My problem with multiplying my estimates by Pi, or any n > 1+epsilon, is that one's boss often looks at such estimates and asks "How can it take you so long to do just XYZ?" and they would rather, in effect, be given a shorter estimate initially and settle on delays than be given a more conservative/inflated estimate, even if in the latter case there is rarely a delay.