Kathryn McKinley

Kathryn McKinley

Kathryn McKinley

A new programming model under development at Microsoft introduces "uncertain types" to address critical inaccuracies in sensor data caused by occlusion and signal interference. By embedding statistical semantics directly into the runtime and utilizing hypothesis testing, the system replaces strict boolean logic with probabilistic assertions that can snap noisy GPS data to road networks with minimal code. Early testing demonstrates that this approach significantly reduces false positives compared to naive implementations while enabling developers to specify acceptable confidence thresholds for applications ranging from agriculture to news verification.