> Since there is no negative power of ten for "k", there is no ambiguity,
> so in a practical sense it doesn't matter.  Since many-to-most of our
> uses of K (or k) are for 1024 rather than 1000 anyway, we are bending
> the SI rules already, so I'm inclined to ignore the whole issue.

I doubt the "most" part, unless you call the 115,200 bit per second baud
rate to an async modem 112.5 kbps, or a 64,000 bit per second ISDN
channel a 62.5 kbps channel, or a 384,000,000 bit per second DSL circuit
a 366.2 kbps circuit (or, for that matter, a 100,000,000 bit per second
ethernet 95.4 Mbps).  Since the `k' in kHz and the `M' in MHz are always
the SI-defined power of 10, and since the phone company uses the SI definitions of `k' and `M' for bit rates, rates involving bits and bytes (ftp excluded)
are almost universally expressed in powers of 10 rather than powers of 2.

Since `k' means both 10^3 and 2^10 when talking about computers, depending
on what you are talking about, the rule I learned a long time ago was
that you used `k' for 10^3 and `K' for 2^10.  This was a very long time
ago, however, when there was less need to talk about `M', let alone
`G', than there is now, so it didn't matter so much that the rule only
disambiguated `k'.  Now things are just confusing.

Dennis Ferguson