>And then we have the lovely example of staring with one... 1 C.E. is preceded by 1 B.C.E. -- and as a result, computation of time passed between C.E. and B.C.E. will need to be handled as a special case.

And even there it's all OK but we're looking at it the wrong way. The year is not a single piece, a nail in the box to count one by one, it's a segment between two particular midnights, namely those that are the border between 31. december and 1. january. And these midnights, being points, are numbered regularly, -3, -2, -1, 0, 1, 2... It's only that the year one is the time betwen NY midnight #0 and #1, year two between #1 and #2 etc.

Just think of how much of computing is influenced by people's unwillingness (or, in some cases, inability) to deal with simple mathematical concepts such as interval, fraction (and the special case, fraction where denominator is a degree of ten). The day with twice twelve, because thirteen hours was too much. The percentage (which has created more confusion than the problems it solved). The distinct names for pieces of the menu system in the horizontal and vertical case - even though their structure and behavior are identical. The funny names of levels in biological taxonomy ("empire", no less! - well show me the emperor then). The different numbering of floors in the US and EU (refusing to have ground floor and 1st above it, because it assumes a zero-based, so they start with 1). The Fahrenheit who was already mentioned (ridiculous scale but no fractions!). Banks and accountants who still stick to their tables even though they don't need them at all, their spreadsheet or even pocket calculator (which they have for decades now) can calculate it with better precision, and will discard a more precise result because it doesn't line up with the rounding error they produce from the tables. Etc etc.

It does keep us in business, but it's largely the part of business where we have to keep doing stupid things because the requirements are stupid. But hey, this is only a rant thread...