The computers seemed to be experiencing wild fluctuations in speeds. When the speeds became very low, the computers marked them as unreliable. But they were consistent. Surely icing has the opposite effect - readings would be very slow moving, and inconsistent (- why would two sensors fail simultaneously) ?
I believe it depends on the mode of icing - if the drain-holes did NOT become blocked and the forward opening DID, with the static ports unblocked, then the pressure within the pitot drops toward static, leading to a low speed (ultimately zero) reading. Given that pitot heating was ON, then icing of the leading port ONLY is the most likely scenario. Particularly given the fact that the speeds appeared to recover and read correctly a little later in the sequence (assumption being that heating led to the ice melting...).
Given that you note you have NOT read the entire thread (which is a task, but really the only way to avoid rehashing half-baked ideas), I'm loath to state anything more than you are quite a long way from the current line of thinking. There is little contention that the pitots iced, leading to AP and AT disconnect (per design) with the crew taking manual control in ALT law... it was uphill from there out.
