Apparently the optimum information-theoretic alphabet would have e (2.718...) bits, but it's a little awkward to work with a transcendental base. The pain in wo

Apparently the optimum information-theoretic alphabet would have e (2.718...) bits, but it's a little awkward to work with a transcendental base. The pain in working with trinary circuits is why we use binary instead.