The question

In an article about Joe McGinniss in today's Post, Gene Weingarten dicusses the genesis of "Fatal Vision" and the way it and its author were pilloried after its publication. Among other things, Weingarten asks

What was McGinniss supposed to have done when he realized, midway through the reporting, that the man he was writing about had lied to everyone? That he had killed his wife and older daughter in a rage — and then calmly, methodically hacked to death his sleeping two-year old, stabbing her 33 times with a knife and ice pick, just to strengthen his alibi? Was McGinniss required to dutifully inform the murderer that he now believed him guilty, and invite him to withdraw his cooperation if he wished, possibly killing the book outright, but certainly killing it as a meaningful, enlightening, powerful examination of the mind of a monster?

The answer is "No" - and I would agree with that. Weingarten notes

There is an implicit covenant between a writer and a subject; in return for whatever agreement you might make for the telling of the story, the subject must tell you the truth. If he lies, all deals are off. It is impossible for a subject to be less truthful than Jeffrey MacDonald was with Joe McGinniss: he misrepresented the central fact of his story, his own guilt.

But here's something I've never understood. As Weingarten notes,

One of the main reasons that there is still doubt about Jeffrey MacDonald’s guilt – 44 years after the crime — is the degree to which “Fatal Vision” was unfairly pilloried by Janet Malcolm, and in a tsk-ing generation of journalistic self-righteousness that followed.

This is, I think, quite true, but I don't understand it. "Fatal Vision" was published after the trial, after MacDonald was convicted. Even had it been a hatchet job, how could McGinniss's views be held to in any way cast doubt on the verdict?