Journal — Method
What makes a research prototype usable?
The release standard: a person outside the Lab should be able to install the tool, inspect a sample, and repeat a documented analysis.
A research prototype becomes useful to someone else when its operation no longer depends on the person who built it. The Lab's August 2026 readiness plan sets that standard for the sensing platform: an outside user should be able to install it, inspect a sample, repeat a documented analysis, and report a problem.
That changes what counts as a release. A working screen needs a documented data path. A dataset needs its units, timestamps, device history, and permitted uses. A result needs a procedure that another person can run.
What travels with an observation
The record design includes device metadata, calibration status, timing information, quality checks, and reasons a value may be missing. These fields give a later user the context needed to decide whether two observations can be compared.
Declaring a field in a schema is the starting point. The release also has to show whether it is populated, how it is checked, and what happens when the information is unavailable. The storm dataset is a useful test case because its uncalibrated devices and interruptions make those questions concrete.
What travels with the tool
The planned package includes a versioned schema, an ingestion path, quality checks, sample data with clear permissions, a reference analysis, tests, and contributor documentation. Location handling belongs in that documentation: how precision is reduced, where the change occurs, and what the exported record retains.
A detector also needs an evaluation appropriate to its claim. Recording a change during an event is different from predicting that event. A forecasting claim would require an evaluation against declared alternatives on events not used to develop the method, including missed events and false alarms.
The immediate release milestone is more practical: an outside person completes the documented workflow and can explain any point where it fails. That run would provide evidence about whether the platform is usable, as well as a clear list of what to improve before recruiting more field operators.