Establish superiority or equivalence in the discovery pilot

Determine whether the native compaction approach is superior or equivalent to the application approach for the reported discovery task, including answer time, required reads, and required-fact retention, using sufficient raw repeated-run evidence to support the comparison.

Background

The discovery pilot contains one weather task with only two executions per arm and provides summarized timing values without raw timings. Both arms pass the reported required-read and required-fact checks, but the available documentary sample cannot support a reliable comparison of performance or tail behavior.

The paper explicitly leaves superiority, equivalence, and tail-latency conclusions unresolved. A definitive study would need raw repeated-run measurements, a prespecified comparison design, and a clearly defined target population.