U brilyent ilustrraeshun uv wie peepool ignore standerdz baadeez. 
(That's one group's "standard" English spelling for "A brilliant
illustration of why people ignore standards bodies".)

> > Interestingly they seem to use B for byte and b
> > for bit as you and Wojciech both suggest.

That's been standard for at least fifteen years.  (That's
"standard" in the sense of being the agreed usage among the
linguistic community of people who read and write about 
computers, not in the "standards body" sense.)  Real standards
in human systems arise out of usage, and a standards body has
to have a damned compelling case to impose new ones.  Many
elements of SI cut the mustard on that front and have been
widely adopted; many others did not, and have not.  "kibibit"
and "gibibyte" don't even pass the giggle test.

> Besides the general 'rule' that k is 1000 and K is 1024

I never heard that one til today, and I've seen plenty of
counterexamples.

> Data rates are typically defined in decimal numbers, storage capacity
> has transitioned from binary to decimal numbers over the last decade.

No; nerds have always meant 2^20 bytes when they said "megabyte", and
marketers have always meant 10^6.  The reality is that when talking
about storage capacities these days, it rarely matters --- and when
it does, you specify.  But if you specify by saying "Mebibytes", I'll
laugh.

---Alex