis: evidence ratio — how much of the discourse is substantive evidence vs hot-take rhetoric. Top-right is where reasonable people are still arguing well; bottom-right is where they aren't.

This week's map clusters around two contested fault lines: the contributor-generated-content debate (top right — high disagreement, high evidence) and the critic-vs-audience divergence axis (mid-band, contested but increasingly evidenced). The settled corner is dominated by craft-of-reading topics where top experts and popular voices actually agree — even when neither thinks the other does.

↖ Settled · Substantive
Quiet consensus, citable
↗ Contested · Substantive
Where the synthesis earns its keep
↙ Settled · Noisy
Settled, but the takes still come
↘ Contested · Noisy
Highest disagreement, lowest evidence

01 How we measure

How contested (0–100)

Combines within-tier variance (do top experts agree with each other?), between-tier max gap (do top experts and popular voices say the same thing?), and an expert-vs-popular bonus. Derived live from the tier breakdown of every signal we ingest.

Evidence ratio (0–1)

Tier-weighted share of substantive signals (cites data, studies, firsthand experience, falsifiable claims) over rhetorical ones. A top-expert substantive citation counts more than a forum one.


02 All 32 topics — sortable

All 32 topics — sortable

Topic Niche How contested Evidence Signals Trending (12h)
Synthesis by SiftingSignal Independently verified 30-day window
Legal/non-task pages should set: instead. -->