My high school chem teacher, in the late-80's, forbade us from using calculators. Instead, he made the entire class buy slide rules, and do /all/ of their calculations with those. At first, we thought he was nuts -- why waste time learning to use a slide rule when you could do it all that much faster with a calculator. Then, after a while, we realized that there was a method to his madness -- that getting "good enough" calculations that were within the right order of magnitude, was much more important than getting perfect calculations that were off by a factor of 100 or 1000. Knowing that there are exactly 5,515,631,995,531,583.83 atoms of carbon in 1 microgram of sugar is much less useful than knowing that there are about 5-7 x10^20 atoms of carbon in one gram.
More generally, this attitude gave us the mindset of thinking in terms of being able to do quick estimates that were generally "good enough" approximations, and, more importantly, being able to do those order-of-magnitude estimates to know when something /isn't/ right.
Thanks to how ingrained that thought process became, it's become extremely useful in the 20+ years since, whether I'm doing some estimate involving finance, economics, computer science, or most anything involving numbers.
And, having talked to some other students who'd taken the class with me, they have the same memories -- thanks to being forced to think in terms of appropriate orders of magnitude thanks to the slide rule, it's helped them thru the rest of their professional lives.
