Keep in mind that this is based a quarter of a planet away, through documents translated by others and trial analyses done by others. Also, IANAL.
In response to a comment, “But that is the job of the defence I suppose, to continually raise doubts about the prosecutions evidence.”
I think an effective defense weaves together a story that effectively counters the prosecutions, or an effective defense focuses on one thing that undermines a case. In the O.J. trial, the defense focused on a glove not fitting, and you saw how that went. In the Scott Peterson trial in California, the defense tried to piece together a story that someone else must have done it, based on the testimony of two witnesses. In that case, the defense’s story was undermined by the lack of evidence to support their counter-claim, since it was based on a few witness testimonies; it was also undermined by the limited evidence the prosecution presented- in particular three missing boat anchors in peterson’s boat garage.
In the Knox/Sollecito trial, it looks like the defense tried a shotgun approach to discrediting everything presented. This approach was undermined by their clients’ own statements. In the end, the defense probably appeared desparate more then a coherent counter-claim to the story of the prosecution.
The defense couldn’t effectively weave many counter-stories; trying to take the tact that Knox was a pious girl to counter the sex gone wrong claim is undermined by Knox’s own past. Trying to weave together police negligence through all the data while your clients are coming across as untruthful also probably came across as desparate.
If the defense had followed the O.J. approach and focused on one item, such as the DNA on he knife, it may have been more successful. A quick internet search revealed in the UK a large trial was overturned on the basis LCN DNA testing is unreliable. If they’d brought in one of those experts, they’d risk the image of a ‘foreigner’ coming to the rescue, but I think that could have been countered by the expert testimony provided. Apparently LCN DNA testing can be unreliable because in research (from what I can tell) they reportedly found that repeated LCN DNA tests on the same amount of starting material can come up with different results; what you get one time may not be what you get the next. This would introduce a high statistical improbability of a single LCN test producing an exact match to the victim, which was the primary forensic evidence tying the two together. Once that point is made, then it is the prosecution’s case that starts to look obsessive and almost desparate. And as the OJ trial shows, the contradictory stories of the defendent fall in weight.