Wow - that's excellent!
Nope, I can't take the trophy.
But some comments anyway - 3db is not twice as loud - it's "slightly louder" according to Axiom's own articles. 10db is twice as loud. Your point is still valid though - it's much greater than even an average HTIB.
6 jitters - each less than 23microseconds long. If there were simply six wrong notes that long, you wouldn't hear those either.
To put it in perspective (if ya don't wanna bother doing the math) there's room for 264,000 errors in that six second clip. Roughly.
This isn't to say you're aren't 100% correct in your test idea - it's a GREAT idea! I just think the sample isn't suitable for this type of test.
Try the intro of Beethoven's Pathetique. Or nearly anything off Johnny Cash's first American Recordings CD (can't recall the name of it...starts with Delia's Gone).
A blatant error by the guitar player would barely be noticable in your tracks - that music is busy!
One sample jitter is 23 microseconds of change. The noise floor, while HUGE by the standards of silence, is tiny compared to the busy tune. These things are heard during subtle quiet accurate clean nearly-naked pieces of music, especially during decays and rests.
I think your effort is fabulous - and although I didn't hear a difference, we all know there IS one (or two) now!
I'm just not convinced (yet) that I'd NEVER hear it, given the right piece of music, whatever that is.
I not really sure one "jitter" (the 23uS) is even audible - I think our limit of discrimination is more like 30uS. But I think my kmart cheapie on a bad day strings all six screw ups together, and THIS, you might hear.
