Open thesis
Publish the problem, the present limits, the proposed leap, the risks, the budget, the duration, and the tests before hindsight can improve the story.
Independent project · No government affiliation
Registry preview · Agencies and program names are public record · The third column reports the state of this registry, not the conduct of those agencies
Founding charterVersion 1.0 / 2026-08-18
The method
must be public.
So must the kills.
OpenARPA is an independent public registry and operating method for ambitious, time-boxed research programs—including what they stop, why, and who is waiting to use the result.
STATUSFOUNDINGRegistry first
01
What the name promises
ARPA is an operating model, not an aesthetic.
A program belongs in the registry only when its authority, objective, duration, exams, and transition path can be stated and tested. The name is earned through structure—not ambition alone.
- 01
A focused technical objective stated without jargon
- 02
An empowered, term-limited program leader
- 03
A fixed duration, budget, and set of measurable exams
- 04
A portfolio designed to stop weak paths early
- 05
A transition plan beyond the program itself
02
What “open” requires
The proposed standard
Transparency must survive contact with failure.
OpenARPA begins with four commitments. The fourth is the hard one: a public record of what was stopped and why.
Open calls
Make the opportunity legible beyond familiar institutions. The question is whether a proposal can clear the program’s bar—not whether its author already belongs to the network.
Open results
Record what was attempted, what changed, what transferred, and what remains unresolved in formats that other programs can actually reuse.
Open kills
A stopped program is not an embarrassment; an undocumented decision is a lost public asset. Today OpenARPA can only reconstruct terminations from public evidence, and for most programs that evidence does not exist—which is the gap this registry is for. Every record marks what is sourced, inferred, and missing. When OpenARPA funds programs, publishing their terminations becomes unconditional.
03
The first product
Registry before institution
The public record is the first piece of infrastructure.
The registry will make programs comparable without flattening their technical differences. Every entry must expose the same minimum decision architecture.
PROGRAM RECORD / SCHEMA 0.1PUBLIC BY DEFAULT
- THESIS
- What is the program attempting, in plain language?
- AUTHORITY
- Who can fund, redirect, and stop the portfolio?
- WINDOW
- When does the program begin and end?
- EXAMS
- Which measurements trigger continue, redirect, or stop?
- STATUS
- Proposed, active, transitioning, completed, or terminated
- TRANSITION
- Who can carry a successful result into use?
- DECISION LOG
- What changed, who decided, and on what evidence?
04
Earn the next claim
Start with evidence. Add authority only after the work earns it.
- 01Now
Registry
A structured, source-linked record of ARPA-model programs and their public decision trails.
- 02Next
Method
Reusable schemas, program templates, exam design, red-team practice, and transition playbooks.
- 03Then
Practice
A learning and talent layer for people trusted to lead high-autonomy, high-accountability programs.
- 04Then
Market
Every transition record names someone who can take research into use. Enough records become a map of demand. OpenARPA can convene those buyers into commitments builders can rely on.
- 05Earned
Funding
Capital comes last, and only against demand OpenARPA can evidence. A program funded with no named customer at the end is a grant with better branding.
05
Why this is not a government agency
The missing half
Everyone copies the catechism. Almost nobody copies the customer.
The ARPA method does not end at a successful exam. DARPA begins transition planning before programs launch and works with operational and commercial partners who can carry results into use. That demand environment is part of the machine—not scenery.
Strip out the customer and the model degrades: programs become grants, exams become reporting, and transition becomes somebody else’s future problem. OpenARPA has no procurement budget and does not pretend otherwise. The buyer has to be built rather than borrowed.
DEMAND AGGREGATION
Organizations that cannot commission a capability alone may be able to commission it together. Repeated transition needs reveal a market that no one has convened.
QUALIFIED PERFORMERS
A source-linked record of teams, methods, exams, and transitions can become a list buyers trust and capable builders want to join.
ADVANCE COMMITMENTS
A credible undertaking to purchase on delivery can create demand before a product exists. Gavi’s pneumococcal advance market commitment shows the mechanism at scale.
Primary sourceThe scarce input is not only money. It is a credible customer. Researchers move when a hard problem has both support and a buyer waiting at the end.
06
Founding boundaries
No implied authority
The shortest route to credibility is saying what this is not.
Not affiliated with DARPA, ARPA-E, ARPA-H, ARIA, or any government agency.
Not a generic open-science campaign or a publication shop wearing an execution name.
Not a grantmaker, accreditation body, or certification standard at launch.
Not a promise that every program succeeds. Failure is expected; decision opacity is not.
Not a claim of lineage. The name describes a method, not a relationship to any agency.
07
Source lineage
Built from public operating precedent.
OpenARPA is independent. Its starting vocabulary is grounded in primary material published by organizations already practicing high-autonomy, milestone-driven research management.
A program may fail.