If you try and follow things through from the police / prosecution point of view it is interesting to see how the errors occurred. Now it is difficult to know exactly who knew what when so there are some assumptions.
One of the key problems seems to be that the police make assumptions and once made have difficulty in revisiting the assumptions. So very early the investigators conclude that the break in was false. It is not unreasonable to think this. Evidence pointing to this was the suggestion that the outer shutters were closed which would mean that the stone could have not been thrown through the window, the stone appeared big and the window high for a break in and other routes easier. The fact they were led to think nothing had been stolen (not true as it happened), clothes were scattered on the floor with glass fragments on the top as if the clothes had been thrown around first then the window broken. Although grounds for suspicion, none of these are proof of a staged break in. Detailed forensic examination of this room did not occur. There was not detailed examination of the route of entry from the outside, no one looked under clothing to see if glass was under clothing, the glass was not examined to confirm direction of breakage, and the distribution of glass fragments was not analysed which would have revealed the direction of breakage. The police went with first impressions, psychologically these are very strong and difficult to reverse. Our own Vixen is a good example, Vixen always references the first reports, and has difficulty moving on from the first impressions. This assumption of staging then led to the suspicion that the case was an 'inside job'. One confounding issue was they may not have initially been aware that Romanelli had entered the room , and removed item(s)disturbing the crime scene. (I suspect in addition to her computer she removed drugs as she was known to supply cannabis to Knox, and none was apparently found. So the police looked to Kercher's flatmates, indeed most murders are not by strangers, as was the case here. If we then add in the odd to the police behaviour of Knox and Sollecito their initial suspicions were reasonable.
I think that it was deliberately planned to bring Sollecito and Knox to the police station together for interviewing. To try and pick a hole in the story. I suspect the casual discussion in the corridor moving to an interview was a deliberate ploy to try and catch Knox off her guard. We know how Sollecito because of confusion over the day appeared to withdraw Knox's alibi, then Knox appeared to have an arrangement to meet someone that evening that she initially denied. leading to increasingly confrontational questioning the collapse of Knox and an apparent admission of involvement. Case solved, confessions trump evidence, then things started to go wrong. What never seemed to happen was a case review to say 'did we get this wrong?'.
Initial evidence supported the police view (and shows the why forensic scientists have to be neutral and not seek to provide evidence of guilt). The initial report on the bloody prints was that the prints matched Sollecito's shoes. This as everyone accepts was clearly an error. But it clearly demonstrates the bias of the forensic scientists. (In the UK the police have a statutory duty (PACE) to seek evidence to exonerate suspects). I suspect the postal police claimed time of arrival added to the evidence of guilt. The problems with computer forensics meant the police did nt have early evidence of Sollecito's being at home after 21.00.
Mignini should have followed the spirit and letter of the law, when rung the night of questioning he should have said arrest them and I will question them tomorrow in the presence of their lawyers. Instead trying to skirt round their right to counsel Mignini elicited a spontaneous statement, then kept Knox without access to counsel during which time she produced further written statements.
So what went wrong? Guede's friend reported him to the police as the possible murderer, forensic investigations came up with an unknown male, subsequently the handprint being identified with Guede. Lumumba had an alibi. The shoeprint evidence against Sollecito collapsed. The bloody print on the bath mat was not matched to Sollecito until later. The knife DNA result probably became available about this time. 17 November? The errors in documentation and oddities in processing suggest this was hurried through. The report from Steffanoni would have reinforced the police view that Solicito and Knox were involved. It is just at this point Guede is arrested and interviewed, presumably some recognition of his possible involvement in burglary comes to light.
What never seems to happen is a review and an attempt to question what the evidence says. But to be fair to the police, the bias in the forensics misled them.
Looking back, the pursuit of Knox and Sollecito is comprehensible. It was a consequence of an immediate assumption about the breaking that was never properly assessed or investigated. Mistakes in the forensic science that delayed exonerating evidence like the computer data (and possibly CCTV), and a focus on finding evidence to support the police assumptions rather than realising the importance of testing assumptions by looking for exonerating evidence and knowing the limits of interpretation of the evidence. (Luminal positive prints may or may not be due to blood, DNA does not necessarily mean direct contact, a blurry footprint is a blurry footprint.
