Additive system memberships v1

Scope and frozen inputs
The first index is a point-in-time pilot using the existing classification v3 cohort: 62 catalogue system entries with direct mappings to both Discord and Reddit. This is a source-availability cohort, not a claim of exhaustive eligibility or market coverage. Discord observations were captured on 18 October 2026, 19:52:36–19:53:44 UTC; Reddit observations on 19 October 2026, 05:13:51–05:20:10 UTC. These two frozen snapshots are used together for this release. The observation month is October, not an assertion of a monthly average or month-end measurement. No new live collection is performed in this calculation.

System arithmetic
For each platform and system, add memberships from every distinct directly mapped community in the frozen system-source table. An identical system/platform/community edge must occur only once. A community mapped to multiple systems contributes its entire count to each system. Scope-limited and publisher-wide communities retain that attribution; their audience cannot be interpreted as players of each game. Family-only mappings do not create new system mappings. The main ranking uses the arithmetic combined total: the Discord system total plus the Reddit system total. Discord and Reddit are also ranked separately as breakdowns. Cross-platform overlap is intentionally retained; the combined count is not unique people.

Family arithmetic
For each family, add totals of its tracked systems, using only the pinned v3 family assignment. Shared sources are intentionally repeated through each system: if two systems in a family share a 100-member server, that server contributes 200 to the family total. A source can contribute to several families. This implements the owner's additive rule without a within-family deduplication adjustment. It increases totals for families with more tracked systems and repeated shared communities. It is a membership signal, never a unique-person or unique-player estimate.

Coverage and missingness
53 classified cohort entries contribute to 25 families. Nine unclassified entries remain in system outputs and are excluded from family totals without inventing assignments. All 44 Discord sources and 59 Reddit sources for the cohort are observed. Eight classification families outside the cohort have null members and rank with status not_collected. Unmeasured systems within an observed family are outside the cohort; its value is explicitly a tracked-system subtotal, not a complete family total. The calculator refuses partial source totals if any selected source is not observed. Observed zero remains zero.

Ranks and comparisons
Rank each supplied metric descending by total. Equal totals share competition ranks (1, 2, 2, 4); stable family ID determines display order within a tie. Missing families follow observed families, ordered by ID. The top-ten display contains ten rows and may truncate a tie at the boundary; downloads include all rows and their shared ranks. Every change is null for the first release. Future growth requires the same cohort, edges, method and comparable capture windows, or an explicit break in the series.

Evidence and reproducibility
Discord counts are approximate total guild memberships, not online users. Reddit counts are subscriber integers visually transcribed from Chrome metadata-page screenshots, not weekly visitors. Original screenshots contain account/browser context and are stored locally outside Git. Structured observations carry screenshot hashes; these do not independently establish transcription accuracy. No new pixel-level review is claimed by generation. The published transcription has not been independently rechecked against the original screenshots; corrections should cite the affected source and retain the prior revision.

calculation.json pins the input files, calculation script and generated output hashes. contributions.csv lists one observed count for every system/source edge. sources.json lists each contributing family/platform/community, its system IDs, multiplicity, original count, weighted contribution and timestamp. systems.csv lists all cohort entries and platform/combined subtotals; metrics CSV and JSON contain ranked family results. Public outputs contain no original screenshots or account context. Classification and observation archives remain unchanged.

Publication and review
This is published snapshot r1, released on 28 October 2026. Explicitly dated snapshots may be published once their observation window closes; ordinary full-month releases remain publishable only after month-end. The October 18–19 dates describe the measurements, not a monthly average or month-end count. Future releases require a documented capture schedule and repeat collection process. Corrected published releases must retain earlier revision files and disclose changes.