Limitations
What should not be inferred from this data.
Counts are not people
Listings are not worker counts. Reposting, agencies, rotating advertiser identities, moderation, and platform behaviour can all affect how many listings appear in a month.
Prices are advertised, not observed
Advertised rates are not transaction prices. They may reflect positioning, bundling, negotiation room, convenience pricing, or strategic self-presentation.
Identity continuity is proxy-based
Repeated phone numbers and user identifiers are advertiser-level continuity signals, not person-level truth. Shared numbers, changed numbers, and missing identifiers all affect overlap metrics.
Service tags are self-described
Structured service fields and title language are marketing signals. They are useful for tracking positioning, not for asserting standardized service delivery.
Interpretation guardrails
Regional movement may reflect reposting, agencies, temporary rotation, or changing listing strategy.
Outcall premiums should not be described as direct measures of danger or harm.
The intimacy-service index measures advertised positioning, not emotional reality or personal relationships.
No output from The Daylight Project should be used for individual targeting, enforcement-style surveillance, or neighborhood-level inference.