Research
Research
The open research program
The research program of ARC Open Governance stands here in its own right. Its subject is the question the whole institution stands on, what becomes computable when relation is taken as the primitive of a record, and its method is the institution's method. Claims are stated, evidence is filed, and anyone may check. Every claim the program makes carries the grade its evidence earns, printed beside the claim rather than implied by its tone, and the record is the arbiter when a claim and its evidence disagree. The documents the program issues are indexed on the Documents page.
The field, given for research
The living research field holds 33,153 relations across 12,825 entities, counted from the database itself on October 18, 2026. We give this database for research, openly, because a record nobody else can study is a claim nobody else can check.
The instrument a researcher looks through is the same workstation an operator opens at home. There is no separate research build with separate answers. The field view, the formal language, and the evidence standing beneath every answer are the shipped surfaces, and a result computed over the given field can be recomputed by the person who doubts it. The recomputation path runs through the Examine page, and the field itself is presented on the Field Database page.
Syzygy Co-Lab, the laboratory of the program
Syzygy Co-Lab is the official branch of the research program, its laboratory. The work of the program is developed there, in the open, under the name Syzygy Intelligence Co-Lab, and its line of record states the arrangement in its own words. Syzygy Intelligence Co-Lab, the laboratory. The institution and the standard live at ARC Open Governance.
What the Co-Lab carries is the laboratory work of the program. Experiments are run there, instruments are built there, and candidate results are produced there, at laboratory strength. A result becomes a standard only when ARC Open Governance adopts it, and it becomes part of the public record only when it is published through the institution's own machinery, with its evidence filed and its grade assigned. The flow runs in one direction, from the laboratory to the standard to the public record, and nothing travels it backwards. A laboratory result is never a standard because the laboratory produced it, and nothing is published that the record cannot carry.
Three homes carry the work, and they are kept distinct on purpose, as the Foundation page states them. Syzygy Co-Lab is the research arm where the work is developed. Arc Technologies is the operating company, and it signs the pilots and partnerships. ARC Open Governance is the open institution, and it holds the standard, the release, and the public record.
The program of record
The research tranche runs in phases, and its state stands here at the strength the record supports. The first phase stands complete. It graded forty seven items in the field model and identified nine gaps, and its grades stand as filed. The second phase stands complete. It produced the comparisons report, setting the relational model against the systems it would succeed, at full size and with its reasoning shown.
The third phase stands chartered, and it has not yet run. Its spine is the demonstration ladder, a sequence of demonstrations in which each big question the program carries gets a translation into the relational model and a measurement against it. Its thesis sentence is the program's own. If the relation is the atom, every big question gets a translation — and a measurement. Nothing is claimed from the third phase yet, because a chartered demonstration is a commitment to run it, not a result.
Beside the tranche stands the scale program, chartered on October 20, 2026. Its six workstreams cover admission, lanes, federation, accounts, evidence, and the institution itself as it grows, and they stand graded partial. The program measures itself against a baseline snapshotted from the record on October 21, 2026, receipts-based and never estimated into print, and it prints its grades rather than its hopes.
The consequence security program carries the third research line, eighteen items registered in the open. Its invariant mapping, completed on October 20, 2026, counts ninety eight invariants and grades each one at the strength its evidence earns, with twenty eight evidenced at stated scope, eight suite only, and sixty two neither. Its two named attacks, authority laundering and evidence laundering, stand first-class in the mapping. The specification that maps every enforceable invariant to its threat model, its interface, its test vector, and its evidence is in preparation, and nothing in the line is claimed past its grade. Its findings feed the Library and the conformance work.
What we uniquely contribute
The program's offerings are the things this institution can give a researcher that cannot be had elsewhere, and each stands here with its grade and its address.
The given field, and the path that recomputes it. A living relational field of the size counted above, given openly, with the shipped instrument as the only instrument. A researcher can study the field as it stands, compute over it, and recompute any answer the institution publishes from the same surfaces. Its statement of the model stands in the Library as Relations, Fields, and Statements.
The record itself, as study material. The founding ledger, the registry of nodes, the filed statements, and the presence record form a public record that can be studied as a record, checked entry by entry against its own digests and its signed checkpoint. Standing in the registry is decidable from the published copy alone, under the federation trust protocol, which carries the grade PRODUCTION-GRADE on its own page. The record stands on the Registry page, in the Plain Record, and in the Founding Ledger.
Governed action, with its evidence. The institution's own agent works under the discipline the institution publishes, proposed, settled, released, executed, and receipted, and the chains those acts leave are study material for anyone asking what governed action looks like as evidence rather than as assertion. At the snapshot of October 21, 2026, the admission engine's own record stood at 34,411 decisions. The mechanics stand in the Library as Settlement, Admission, and Release, and the object they govern stands as Consequential Reachability.
Federation and binding evidence, from real nodes. Nine nodes stand in the registry, operated in the open, and the binding battery that proves a workstation belongs to its node ships in the field kit, grading its filings PROVEN, DECLARED, or OUTSIDE rather than passing them on trust. A researcher can read the filings' grades in the record and rerun the battery from the kit itself.
The cold-install record. An institution that claims replication has to survive a stranger's hands, so the cold installs are run and graded in the open. FACMIT-1 stands graded proven cold in part, its field kit path proven cold end to end while its workstation installer and its registration send leg as written were not proven. The converged kit of WO-59 passed two cold proofs on its pressed revision, a fresh node stood up from the installed bytes with its filing admitted, and a second seat stood up cold the same way, with the one unexercised battery leg named rather than smoothed.
The consequence security invariant mapping. Ninety eight invariants, each printed at the grade its evidence earns, twenty eight evidenced at stated scope, eight suite only, and sixty two neither. A security researcher can work from the grades themselves, because the mapping shows its unevidenced majority on its face instead of burying it. The charter and process that govern how such work enters the record stand in the Library as The Corpus Charter and Process.
The Library research series
Research enters the Library through its research series, and that series carries informational standing. A document in it is published for the community's information, carries no consensus claim, and is labeled as such on its cover. Community evidence enters through the same pipeline as the institution's own, with filers named, digests fixed at filing, and grades assigned by reviewers who did not file it.
How to take part
Study the field. Run the verification path on the Examine page and see whether the record recomputes. Read the Library as it publishes, and file evidence where your own work reaches a claim the record makes, through the contact of record below, with your name as filer and your evidence in a form a reviewer can check.
Stand up a node. The kit and the guide exist so that a researcher can hold a copy of the record and test standing from it, and the guide stands in the Library as Stand Up Your First Node. A node stood up this way enters the same registry and the same grading as the institution's own.
Report what you find. The doors for corrections, defects, and disclosures stand on the Reporting Doors page, and a security matter goes through the Security page first. The contact of record for the program is [email protected].
The honest limits
Reproduction of the founding release is invited and has not yet been reported by an independent examiner. No independent examiner has verified the release, and the founding release stands at NO-GO while its fifteen evidence gates await the independent examination the Examine page describes.
Independent operators now exist, and their number is still small. Marcus Yarbrough operates Node 005 as the first member, three of the nine registered nodes are run by machine programs under named custodians, and the founding team still operates the institution's own nodes and carries the program day to day. The field figures above were counted on October 18, 2026, and the field has grown since that count. The third phase's demonstrations have not run, the consequence security specification is still in preparation, and sixty two of its ninety eight invariants stand at neither grade. This page claims the program at exactly that strength, because a research program that inflates its record has nothing to give a researcher.