Once and for all, in the context in which this discussion started, we were
talking about bits as they exist in the architecture of computers and other
binary-based devices -- storage elements having an on and an off value (1
and 0). We were not talking about the variations or nuances in the meaning
of the term as it exists in information theory.

They are the bits, of which 8 make up the customary byte (ignoring, for the
sake of simplicity, other n-bit bytes). The other use of the word bit is not
even tangential to the discussion; it's irrelevant.

On your next sub-topic, I have never, in this conversation, said that I
thought we should use prefixes such as gibi (nor, to be honest, have I said
that we shouldn't). I simply referred to the sources where that is proposed
and dealt with the multiplier values expressed by the prefixes.

As for using b for bar, rather than for barn, I wouldn't use it for either.
As you point out, there are perfectly good (and recommended) SI units for
both. However, if I had a need to refer, for historical reasons, to either,
I would follow the SI bible and use b for barn and bar for bar. I find
violating a standard just for the sake of doing to somewhat wrongheaded.

Regarding the fact that millions use the SI prefixes to mean something other
than is meant within the SI definitions, the problem is that they are using
them inconsistently. The capacity of direct access and linear storage
devices has traditionally been expressed with conventional prefixes
(multiples of 1000). So have data transmission speeds. Because internal
memory has a binary structure, it has been convenient to use those same
prefixes to represent powers of 2, in exponent steps of 10. In general
conversation, the conflict in usage doesn't matter too much. However, for a
formal bid on a contract (for example), it could be critical. Hence, it's
not surprising that there is a movement to standardize (at least for
critical documents) on a specialized set of prefixes.

Finally, congratulations on your computer science degree. I've been in the
computer field for 45 years, by the way, and have been a Fellow of the
British Computer Society for over 34 years. I have a feeling that may
qualify me somewhat to talk, with at least a modicum of authority, about
computers (as opposed, in this instance, to the more general topic of
computing -- which, of course, predates computers by a long time).

Bill Potts, CMS
Roseville, CA
http://metric1.org [SI Navigator]


>-----Original Message-----
>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
>Behalf Of Carl Sorenson
>Sent: Thursday, December 25, 2003 20:58
>To: U.S. Metric Association
>Subject: [USMA:27984] RE: Binary prefixes--not strictly an SI topic
>
>
>Bill, I've got to take issue with a lot of what you said the past
>few emails
>(although not the one to Michael).
>
>You said, "A bit is a bit is a bit. The fact that there can only be an
>integral number is a simple mathematical fact and is implicit, not
>specifiable, in the definition."
>
>Nope.  There are nuances of the term "bit" that can allow for fractions.
>Try reading up on information theory.
>
>You said, "If a specification is actually gibibytes (but
>incorrectly labeled
>as gigabytes)..."
>
>Just because you think that we should use these different prefixes doesn't
>make it incorrect to use different prefixes.  In computer science (in which
>I just got a degree), no one uses "kibi" gibi" or anything like
>that, that I
>have ever seen.  Maybe some Linux people do that, but that is hardly a
>representative sample of the industry as a whole.  We have a
>pretty standard
>way of communicating.  When referring to bytes, "giga" is generally
>understood to mean 2^30.  Perhaps a lot of hard drive manufacturers are not
>following that, but the hard drive I bought two years ago was
>indeed labeled
>that way.  Try looking at the properties of a folder in Windows
>Explorer and
>you will see an example of the traditional terminology.
>
>We have millions of people using a particular terminology, and a handful
>that think it should be changed.  The terminology of the millions doesn't
>become wrong simply because the handful has a "better" idea.
>
>Likewise, I see nothing wrong with using "b" as the symbol for "bar".  I
>seriously doubt anyone will get it confused with a "barn".  Pragmatically,
>"mb" is a convenient symbol for "millibar".  (Personally, I tend to use the
>symbol hPa, but I don't get religious about it).
>
>Carl Sorenson
>
>
>
>

Reply via email to