This is how i understand how the camera works for the view transformation. I was gettig confused. Initially the camera will point at the orgin in the centre of the screen usally. If I want the camera to move to the right 2 unts along X axis. The view (screen)shifts to the left and I am looking at (2,0,0) in the centre of the screen . If I want the camera to rotate 45% as I look up along y axis. The view (screen)rotates down and I am looking at (2, 45 deg up, 0) in the centre of the screen but I am looking up and the terrain has shifted down. If I want output the view matrix I should see the coordiantes in theory? Is this right?

Basically yes.

But i think what you should do is read about matrices.
Especially doing transformations using matrices.

Clicky : first result returned by google for "matrix transformation"

if i want to know what point I am looking at eg a centre of the screen then what do i do?

There is little value in knowing what you are looking at unless:

a) You are setting your initial lookat point
b) You are doing intersectoin

Otherwise it is just a case of doing rotations and translations of the view matrix.

what i didn't like is that you are just plugging a few matrix values and the rest you take as it works fine. There just seems 1 way only to move the camera so accept it.