# 30-Day Community Activation Plan

Baseline date: 2026-08-16  
Primary hub: https://github.com/mmjbds/mianzhang.org/discussions  
Artifact layer: https://huggingface.co/MMJBDS  
Owned entry: https://mianzhang.org/community/

## Objective

Test whether the public research and artifacts attract meaningful external participation and qualified design-partner problems. The objective is not to maximize raw members, impressions, or repository noise.

## Thirty-Day Experimental Thresholds

- 10 distinct meaningful external participants.
- 5 public results, use cases, or artifact improvements.
- 3 returning participants.
- 3 qualified design-partner conversations.
- First useful maintainer response within 48 hours.

These are experiment thresholds, not forecasts or public claims of existing community size.

## Definitions

**Meaningful external participant:** a person outside the maintainer team who produces a substantive public question, use case, experiment, Issue, pull request, artifact discussion, or a qualified private design-partner conversation.

**Returning participant:** a meaningful participant who performs a second substantive action on a later date.

**Public result:** a versioned experiment, baseline, use case, tool, documentation repair, safe fixture, or extension with a public route and enough context to inspect.

**Qualified design-partner conversation:** a private conversation with a named problem owner, a concrete reliability or evaluation problem, a plausible decision horizon, and authority or access sufficient to run a bounded evaluation. Private identities and details never enter the public ledger.

**Useful response:** a maintainer reply that answers the question, requests the minimum missing information, routes the work correctly, or states a reasoned boundary.

## Baseline

The 2026-08-16 public API snapshot found seven active public research repositories with seven stars in total, no forks, no open Issues, no enabled Discussions, and no machine-recognized license on the main website repository. The Hugging Face Space and WisdomBench dataset each had zero likes; the dataset API reported 106 downloads. Downloads do not establish use, understanding, or participation.

No public repository event in this snapshot establishes a meaningful external participant. Private design-partner activity was not measured and must not be inferred as zero.

## Week 1: Make Participation Possible

- Publish the bilingual community gateway, contribution guide, governance, release boundary, and license map.
- Enable one GitHub Discussions hub with Questions and Answers, Research and Extensions, Ideas and Use Cases, and Show and Tell.
- Seed five discussions: welcome, KDD research, first contribution, real-world use cases, and show-and-tell.
- Update Hugging Face cards to route artifact questions to the artifact and general discussion to GitHub.
- Send 10 to 15 individual invitations to people with direct topical fit. Each invitation links one artifact and asks one bounded question; no bulk message and no request for a generic endorsement.

## Week 2: First Public Research Session

Run a 45-minute public research session around one question:

> When an agent changes users, evidence, incentives, or institutions, what should an evaluation measure next?

Agenda:

1. Five-minute problem definition.
2. Ten-minute ReflexBench protocol and result boundary.
3. Fifteen-minute participant use cases or counterexamples.
4. Ten-minute selection of one public extension or fixture.
5. Five-minute owner, version, and next public receipt.

Publish a summary only if the session produces substantive questions, a result, or a committed artifact.

## Week 3: Public Build Session

Run a 60-minute artifact session with three tracks:

- No code: improve one explanation, citation, route, or use-case record.
- Light code: add one safe fixture, parser, validator, or interoperability example.
- Full experiment: publish one baseline or extension with versions, data rights, method, result, and limitation.

No session file may include private implementation, customer data, account records, exact production parameters, or unreleased research.

## Week 4: Evaluate and Convert Learning

- Count distinct meaningful participants and returning participants.
- Review which route produced a concrete public result.
- Identify repeated private problems that cannot be solved by the public artifacts alone.
- Invite qualified problem owners into a time-bounded private diagnostic or design-partner evaluation.
- Publish a factual monthly summary without naming private contacts or inflating inactive members.

## Targeted Invitation Standard

Every invitation contains:

1. Why this person's work or operating context is relevant.
2. One public artifact, not the entire portfolio.
3. One concrete question answerable from their experience.
4. One low-friction response route.
5. No claim that participation implies endorsement, authorship, employment, or commercial commitment.

## Stop Condition

After targeted invitations and both public sessions, fewer than three meaningful external participants triggers a review of the value proposition, target audience, onboarding friction, artifact usability, and invitation quality. It does not trigger another forum, another chat group, or higher posting volume by default.

## Privacy and Commercial Boundary

The public ledger contains public URLs, aggregate counts, response times, and optional public handles. It excludes email addresses, phone numbers, customer names, private messages, account records, proposed pricing, confidential requirements, and private evaluation results.