I would argue that Fitt's law when applied to the touch screen of a smartphone has slightly different results than when applied to a mouse and desktop screen situation.
A smartphone divides into far fewer discrete active areas than a screen, a fact that is compounded further by the huge difference between interaction tool (mouse versus finger) as well as occupying a much smaller space, so there is much less room for Fitt's Law to return a large value. Once focused on a smartphone screen, while using a pointer as large as a finger, there is a lot less room to actually move, both when considering the physical size of the screen and when considering the size of the grid of active areas.
Consider this: if you were using a mouse and screen to point at a set of icons that were same size as smartphone icons and filled the screen and you were asked to click a random one of those icons. Another consideration: if the desktop top screen were filled with the same number of icons as a smartphone (say 20) but scaled to fill the screen. In the first case Fitt's Law would return a high value and the interface would have issues. In the second case Fitt's law would produce a low number. You could also imagine a smartphone screen sized area with 20 icons in which the mouse pointer was confined and think of those results too.
Also, I think smartphone touch screens also react to areas of higher contact whereas a mouse is pixel precise.