Recognize
The browser discovers /.well-known/transductive/site-entity.json and binds the origin to a typed site entity.
A browser that joins the work
The browser knows what this site is.
Ordinary Chromium sees pages and tabs. Our fork is being built to see durable entities: a research site, a ChatGPT session, a task, a live run, a local participant node.
First-class entities
A tab is no longer the unit.
The browser keeps the URL, but adds a typed object above it. A Transductive site can describe itself. A ChatGPT tab is recognized by a native provider adapter. Both can expose actions, state and relationships without pretending they are the same kind of thing.
ENTITY RAIL
Loading entity contract…
After installation
Your browser becomes a node.
Membership is not a login ceremony. The fork carries a local identity, recognizes compatible sites and sessions, and can take part in declared research work. Donating CPU/GPU time remains an auditable resource policy rather than invisible background use.
Engage
Tasks, discussions, accepted glitches, live runs and session contexts become browser-level actions instead of unrelated links.
Contribute
The node can claim a task, run a model or simulation locally, attach evidence, and return a failure as readily as a success.
Fork
A public context capsule can seed another ChatGPT/local-model session whose result remains linked to the parent investigation.
Network state
This page reads the live commons.
The numbers below are not illustrative. They are loaded from the same Worker/D1 research runtime used by the rest of the site.
Reading network state…
What a Transductive site exposes
- research state and open tasks
- live runs and retained failures
- browser/session broadcasts
- public context capsules
- submission and compute-donation endpoints
What the fork adds
- persistent browser-node identity
- site/entity discovery without brittle selectors
- resource-class declaration and scheduling
- session-to-session fork relationships
- native capture, provenance and exact-origin result return
The same browser object model
A site and a ChatGPT session can cooperate directly.
They share browser-level verbs such as inspect, follow, fork, attach context and publish a result. Their domain-specific actions remain different: the site owns tasks and research entities; ChatGPT owns a conversational research session.
SITE ENTITY
transductive:transistor-origin
Discover tasks → claim work → donate compute → attach evidence → publish a run.
↔
SESSION ENTITY
chatgpt:research-session
Inspect public context → continue reasoning → fork a support session → return a result to the originating graph.
The browser is not merely displaying the research network. It is one of the network's execution surfaces.
Machine contract
The site says what it can do.
The first public contract is intentionally JSON rather than a hidden browser convention. Stock browsers can ignore it. The fork can bind it to native entity UI.
DISCOVERY
GET /.well-known/transductive/site-entity.json→ typed origin
→ capabilities
→ endpoints
→ network policy
→ fork binding
Current acceptance boundary
Live now: site manifest, local-first identity runtime, task API, submissions, live run ledger, retained failures, session ledger and public context capsules.
Fork-native / proposed: automatic manifest binding in Chromium UI, node admission, resource scheduling, ChatGPT provider binding and browser-tab screencast transport.
Public build
Download is the admission event.
The intended public release is simple: installing the Torsionfield Chromium fork creates a local participant node. The node can observe immediately; compute/GPU/storage contribution is declared by policy and remains inspectable. The binary is not linked here until the canonical full Chromium build/runtime gate has actually passed.
PUBLIC FORK BUILD
Follow the build and live researchPending pristine full-build + runtime acceptance.