I already said I don't consider alternative explanations on their own to be indicative of lying. I don't know where you're getting this notion that speculation is evasion, here's what I said on the matter:

If a client is either factually innocent or guilty-but-sober-minded, there’s no difficulty getting them to admit the incriminating nature of incriminating evidence. If a client is lying — whether to me, themselves, or just desperately trying to manifest a reality which doesn’t exist — it’s like pulling teeth.

If they have no idea what's going on then there's no need for this exercise. There's other ways to cooperate in truth-seeking.

I don't consider alternative explanations on their own to be indicative of lying, especially if the alternative theory as a whole more accurately comports with reality. This is why there are two parts to this exercise: surviving the gauntlet of facts and dethroning the other survivor (if any).

Why would lying be a natural response for a non-liar falsely accused of lying?

I've never encountered this framework before but I'm curious. What do you find useful about it?

It's certainly possible to just constantly amend a theory and keep it technically cohesive, but I've found that even dedicated liars eventually throw in the gauntlet after their contortions become too much to bear. Even if a liar refuses to give up, they still have to grapple with trying to unseat the truthful (and much less convoluted) theory. That's why there's two parts to this exercise: surviving the gauntlet and dethroning the other survivor.

I didn't take a position on what balance between convenience vs ambiguity we should strike, it's always context-dependent. That said, second-person pronouns like "you" tend to be significantly less ambiguous than third-person pronouns. Because you generally know who is talking to you directly (see what I did there?) whereas "they" can potentially refer to anyone in the world.

I mentioned this in another comment, I used an unrealistically convenient example for illustrative purposes. A real-life application of my rubric on a real-life lie would be much more complicated and take multiple detours.

I'm unclear on what the distinction is exactly. This is a tutorial that works for catching a talented liar but also creating common knowledge between yourself and a bad liar.

The example was intended to be unrealistically convenient, since the goal there was just an illustrative example. Had I used an actual lie narrative from one of my clients (for example) it would've gotten very convoluted and wordy, and more likely to confuse the reader.

I acknowledge there are limitations when you're dealing with unknowable lies. Beyond that, it was really hard to figure out how rare "lies with convenient flaws" really are. I don't know what denominator I'd use (how many lies are in the universe? which ones count?) or how I'd calculate the numerator.