The "3 seconds" stuff (and similar) is subject to gross misinterpretation and misunderstanding by pro-guilt commentators and the convicting lower courts alike.
The call durations given in the logs refer explicitly and solely to the CONNECTED times. That is to say, the time elapsed after a connection to an open line was made. And in turn, that is to say it does not include the time during which the called party's phone was ringing.
See, the ignorant pro-guilt commentators and convicting courts seem to be under the impression that if, say, a call is listed in the records as being of a 3-second duration, this somehow means that Knox (in your case above) called Kercher's Italian phone, let it ring for just three seconds, then hung up. In other words, they interpret it as evidence that Knox was just going through the motions of calling Kercher, without even giving the phone enough chance to be answered (the follow-on "conclusion" being that Knox knew Kercher was dead, but wanted to try to establish a claim that she'd been calling Kercher out of concern etc).
But, as I said, the 3 seconds strictly refers to the time of connection of the call. And in the case of Kercher's Italian phone, this means the time of connection to the Italian voicemail message on Kercher's Italian phone. So what actually happened was that Knox called Kercher's Italian phone, found it diverted to voicemail, and had 3 seconds to realise that this was what was happening before hanging up (Knox had already tried several times to reach Kercher on this number, and there was little point in her leaving voicemail messages at this point, so it was entirely reasonable for Knox to terminate the call after 3 seconds of the voicemail message).
And to answer your further question, Kercher's Italian phone was actually switched OFF from around 10pm on 1st November until the Postal Police turned it on around 12pm (lunchtime) on 2nd November. So in fact Kercher's Italian phone was not ringing at all when it was dialled - it was diverting straight to voicemail. A fact which further negates the incorrect and misleading "conclusion" attempted by pro-guilt commentators.
Kercher's Italian phone was switched off on the night of the murder. By someone. And Kercher's English phone was left switched on on the night of the murder. However, the strange short-code mis-dialling on Kercher's English phone at around 10pm on the night of the murder tends to suggest that someone was fiddling with the phone trying to do something with it. And it's hardly unreasonable to suggest that this someone was trying to turn off the English phone, but was perhaps unfamiliar with the handset and the English language menus and screen displays.
My long-held supposition is that Guede, having killed Kercher, took both her phones, with the intention of turning both of them off before he left the cottage (he didn't want the phones ringing when they were in his possession after he left the cottage, for obvious reasons). He managed to turn off the Italian one easily, but tried and failed to turn off the English one (doing those strange aborted mis-dials in the process). He therefore had no choice but to leave the English one switched on. And then, when that English phone suddenly sounded and lit up with the incoming MMS message, almost certainly while Guede was walking along the road round the outside of the old city wall near to Parc Sant'Angelo, his fears about being caught out by the phones intensified and he decided to simply ditch both phones. He hurled them into what he thought was scrubby hillside, but what was in fact the neatly-tended garden of Sig.a Lara.
: 
