Vela manifesto · 3 min read

Science needs a record that can change direction.

Models are increasing the volume of possible scientific work. The useful unit is not another answer. It is a reviewed change to shared state that others can inspect, reproduce, correct, and continue.

Open the record
  1. 01

    Candidate work is becoming cheap.

    Models can propose proofs, constructions, protocols, and explanations faster than researchers can assess them. The bottleneck is moving from generation to selection, review, and maintenance.

  2. 02

    A result needs an address.

    Claims, evidence, verifiers, failures, corrections, and open obligations should remain connected through a durable history. The next worker should inherit state, not reconstruct it from scattered prose and private threads.

  3. 03

    Verification is not authority.

    A checker can establish that an artifact satisfies a named test. It cannot decide significance, intended meaning, or institutional acceptance. Those decisions must remain explicit, attributable, and contestable.

  4. 04

    Corrections must change what happens next.

    Scientific state is useful when a reviewed correction reaches the next task, model, experiment, or decision before an obsolete assumption does. A record that cannot redirect work is only an archive.

  5. 05

    Build the substrate, then prove the institution.

    Vela provides an open, replayable protocol for this record. Exact mathematical frontiers show that the mechanics can work. Independent recurrence, empirical writeback, federation, and durable stewardship still have to be earned.