Should we map first the current situation or the desire situation?

Enterprise architecture is actually about mapping the current situation defining the desire situation and finally planning how to fill the gap. Almost every enterprise architecture framework argues that you need to map the current situation first. Is that true? While mapping the current situation we are actually learning what has been done by now, and simply block our mind to new and even better solutions. Maybe it’s better to understand your enterprise business needs and first to find out the best information, system and infrastructure architecture. Maybe, just after you've got the best architecture its time to map the current situation, make gap analysis and plane how to close the gaps? Just a thought...