Hence the following words in my post, "I remain highly skeptical."
I'm not saying the problem does not exist, period. I'm saying that because 1) I have not seen evidence for it and 2) its existence is not predicted by theory and would contradict the defined behavior of Windows memory management, I am highly skeptical of these claims.
If you told me, "running an EXE file from a spam e-mail will probably hose your computer", I will believe you even though I've personally never tried it, because it fits well-known expectations of what what spammers do. But this is different because, as I said earlier, applications have absolutely no access to the page file and is blind to what can get paged out (well, okay, they can mark a page they own as please-do-not-page-this-out), so this would be like someone telling me, "running a signed EXE file from a trusted source will hose your computer"--I haven't experienced it, and I would be right to doubt such a claim and to ask the questions, "Are you really sure it was signed?" and "Are you really sure it was a trusted source and not something that just looked like a trusted source?"
And as I said earlier in the thread, if someone can post a screenshot of a pagefile-related error in which there isn't genuinely high memory demand, I'd love to see it--show me the evidence, and I'll sing a different tune. I want to see the exact error in question and also see evidence suggesting that it's the result of the page file not existing and not the result of some other cause.