Re: Types of numerals; bases in natlangs.

On 1/14/06, Paul Bennett <paul-bennett@...> wrote:
> It's worse even than that.
>
> Some systems use base 10^3 and others use base 2^10.
And some use a mixture of both!  Such problems start around the "mega"
range, where a "megabyte" has three different interpretations:
1,000,000 bytes, 1,024,000 bytes (10^3 * 2^10), and 1,048,576 bytes.
The middle one is relatively rare, but its analog at the Giga level
(1,048 ,576,000 bytes) is pretty common.  Usually, if someone's trying
to sell you a hard drive, the reported size uses the smallest possible
unit so as to maximize the reported number.


> There have been efforts to introduce -i and -o suffixes (thus Mi for 2^20,
> or Mo for 10^6, for example) to help cut through this confusion, but I've
> only encountered them sporadically, and frankly only in cases where the
> conversants ought to be familiar with which base was in play anyway.
> They're fighting 50 years of computer-geek groupthink, and frankly that's
> seldom a good idea.
They're fighting a lot more than that with the Mo- stuff - you've got
over a century of metric tradition behind the bare M-.  The proposal
I'm familiar with introduces kibi, mebi, gibi, etc (the first syllable
of the SI prefix followed by "bi" for binary) with abbreviations Ki,
Mi, Gi, etc, for the decimal powers of 2, while leaving the standard
SI prefixes alone for their usual meanings.

--
Mark J. Reed <markjreed@...>