The concept is technically credible enough to justify a controlled feasibility experiment, but not enough to justify full development. The next dollar should buy evidence that separates signal recovery from model overfitting and measurement artifact.
SANITIZED SAMPLE · TECHNICAL FEASIBILITY
Can optical state be recovered in a highly scattering process?
Illustrative technical question. The sample intentionally omits proprietary methods and project-specific experimental details.
The underlying physics is understood.
Absorption, scattering, path-length effects, and re-emission can all distort optical measurements. Existing approaches provide pieces of the solution but do not automatically resolve state under every dense-process condition.
Inference under changing optical conditions.
The useful question is whether a constrained model can recover a decision-relevant state variable as density and optical properties vary—not whether a model can fit one calibration set.
Predefine quantitative success.
Specify error, repeatability, holdout performance, and perturbation thresholds before testing. A feasibility result must survive conditions it was not trained to memorize.
Correlation masquerading as measurement.
The highest-risk outcome is a model that appears accurate because experimental conditions are confounded. Controls must deliberately break those correlations.
- Define the state variable and quantitative acceptance threshold.
- Design independent perturbations of density and optical properties.
- Reserve holdout conditions before model fitting.
- Kill or redesign the concept if it cannot generalize across those controls.