[ih] Nit-picking an origin story

[Clem Cole]

On Sun, Aug 24, 2025 at 10:43 AM Barbara Denny via Internet-history <
internet-history at elists.isoc.org> wrote:

>  You might hear 56 kb/sec from other people.  I was surprised to hear 50
> kb/sec from this list somewhat recently.  Of course my memory could be
> wrong but i always thought it was 56 kb/sec.  Did anything change by the
> mid 80s? Your quotation says normally 50 kb/sec from the ARPAnet brochure
> in 1980.
>
That's because it was two very different technologies.

The earlier [circa 1970s] 50K links were created by "bonding" 12 dedicated
analog voice circuits, and the reason they could not get more than the 50K
was the overhead used for the mechanism used to make them appear as a
single faster circuit.

In the 1990s, thanks to the microprocessor revolution, it became possible
to use digital signal processing, and more than 1 bit could be sent at a
time, so even though the Western Electric circuit under the covers only
allowed 1200 BAUD, each BAUD contained more information.  Thus, a
traditional voice-grade POTS line could send as fast as 56K bits/second.

The UUCP world took this into account when running on Telebit Trailblazer
which put its "g" protocol into the modem, and faked out the transfer code
[uucico] since the Trailblazer created a reliable 56K channel. Does anyone
know if anyone ever did for SLIP and TCP?