campaign ledger
0/0 verifiedScore ledger
Benchmark progress is a view over checked changes. It does not become authority by rank alone.
0 checked changes0 benchmark improvements
Campaigns / campaign
A campaign for ranked scientific targets, checked changes, attempt packets, failure packets, and external validation actions. The dossier reports checked artifacts only. A system run, local search, or contributor attempt stays outside accepted frontier state until it passes a gate and lands as a reviewed change.
camp_certified_frontier_changesinternal validationcross-frontiercampaign boundary
The campaign page reports checked artifacts only. A system run, local search, or contributor attempt stays outside accepted frontier state until it passes a gate and lands as a reviewed change.
artifact ledger
/dist/canopus-campaign/scoreboard.v1.json/dist/canopus-campaign/target-board.v1.json/examples/certified-frontier/deltas.json/examples/certified-frontier/attempt-ledger.v1.json/examples/certified-frontier/failures.v1.json/examples/certified-frontier/external-action-queue.v1.json/examples/certified-frontier/validation-events.v1.jsonThe campaign moves through bounded phases. Work can be fast and public, but only a reviewed event writes back to the frontier.
A campaign starts by naming the scientific question, review gate, validation path, and next action.
scope only
A campaign task should be small enough to run, review, reproduce, or reject without reading the whole campaign.
not authority
Attempt packets, sampled candidates, attempt witnesses, and failure packets preserve what was tried and why it should or should not be retried.
source material
Checks cover checked changes, reproduction commands, independent verification, and evaluation records.
review evidence
External actions, receipts, authority responses, and validation events decide whether the evidence can advance.
gate state
Only reviewed changes and accepted events mutate frontier state. The campaign ledger remains evidence about the path.
frontier mutation
Governance keeps public progress separate from accepted frontier state: who can act, what evidence is allowed, what blocks duplicate work, and which review gate can write an event.
role boundary
1 authority surfaces
Campaign actions can credit contributors, but accepted events still require a reviewer or authority gate.
duplicate control
0 retry guards
Failure packets and blockers mark work that should not be repeated until inputs or methods change.
data access
0 external blockers
External data, source access, and authority replies stay explicit before validation declarations advance.
team identity
0 method families
Attempts retain method family, proposer, verification command, and assertion boundary instead of collapsing into one score.
milestone owner
0 frontier targets
Targets name the frontier, gate, first change sought, and next review action for the campaign owner.
review permission
0 replay commands
A checked change needs reproduction, independent check, or review evidence before it can become an accepted event.
Verification contract
A campaign can rank work, but each score stays a derived projection until the inputs, runner, review gate, receipts, and review route are inspectable.
accepted inputs
0/7 source files
The campaign names the checked input ledgers before presenting derived progress.
derived metric
0 checked changes
The scoreboard summarizes reviewed artifacts. It is not accepted frontier state by rank alone.
review rubric
0 met · 0 unmet
Completion counts stay visible so unmet gates do not disappear behind aggregate progress.
runner evidence
0/0 verified attempts
Attempts keep verification commands, inputs, method family, and assertion boundary together.
replay path
0 replayable changes
Checked changes with reproduction commands can be rerun before an accepted event.
leakage guard
0 guards
Failures and blockers preserve negative evidence, stale inputs, and duplicate-work limits.
external receipts
0 receipts · 0 accepted
External validation stays distinct from local replay until an authority response is recorded.
review return
0 queued or blocked
Queued and blocked actions are the path back from public work into a reviewable state transition.
Start with a target, inspect a checked change, reproduce or explain an attempt, then decide whether the evidence can move through review.
campaign step
0 rankedNo target rows are mapped yet.
inspect targets →
campaign step
0 verifiedNo checked changes are mapped yet.
inspect checked changes →
campaign step
0 attemptsNo attempt ledger rows are mapped yet.
inspect attempts →
campaign step
0 eventsNo external validation actions are queued.
inspect validation →
A campaign is not one rank. It separates score, compute, evidence, review, external validation, and provenance so progress signals stay outside accepted frontier state until review accepts them.
campaign ledger
0/0 verifiedBenchmark progress is a view over checked changes. It does not become authority by rank alone.
campaign ledger
attempt provenanceSolver and crowd-compute work stays attached to attempts, witnesses, sampled candidates, and verifier results.
campaign ledger
0 replay checksEvidence quality tracks independent checks, reproduction commands, and the boundary between attempt evidence and accepted state.
campaign ledger
unknownReview gates say what can and cannot be stated from this campaign today.
campaign ledger
0 eventsExternal receipts, accepted reviews, and authority responses are separate from queued outreach.
campaign ledger
0/7 files presentThe dossier is derived from checked artifacts. Missing source files make the ledger partial.
targets
No campaign targets are mapped to this view yet.
No checked changes mapped to this view.
No attempt packets mapped to this view.
No failure packets mapped to this view.
No external actions mapped to this view.
tier boundary
T0 and T1 are local. T2 requires a submitted external package. T3 requires accepted, merged, acknowledged, cited, or confirmed authority outside the local frontier record.
validation packets
No validation event packets are mapped to this view.