# MyFavoritePlay public-page information audit — prespecified protocol

Protocol recorded on 19 October 2026, before selecting or collecting game-page observations. Commissioned by MyFavoritePlay; not independent journalism.

Question: What information is publicly available on MyFavoritePlay to discover and compare games, and which decision-relevant information is absent or unconfirmed?

Scope and frame: use the bundled game catalogue in the deployed source commit a1719478b5ebfdba8f7e54c87376ce2803ed9ccb as a reproducible candidate frame, not as a claim to enumerate every database-backed live record. Deduplicate by stable game ID. Exclude casino informational records, records without a game title or ID, and platform-unclassified entries. Public pages are the unit of observation; editions or mobile records with separate IDs remain separate pages. No user or business activity metrics are collected.

Selection: stratify by kind first (Roblox, Fortnite, Mobile); remaining records use all listed platforms and mutually exclusive priority Nintendo, Xbox, PlayStation, PC. For each stratum, sort by ascending SHA-256 of the UTF-8 stable game ID and select the first three records (or all if fewer). Publish the full frame with stratum, hash and selection flag. No substitutions for sparse, unavailable or unfavourable pages. Equal sample sizes support category coverage, not population estimates. Cross-platform records are assigned once; stratum is not an exclusivity claim.

Data collection: retrieve each selected production game URL as a signed-out public page using the available text retrieval service. Record date, URL, retrieval reference, observed platform text, genre, summary, solo/multiplayer passage, price/free statement, official-destination URL, Fortnite code where applicable, sources/check-date availability and public-review state. Retain raw retrieval text. Inspect public catalogue filters separately. Do not register accounts or collect personal data. Do not test a purchase or redeem a pack.

Coding: present means visible concrete information in retrieved page text; missing means absent from that representation, not absent from the game. Generic advice such as 'check the official listing' is unconfirmed, not verified feature data. Multiplayer distinguishes explicit supported modes, generic online guidance and unconfirmed. Reviews distinguish review UI from actual published review evidence; client loading is 'not assessed', never zero reviews. Price/free status is explicit only if the page states an amount/currency or a free entry/base-game statement; optional currency/add-ons do not count. An official destination is a visible link to a publisher, store or platform for the game; target correctness and current availability are not independently certified. Fortnite codes are explicit four-four-four digit codes. Any new field must be labelled exploratory.

Analysis: report integer numerators and observed denominators only, separately for all selected pages and relevant subgroups; no market extrapolation, significance testing, behaviour, satisfaction, time-saving or competitor claims. Charts summarize coded fields and include their denominator. Unavailable pages stay in the sample and appear as unavailable. If live access fails completely, publish a not-yet-executed protocol instead of findings.

Limitations: source-defined frame can omit database-only additions; small disproportionate deterministic sample; static text retrieval cannot establish client-loaded review or price state, logged-in flows, link validity or factual correctness of game claims; no user research. Results are one-day observations and may change.

Update policy: retain the dated CSV, frame, protocol and raw observations unchanged; re-run with the same frozen frame and selection for comparisons, clearly date a new version, disclose corrections and never silently overwrite the original findings.