Except if you've already done a very similar project before, the unknowns are now knowns. And more importantly, problems already have developed solutions that can be copied or reused, and not developed. So a very similar project should be an overestimate, and a repeated task should take a fraction of the first time.