I don't understand how we can assume that the doubling in speed that occurred during the mid-1990s was also achieved by doubling transistors. Therefore, we can't be certain doubling cost accompanied doubling transistors. What if they developed a different way to speed up a chip's computing speed (besides doubling transistors)? That explanation would also align with the sudden onset in rising costs in the mid-1990's, too. Is the wording/grammar of "each SUCH doubling in a microchip's computing speed" supposed to indicate that the doubling was achieved in the same way?