You state that "some people" are innocent and yet change their story. I don't know whether that's true, but it does not change the fact that story-changing is, in my view, at least, strong evidence of guilt. You state that it "can probably [be taken] as weak evidence of guilt," and the author of the present post seems to grant even less than that. Story-changing by someone caught in a trap is a significant element of the human condition, an element you seek to dismiss with the phrase "speculative psychological evidence."

I also don't know whether it's true that "lots of confessions" are coerced and false; but unless you have something specific to point out with respect to such supposed coercion and the changed stories of Knox, you're just throwing up dust.

It's extremely common for people to give false confessions under duress, and some judicial systems take this into account. In Japan, for instance, no person can be convicted on the basis of a confession alone; they must be able to reveal verifiable information about the crime which the police could not have already provided to them.

I participated in a psychological study a few months ago regarding interrogation. I was interrogated for a minor offense which I did not commit by a person who was led to believe that I had committed it. I pretty much had the deck stacked in my favor here; I knew that the stakes riding on the interrogation were trivial, I had, if not an actual alibi, then at least extremely strong circumstantial evidence in favor of my innocence, plus strong material evidence. I was accused of stealing money, which I would not have had time or opportunity to spend or hide, but I did not have any bills on me in the denomination I was accused of stealing, and was free to demonstrate this. I am also pretty good at retaining my reasoning and rhetorical faculties under pressure, making me better at arguing in my own defense than most people.

Nevertheless, it was a really surprisingly stressful experience. If you've never been interrogated for a crime you haven't committed, then I would say that this is absolutely an area where you shouldn't try to state with any kind of confidence what you would or would not do.

As for the evidence that more than one person was involved in the crime, to make a not-that-long story short, that's not really the case. The prosecution had already flagged Amanda Knox, and Raffaele Sollecito by connection, as suspects, on very tenuous bases, prior to processing any forensic evidence for the case, and they continued to interpret the available evidence in light of that hypothesis, but on the whole there was really an absence of evidence favoring the involvement of multiple people over a single murderer.

The evidence for the involvement of other people besides Guede pretty much boils down to

*The number of stab wounds and general brutality of the murder *Evidence of the break-in being staged *DNA of other individuals aside from Guede at the scene.

However, had the prosecution not already had a vested interest in flagging multiple suspects, they would almost certainly have noted that the number of stab wounds on Kercher's body was not actually unusual for a knife murder. Knife murders routinely involve very large number of stab wounds, because it's much harder to kill a person with a knife than people generally expect, and when a person who's not trained or experienced in killing people with knives attempts to do so, the result is usually a frantic extended struggle. The "evidence" that the break-in was staged did not stand up even cursory analysis by the defense, and could reasonably be described as "completely made up," and the DNA analysis only suggested involvement of multiple people after egregious mishandling.

Had the police not walked into the case with a vested interest in tying Knox and Sollecito into the case, it's unlikely that they would have concluded from the available evidence that any perpetrators other than Guede were involved.

Edit: With regards to Knox having specfic reason to alter her testimony in absence of her guilt, it's important to realize that the person she flagged after her extended interrogation was the same person the police had already decided to treat as a suspect based on their interpretation of her phone messages to him0, messages which were in fact quite innocuous when viewed in light of English language colloquialism. They had a suspect they wanted her to finger, and after an interrogation in which she alleges extreme duress, she eventually does so. Then, after having a short period in which to recuperate, she repudiates her own testimony. The suspect then proves to have an airtight alibi.

Under these circumstances, if her changing testimony should be taken as indication of anything, it's that the police probably did push her extremely hard to give a response they had already settled on. It's very unlikely that she and the police would independently narrow down the same suspect who was unambiguously uninvolved, and the content of the proceedings clearly shows that the police, rather than Knox, were the first to point to that suspect.