Detect unsupported, contradicted, and over-generalized statements.
PrerequisitesW01 object and scope card.One questionable GEO claim from a paper, blog, report, or platform page.
Builds on
W01 object and scope card. · One questionable GEO claim from a paper, blog, report, or platform page.
Investigates
Which source can support which claim?
Feeds
L04 seed ledger and adjudication questions.
02
Before class
Arrive with a prepared artifact
Reading route
Core PAPER-20/PAPER-40; inspect the controlled 86-source status and ceiling legend.
Viewing route
Open the complete text-first lecture packageFive-minute worked example: from a vendor headline to four atomic claims. The current equivalent is notes, slide script, worked case, and no-video transcript; no recording is claimed.
Readiness check
Underline every quantifier, causal verb, platform name, date, and population in the brought claim.
Bring
Ten-row seed claim ledger with one unresolved contradiction.
03
Explanation
Concepts, assumptions, and boundary
Claims are the unit of audit
A polished paragraph can hide several independently testable claims: that a platform retrieves a source, that a feature caused the retrieval, that the effect generalizes, and that the result produces a business outcome. Claim atomization converts prose into units with a subject, predicate, scope, population, time window, and uncertainty. Each unit receives a disposition—supported, partially supported, contradicted, unresolved, or out of scope—and a source edge. This preserves useful nuance while preventing one strong citation from laundering adjacent unsupported statements.
Authority is claim-relative
Source quality is not a single prestige score. A first-party platform document is authoritative about the named product behavior it documents, but cannot establish cross-engine effects. A national standard can establish a requirement within its scope, but cannot prove ranking improvement. A peer-reviewed experiment supports conclusions bounded by its design, data, models, and date; it does not automatically describe a changing commercial surface. Practitioner tutorials and vendor reports can supply terminology, implementation hypotheses, and observational cases, but require primary or official corroboration for scientific, legal, or causal claims.
04
Primary visual
Inspect the mechanism or evidence structure
Conceptual graphClaim–Evidence–Source graph
Figure design. Three vertically separated node bands for claims, evidence passages, and source identities; solid support, dashed partial support, red contradiction, and grey unresolved edges. Authority and disposition belong to the edge, not to prestige alone.Long description
Atomic claim nodes connect to exact evidence passages, which connect to versioned source identities. Edge shape and label carry disposition so color is never the only signal. A side legend states the L/N/A/B/C/D/E evidence ceilings.
05
Interactive check
One action, one feedback state
Action
Map six claims to nine source cards and choose an evidence ceiling.
Feedback
Feedback explains why source authority changes with the claim and why one source may support only part of a sentence.
Accessible alternative
The mapping is available as a printable matrix with the same explanations.
Describe one source that is credible yet still unsuitable for the claim attached to it.
Revision
Split every row containing more than one falsifiable predicate.
Low-compute route
All source cards and excerpts are supplied as static, licensed or citation-only fixtures.
Five retrieval questions
01Why is authority claim-relative?
A source may be authoritative about one scope—such as its own product or a legal requirement—while being unable to support a causal or cross-platform claim.
02What is the minimum resolvable evidence edge?
A versioned source identity plus an exact page, section, paragraph, table, or record location supporting the atomic claim.
03How should a contradiction be handled?
Preserve both sources, compare scope and method, mark the claim unresolved or narrowed, and record adjudication rather than silently selecting the convenient result.
04What is the evidence boundary for W02?
The papers motivate evidence integrity and manipulation risk; their taxonomies must still be checked against the course’s operational definitions.
05What must you submit or revise after this week?
L04 seed ledger and adjudication questions. Split every row containing more than one falsifiable predicate.