http://www-users.math.umn.edu/~arnold//disasters/patriot.htm...
No they weren't using floats. The article you've linked to says this really clearly.
> This calculation was performed using a 24 bit fixed point register
They meant fixed precision not fixed point math.
