Evidence
Not Every Result is Evidence of Success. Knowing the Difference Matters.
The HEOS Evidence Engine produced results in more than one category, and I don't want that distinction blurred just because one of the numbers is dramatic. Here's what each category means — and where the $251,000 finding actually sits.
Three Kinds of Results
- Observed Finding
- A real pattern the analysis surfaced — not necessarily a planted failure.
- Benchmark Result
- The formal outcome of testing whether a planted, hidden problem was recovered.
- Control Result
- The formal outcome of the Healthy Baseline negative-control test — whether the system stayed quiet when no planted failure was present.
An Observed Finding does not equal a Benchmark success. A Benchmark miss doesn't mean there were no useful findings. A clean Control Result doesn't prove general reliability.
Observed Finding
The $251,000 Observed Finding
One result mattered immediately: an estimated $251,000 additional collections opportunity if out-of-network claims had realized at the in-network rate.
That finding was real within the synthetic data. It was not a Benchmark Result, and it wasn't one of the two problems I'd deliberately hidden for the system to find. When I later tested the Evidence Engine against those specific hidden problems, both came back MISS. The Healthy Baseline came back CLEAN. The $251K finding sits in a separate category from all three.
The goal isn't to make the $251,000 number impressive. It's to make it inspectable.
Sourced from Harbor Ridge Scenario 1's frozen blind-test output and its evidence-trail reconstruction. Full arithmetic and underlying claim-level records are at Evidence Trail.