Wait, was this the analysis where you compared results across two contests in different years in completely different geographies with a commonality of about 15 percent in the participating pilot population and using the the highly nebulous "PRL to Performance Metric"? The one where you claimed it was "only 259 data points" (when one could as easily argue it was 2 data points; contest A and contest B). Where I think you used "Bayesian" in a sentence... a tactic I thought only an economist would resort to?? Yes - just poking fun at your expense.

But seriously, the statistics (lies, damned lies) you are citing strike me as only one aspect of the analysis. I think the other is either unmeasurable after-the-fact or would need a much more controlled environment to study conclusively. All I can tell you is what I've observed in ACTUAL contests using Flarm Stealth and not using Flarm Stealth and... more importantly.... what is very achievable in the realm of software development and user interface development in the near future.

I fundamentally believe this is more of a philosophical discussion than a technical one at this point. I'm afraid it may be hard to "win" this one on technical analysis alone.

Back to real work. Q1 revenue estimates are due, and unfortunately I don't have Flarm to use to leech off the other consulting firms.

P3