You think that's bad, try talking to computer people. They think a kilo of their bits means 1024 of them, not 1000!
After many years and different widths the smallest common unit of exchange evolved to be a power of two: 8 bits are a byte, not 9, 10, 27, 30, etc; though Unicode might have been better off if 32 bits were the base character. However then older systems would have been at least 75% less useful.
Physical addressing units are also binary, because literal address lines (or their virtually latched equivalents on a serial bit-shift register) are used.
ROM, RAM, and correspondingly non-volatile writable storage inherit this basic premise.
So a non-sales computer person speaking 'kilo' or 'mega' or any other SI unit in relation to a computer part means it will hold or be able to transfer at least that many SI units in the quoted units. Everything gets rounded to the nearest useful neighbor. A digital kilo is not 1000, it is 2^10. A digital mega is correspondingly also 2^20. This is very similar to the exponents in SI units.
For that matter, humans have 8 fingers and 2 thumbs. Wouldn't an octal numbering system make more sense? Though for that we'd have to come up with new unit names near 2^3, 2^6, 2^9, etc that don't sound silly.
Each of the 4 long fingers has 3 phalanges, so an interesting possibility is to use each thumb to count up to 12 on each hand. Given that many ancient units of measurement where base-12, I wonder why we didn't develop base-12 languages.
Wish we could all start understanding the difference between kilo and kibi, mega and mebu and so on for bits and bytes.