First, as I said, the 3 miles range was taken as a hypothetical range at which I may well have concluded my preliminary collision risk assessment. It was not intended to be precise. The actual range could be 4 miles or 2 miles. Indeed, I may not know with any substantial accuracy what the distance is. Why should I need to? It's like the braking/stopping distances in the highway code. They have absolutely no practical relevance or application in everyday driving.
Second, my hypothetical illustration is that it appears that I will pass ahead. But am I sure? Snowleopard's analysis indicates that the actual change in bearing for a CPA of 400m between 4 miles and 2 miles separation is only ~3°. And even if I am sure, does the CPA (which I cannot measure accurately anyway) provide a safety margin that I am willing to accept? I know I said I could manoeuvre out of the way in nearly all cicumstances, but the risk of engine failure or hitting something in the water at the critical moment, even if remote, does exist, so I do not discount it entirely. So my real choice (it seems to me) is to eliminate the possible conflict with certainty (the sooner the better), even if that means [bending the rules a bit] [perhaps "a flexible interpretation" is better], or to wait until the range has reduced to ~1 mile, at which point the avoiding action, if needed, will have to be more substantial.
Third, why is 2 miles the range which I should set as my rule 17(a)(ii) decision point. Why not 2.5 or 1.5? You said (I think) that ship's OOW will usually have commenced avoiding action at ~2 miles, but should I rely on that?
Agreed. But we can debate the interpretation of Colregs. We can debate the scope of meaning of "appreciable" in 7(d)(i) and the intent of "if the circumstances of the case admit" in 17(c). I think you felt it otiose to debate the scope of meaning of "apparent" in 17(a)(ii), but what about "appropriate" in the same rule. We can debate the point at which a CPA crosses the line between "no risk of collision" and "risk of collision"; and whether that distance may or may not vary depending on one's perspective (small vessel or otherwise), appetite for risk or simple subjective judgement. We can also debate whether and how CPA can be reliably measured (by a small vessel or otherwise), and how that influences the interpretation of rule 17 and the previous judgements.I thought that, too. But whilst we can discuss opinions, there is nothing to be gained by discussing facts: it just becomes boring. And the fundamental fact is the text of Rule 17. We can't change that.
You may be persuaded by my arguments and acknowledge that early avoidance, albeit not in accordance with the narrow interpretation, could, in some cases, actually be 'a good thing', or just an acceptable, practical measure (I'm not holding my breath) :-). We could debate which of the available alternative avoidance measures is more or less appropriate. I suggested a 10° turn to port, which geometrically would work (comfortably I think) if made at ~3 miles range. However, I can see the argument for a more decisive turn (showing the other bow), hold for a while, then equal and opposite turn back on to original heading. Or something else (I don't like 360° turn for several reasons but it could be discussed). And would that 'appropriate action' be different if the small vessel is a yacht at 5kts rather than a power vessel at substantially greater speed?
Plenty to talk about. I respect your experience and expertise and I am genuinely interested to hear your opinions, but I will keep challenging you to justify them if I see any weaknesses or gaps. That's just the way I am.