Welcome to PARS Discussions!

[rookepoole]

👋 Welcome to PARS Discussions!

Welcome to the community around PARS — the Poole Adaptive Reasoning Stack.

PARS is an experimental reasoning and verification architecture built around a simple idea: difficult work should not just produce a plausible answer — it should preserve the real task, protect hard constraints, challenge weak assumptions, respond to new evidence, and verify the final result before making a claim.

We’re using GitHub Discussions as a place for researchers, developers, AI builders, experimenters, skeptics, and curious newcomers to explore that idea together.

💬 What belongs here?

We encourage you to:

• Ask questions about PARS, its architecture, protocols, terminology, or implementation.
• Share experiments you’ve run using PARS or $apply-pars-deep.
• Propose ideas for improving the reasoning architecture, verification methods, tooling, or documentation.
• Discuss benchmarks and ways to test whether PARS actually improves reasoning or artifact quality.
• Share failures and unexpected results. Negative evidence is useful evidence.
• Explore exact-binary research, provenance verification, reconstruction, and related experiments.
• Show what you built using PARS and explain how you evaluated it.
• Challenge assumptions. Constructive attempts to falsify PARS claims are welcome.
• Help other community members reproduce experiments, understand results, and improve their work.

🧪 Evidence over hype

PARS is an experimental project, and this community should reflect that.

When possible, distinguish between:

• what you observed;
• what you inferred;
• what you verified;
• what remains unknown or unverified.

Receipts, hashes, benchmarks, reproducible procedures, counterexamples, failed tests, and well-defined experimental conditions are especially valuable.

You do not have to agree with PARS to participate here. Serious criticism, alternative architectures, competing explanations, and evidence that contradicts existing assumptions are welcome.

🛠 Discussions vs. Issues

Use Discussions for:

• questions;
• research ideas;
• experiments;
• proposals;
• architecture conversations;
• benchmark results;
• showcases;
• general PARS discussion.

Use Issues when you have a specific, actionable repository problem such as:

• a reproducible bug;
• broken documentation;
• an installation problem;
• an incorrect reference;
• a clearly scoped implementation change.

🤝 Community expectations

Please:

• Be curious and constructive.
• Critique ideas and evidence, not people.
• Clearly label speculation as speculation.
• Avoid presenting unverified claims as established results.
• Give enough information for others to understand or reproduce an experiment when possible.
• Be open to changing your conclusion when the evidence changes.
• Welcome newcomers — you do not need to understand the entire PARS architecture before participating.

The goal is not to build a community where everyone agrees.

The goal is to build one where good ideas survive serious testing.

🚀 Introduce yourself

To get started, reply below and tell us a little about yourself.

You might include:

• What are you building, researching, or experimenting with?
• What brought you to PARS?
• What kinds of AI reasoning problems interest you?
• Have you tried PARS yet?
• What would you like to see explored, tested, or built next?

Whether you're here to use PARS, improve it, test it, challenge it, or simply understand it, you're welcome.

Let’s see what survives the evidence. 🧠🔬