[e2e] evolution of bandwidth as a term

David P. Reed dpreed at reed.com
Fri Oct 3 08:08:44 PDT 2003

In a side conversation about a more subtle ambiguity, a list reader 
reminded me that there are two interpretations of the word "bit" that we 
tend to use.   One kind of bit is a symbol which has two states 
(conventionally called 0 and 1).   Another is a unit of information (or 
negentropy) that can measure the information content of a system 
state.   So, in fact, "bits per second" can be used to measure quite 
different things.   For example, CNN delivers multiple megabits/second in 
the first interpretation, and <10 bits/second in the second interpretation 
(if you watch it all day every day, how many bits of new information will 
have been delivered to your brain in each second?)

So knowing the unit does not tell you what is being measured.   Both 
interpretations are denominated in bits/second, just as axle rotation rates 
and bandwidth are denominated in Hz.

At 08:30 PM 10/2/2003, grenville armitage wrote:

>Christian Huitema wrote:
>         [..]
> > If you
> > want something unambiguous, use Hz or bps, don't use bandwidth.
>"Hz" and "bps" might well be unambiguous, but they are only the units in
>which a measurement is being made. Another word usually identifies the
>network/system characteristic being measured or quantified. And that
>word is....

More information about the end2end-interest mailing list