Quite simply, I am talking about late swing. As I'm sure you know, the ball doesn't always swing in the air before pitching or straight on from the point of hitting the pitch, it sometimes swings late, and if the ball pitches and swings late before hitting the pads, Hawkeye is quite liable to get it wrong. You might say the umpire wouldn't be able to anticipate the degree of the swing either, but at the very least knowing something about cricket and most likely having played himself he could make an educated guess, which is something a machine cannot do.
You are missing the point. The cameras track the path of the delivery upto the point of impact ( with pad, bat, whatever). Any late swing would be recorded by Hawkeye (and I have seen Hawkeye replays depicting late swing). The extrapolation aspect of it is only for the expected path of the ball after the point of impact, and that is plotted based on all infomation gathered till the impact occurs. Basically, it does exactly what the umpire does under ideal conditions.
Yes, usually snicko will show the bat as a thinner, longer line and the pad or clothing as thicker, shorter lines because of the tone of the sound. As I said, in some cases snicko works fine, but usually only in the cases where the umpire could have made a simple decision anyway. In situations like bat-pad catches and situations where the ball may have hit bat and pad at roughly the same time - ie: the situations where the umpire is most likely to be in doubt and refer it to the third umpire - it is inconclusive more often than not.
Chances are that it would prove conclusive more often than the umpires ever will be. That is reason enough for it to be considered superior to the umps. What about instances where the crowd is so noisy that the umpire cannot hear properly ?
