You're right about one thing - you see what you want to see. I'd swear if asked that he's using his thumb and finger to draw a rectangle shape on the desk representing the keyboard he describes at the same time, but there's no denying the circle on the screen, even if just a shape bounding his touch points, rather than an object he selected and zoomed.

The later clip with the two fingered multitouch zoom of the person that recoiledsnake links to is more compelling, even if the gesture doesn't quite match that we know as pinch to zoom.