Start with a signal we can explain
The October 2026 popularity index uses Discord membership counts from 18 October and Reddit subscriber counts from 19 October. Its fixed cohort contains 62 catalogue system entries with direct mappings to both platforms in classification v3. All 44 selected Discord communities and 59 selected subreddits have observations. This is complete collection for that cohort, not complete coverage of the RPG market.
For each system, we add memberships from its directly associated communities. For each game family, we add the totals of its tracked systems. The main ranking combines Discord and Reddit; separate platform rankings are also available. Shared communities contribute once per associated system, including repeated contributions within a family. For example, a 100-member server mapped to two systems contributes 200 memberships to their family. Audiences also overlap across platforms. These totals favour families with more tracked systems and shared communities; they are not unique-person estimates.
53 classified entries contribute to 25 families. Nine unclassified entries appear only in system downloads, and eight other families have missing values. Equal totals share competition ranks (1, 2, 2, 4); family ID orders ties. The top 10 shows ten rows and may end within a tie. Full downloads retain all families and ranks.
Discord counts are approximate memberships. Reddit subscriber counts were transcribed from metadata-page screenshots, which contain private browser context and are not distributed. The transcription has not been independently rechecked against those images. The frozen method and calculation files document the evidence and reproducible arithmetic.
These are indicators of online interest, not measures of sales, active players, or market share. They cannot establish local demand for a shop or an event.
Which games belong in the project?
Our provisional starting criteria are:
- An English release of a core book is available online, either for purchase or as a free download.
- The game is a tabletop RPG system as commonly understood.
The final list and criteria have not been agreed. We still need clear rules for editions, multi-book systems, and games that depend on another rulebook.
The research catalogue is a list of books and editions, rather than a ready-made list of distinct systems.
Systems, editions, and families
A community may discuss a single edition, a whole game line, or several related systems. Its scope needs to be recorded alongside its membership count. Only explicit system mappings contribute to system totals; a family association alone does not create a system mapping.
The index compares broader game families. The game family classification records the system-to-family mapping in independently versioned releases. Version 1 is derived from the core catalogue and documents its grouping choices. Standalone and unclassified systems have no family assignment and are not combined into a catch-all ranked family.
Classification releases have no fixed schedule. Each monthly index explicitly references the version used at the time; adding a new classification version does not change that reference or the older mapping. Changed assignments or corrections require a new classification release.
Version 2 bundles those family assignments with a public Discord server registry and many-to-many system/family associations. It includes verified public invites, stable Discord guild IDs, source evidence, scope notes, and links for later membership checks. Family associations describe the named mapped member systems; they do not imply coverage of every system in a rules family. No confirmed association is missing coverage, not a zero-member observation. The classification contains no member counts.
Version 3 adds a public subreddit registry and associations to that same bundle, preserving the family assignments and Discord mappings. It includes public community links, optional Reddit community IDs, subscriber metadata resources and scope notes. Some evidence is cached and live subscriber API access has not been verified. Weekly visitors and contributions measure activity, not subscribers; they must not substitute for subscriber totals. Independent whole-family associations are distinguished from subset links, and neither automatically assigns communities to every system.
The October release pins classification v3 and methodology additive-system-memberships-v1. Shared rules do not by themselves establish a shared player community. Future changes to coverage or aggregation must be documented before making comparisons.
Make gaps visible
A missing measurement is not zero. Releases distinguish data that was not collected, a source that was unavailable, and a measure that does not apply. Source URLs, collection timestamps, and coverage notes accompany every release.
Changes over time require comparable sources and definitions. Adding a community to the source list must not appear as sudden audience growth. Where the baseline is missing or zero, percentage growth should be left unreported with an explanation.
Monthly snapshots and corrections
The planned cadence for online sources is monthly. The homepage shows the latest published index and its observation dates. A release address such as /releases/2026-10/ identifies the observation month; publication time is recorded separately. Explicitly dated snapshots may be published after their capture window closes, while full-month releases wait until the month ends. October is a point-in-time snapshot, not a monthly average or month-end count. There is no earlier compatible observation, so changes are unavailable.
Each release provides CSV and JSON data, source contributions, coverage notes and a frozen methodology. Revision-named downloads remain available for citation; corrections require a new revision and a public correction note. A repeat collection schedule is still being established.
The annual player survey
We plan a separate, anonymous survey once a year. Its recruitment, questions, sampling limitations, and reporting rules will be described before launch. Respondents to a voluntary survey may not represent all players, so survey findings will not be presented as a census.
Only suitable aggregates are intended for public release. The survey design must address anonymity and the risk of identifying someone through small groups or free-text answers; raw individual responses are not intended to be published.
Help improve the method
Source selection, genre classification, coverage, and eligibility all benefit from community knowledge. If you spot a gap or want to help review the approach, volunteer with the project.