Frontier experiment starter kit
Use with: experiment atlas · detector atlas · open topics
An experiment is ready to leave the atlas when every field below is concrete enough that a skeptical team could price, simulate and preregister it.
One-page experiment card
| Field | Required content |
|---|---|
| Target and topic IDs | One canonical frontier target and one maintained open topic. |
| Conventional question | A useful physics question independent of the programme-specific interpretation. |
| Programme-specific bridge | The exact carrier/map/invariant the experiment would test. |
| Platform | Material, device, geometry, dimensions, temperature, field and fabrication state. |
| Preparation | How the state, sector, orientation or history is physically created and independently verified. |
| Control parameters | Variables swept, held fixed, randomized and blinded. |
| Primary observable | A dimensioned detector output and its calibration route. |
| Added discriminating channel | Reversal, frequency, geometry, waiting time, higher moment or cross-detector signal that breaks the leading equivalence. |
| Null and alternatives | At least one conventional null and the strongest mundane competing mechanisms. |
| Nuisance model | Drift, heating, disorder, contacts, backgrounds, finite size, hysteresis and data-reduction choices. |
| Predicted contrast | Sign, scaling, band, pole, lag, symmetry or likelihood contrast; sensitivity alone is not a prediction. |
| Power / precision | Expected effect, uncertainty budget, sample count and stopping rule. |
| Kill test | A result that rules out or demotes the bridge under its stated assumptions. |
| Ordinary residue | What useful measurement or bound remains after a null. |
| Interpretation ladder | Conventional result → shared mechanism → typed bridge → new physics, with separate thresholds. |
Required pre-computation
- Freeze the operator and port before fitting.
- Propagate the same material/device parameters into every claimed channel without hidden refits.
- Simulate the leading nuisance collision through the same detector fold.
- Run ablations that remove the proposed second datum, orientation or memory channel.
- Declare order of limits, bandwidth, finite-size scaling and calibration transfer.
- Produce synthetic blinded fixtures and verify the analysis recovers the injected contrast.
Decision rule
A result reaches new-physics candidate status only if the preregistered contrast survives calibration, nuisance alternatives, detector equivalence and an independent replication route. Otherwise report the strongest lower rung actually supported. A null can still close a mechanism, set a bound or validate the apparatus model.
Handoff artifacts
- completed one-page card;
- device or sample specification;
- frozen analysis script and synthetic fixtures;
- uncertainty and nuisance ledger;
- preregistration text;
- data and calibration provenance plan; and
- a corpus update naming which frontier edge the result changes.