THE STARK-UP COLLABORATION PLAYBOOK

Turn a shared problem into useful progress.

Start with one use case, an observable outcome and the people affected. Advance through evidence and voluntary decisions.

Problem owners and testers

Search for the issue before opening another Box.
Add your experience, frequency and current workaround without exposing private information.
Define what a successful solution would change.
Try safe, consented candidates against the same criteria; report outcomes and limitations.

Choose a sourced use case and speak to affected people.
Describe a distinct mechanism, resources, risks and a falsifiable first experiment.
Credit contributors, prior work and any tools used; select code licences deliberately.
Record repeatable results, choose collaborators and agree a charter before forming a team.
Assign delivery tasks and prepare milestone evidence for partners.

Define a measurable problem and who benefits.
Separate the prize budget from service costs; confirm authority and ability to meet the pledge.
Publish deadlines, eligibility and fixed assessment rules.
Evaluate evidence consistently, disclose reasoning and record direct payment separately from recipient receipt.

Inspect problem credibility, comparative results and assumptions.
Check ownership, licences, costs, dependencies and milestone acceptance criteria.
Offer specific mentoring, collaboration or non-binding funding interest.
Agree any commercial, investment or equity terms separately. Access requires participant consent.

People may build with or without generative tools. Evaluate the mechanism and evidence rather than the tool label or the volume of generated text.

Problem evidence: distinguish observed needs, source-backed context and assumptions.
Baselines: compare with the current workaround and strongest practical alternative.
Repeatability: record data permissions, instructions, relevant versions, resources and costs.
Robustness: test failure cases, changing inputs, accessibility and human review requirements.
Original contribution: explain what changed beyond wording; credit prior implementations and collaborators.
Tool use: accurately disclose tools used, verify outputs and avoid uploading confidential or personal data without permission.
Decision discipline: record an iterate, pivot or stop decision when evidence warrants it. A polished pitch is not proof.

Automated checks and optional draft assistance support human review. They do not award ownership, certify originality or make investment decisions.

Term	Meaning on Pandora’s Box
Box	A shared problem page with related proposals and discussion.
Research brief	A source-backed starting point. Its proposed experiment is a suggestion that needs testing.
Founder project	A contributor’s proposed solution and project plan. It is not necessarily an incorporated company.
Trial / validation	A structured comparison of solutions under published rules. It is not regulatory approval or a guarantee.
Evidence record	A project’s published observations, assumptions, costs and plans. Assess its sources and limitations.
Independent review	A review by an eligible person outside the solution team. It is not automatically a professional audit or certification.
Prize pledge	A sponsor’s stated payment commitment under the challenge rules. It is separate from payment and does not mean Pandora holds the money.
Payment receipt reported	The named recipient says the payment arrived. The platform has not independently checked bank settlement.
Support offer	Non-binding interest in mentoring, collaboration or funding. A contract or payment requires separate agreement.
Team milestones recorded	A team has completed the platform’s milestone process. Company registration, ownership and investment approval are separate.
Timestamp / fingerprint	A record of submitted content and its timing. It does not grant a patent, exclusivity or legal ownership.
Test tokens / credits	Development records with no cash value. They are not live tradable assets or startup capital.

Check each challenge’s specific rules. Company prizes, comparative tests and external competitions can have different requirements.