Scientific knowledge, it could be said, is based on eyewitness observation. But these are observations that can be made repeatedly, over and over. If one scientist makes an observation, he could repeat the experiment. And other scientists can repeat the experiment. Under these circumstances, eyewitness observation is reliable.
If we had a time machine where witnesses could travel back in time, to confirmed their observations, I would be a lot more confident in witnesses. Particularly if I and anyone else could also travel back in time over and over again to confirm and reconfirm what happened.
But witnesses only witness something once. And their impressions can be false. Or change over time. Or be influenced by what others tell them, possibly within a few minutes of the event. That makes them unreliable.
But we do have a time machine, of sorts. It is the Zapruder film. We can observe it over and over again. We can see the movements of Rosemary Willis, Connally and JFK. We can see the film over and over again to confirm when they move and how they move. And conclude the forward spray seen in frame 313 implies a shot from behind and is not the result of me misremembering what I saw, that the spray when backwards or there was no spray at all.
Good post.
And if a witness has a false memory, it can appear as real and vivid in his mind as an actual memory. Memories are mental reconstructions that can be honesty skewered by impressions, biases, perceptions, distractions, etc.
People may best remember salient items ("I saw the President and Jackie") but be weak--though "real" in their minds--on secondary events ("the limousine stopped"). Problem could be that in an investigation, the salient facts ("the President and Jackie travel on Elm in Dallas") are established easily while the secondary events (the ones more likely to be mis-recalled) take on a new importance.