I can give you other instances of hawk eye being inaccurate. In the 2nd test 4th innings 2010 RSA vs India at Durban. Boucher was judged lbw from Zaheer but hawk eye showed it to completely miss the off stump and so a lot of south africans unjustifiably felt hard done by. A sky sports commentator picked this up, the ball had in fact reversed in and was most probably going to hit off stump but hawk eye carried the ball along its original trajectory. The south african media being selectively analytical jumped all over this and blamed the umpire when in fact it was hawk eye's fault.
Also in the Perth test Aus vs India 3rd test 2008, Kumble trapped Symonds plumb lbw but hawk eye said it was going over and so a lot of aussie fans claimed that this equated to the 8 obvious blunders that Bucknor made during the 2nd SCG test to hand the aussies the match. There have been many other instances when hawk eye has made an apparent blunder, I see it all the time. You may think im chatting a pile of crap? But Adam Gilchrist agrees with me...and so does Mark Taylor judging by how he's laughing.
YouTube - Adam Gilchrist Commentary on 20.20 1:35
The way hawk eye works is that it takes frames of video footage from multiple cameras and uses each frame to assess the flight pattern e.g. from frame 1 to frame to frame 2 its turned 2mm, from frame 2 to frame 3 its turned 1 mm etc and thus we can predict how much it will turn. Therein lies the problem with hawk eye. If we are using it to predict turn after pitching it is severely limited. For example if Harbhajan bowls a ball and it strikes a batsmen on the pad immediately after it pitches, it is almost impossible to assess the amount of turn because there will simply not be enough footage to take enough frames of footage and hence accurately depict the ball's later path. Same applies to bounce if the ball strikes the pad very soon after pitching. Hawk eye is even more inaccurate when the ball strikes the pad very soon after bouncing when bowlers bowl dooras, googlies, topspinners etc as it will probably not even be able to register the fact that the ball was doing something different to all the other balls in the over. Hawk eye only becomes accurate to predict turn bounce etc when there is considerable time between the ball pitching and striking the pad (which will only really applies to exaggerate back foot shots). In such instances the decision becomes so obvious, that even the naked eye will be able to make a judgement...to the point that if you sat a 100 ppl in a room and got them to watch the footage that at least 90 off them would reach a consensus of out/not out. Hence hawk eye is a redundant technology.