Public deliberation plane
Challenge the system before it becomes doctrine.
Aftergraph Discussions is where research reproductions, prior art, architecture critique, RFC formation, roadmap input and implementation questions begin. Concrete work moves to the owning repository once the problem and evidence are clear.
Discussion is not authority. Canonical specifications, benchmark results, releases and governance decisions remain in their owning repository and exact source version.
Choose a route
Start where uncertainty still exists.
Use Discussions for exploration and evidence exchange. Use Issues when work is bounded. Use pull requests for concrete changes. Security vulnerabilities follow the private security route, not a public thread.
Reproduce MISSION-Bench
Report exact versions, environment, method, raw evidence, deviations and limitations. Negative results are first-class.
Prior art / falsificationShow us what already exists
Bring papers, standards, patents, source code or observable product behavior that narrows an Aftergraph claim.
Architecture / RFCAttack the boundaries
Challenge repository ownership, contract boundaries, runtime semantics and the cost of the proposed abstraction.
New proposalStart an RFC discussion
Use an [RFC] title for public contracts, compatibility, cross-repository semantics, governance or new long-lived primitives.
RoadmapEvidence before dates
Current public priorities are gate-driven. Challenge ordering with user evidence, reproducible failures or adoption signals.
General / Q&AAsk, explain, show, critique
Use GitHub's active default categories for questions, ideas, show-and-tell and organization-wide discussion.
Lifecycle
Conversation becomes work only after it earns structure.
Featured threads
Nine conversations already worth attacking.
#9Welcome to Aftergraph#10What comes after Graph Engineering?#11SPEC-001 Mission Contract: Request for Implementations#12MISSION-Bench Reproduction Registry#13Prior Art Challenge: What Has Aftergraph Reinvented?#14Aftergraph Architecture RFC: Where Are the Real Boundaries?#15Sentinel: What Must a SHIP Verdict Prove?#16Aftergraph Public Roadmap: Evidence Before Dates#17RFC Intake: How Aftergraph Proposals Become Changes
Evidence boundary
What makes a useful contribution?
Make it reproducible
Exact source version, environment, workload, method, sample size, exclusions, raw artifacts and limitations.
Make the alternative explicit
Describe what existing standards or a simpler mechanism can already solve, plus what gets worse if the proposal ships.
Count control cost
Verification, policy, state and orchestration are not free. Include latency, tokens, compute and human attention where relevant.
Keep evidence classes separate
Simulation, deterministic tests, live model runs and independent reproduction are different forms of evidence and stay labelled.