Ron Burnett wrote:

> Not coincidentally, it's Friday in this big city, and we've had
> 8-digit telephone numbers for several years (with a two-digit area
> code if you're dialing from within Australia, one digit if you're
> dialing from across the big ponds).
>
> We used to have phone numbers that varied from 5, 6 or 7 digits
> depending on where you were (smaller numbers in the country, but
> they had longer area codes).

The US system, locally, is based on a 3-digit area code, the "NNX"
prefix, and an extension.  The "NNX" specifies a switching office (NN)
and exchange (X), and pretty much holds to this day except for some
legacy overlap of NN existing in different cities within an area code
(but long distance).  This the old movie references to telephone numbers
like "Denizen 5-1234" where the "DE" in Denizen was the prefix
switch code (22)5-1234.  Up through the 70s at least and maybe even
later, in many small towns you could dial locally with just the last
four digits, or the exchange digit + last four.

Area codes used to always have a zero or one as the second digit, and
this is how switches differentiated between area codes and NNX codes,
and consequently no NNX codes had a zero or one as their second digit.
This has now of course changed, and above all else led to the mandate
to dial the area code (the switches can no longer tell the difference).

Long distance provider prefixes have similarly changed.  Whereas you
could once dial 10+provider code+area code+number, the available three
digit provider codes (which could not start with a 0 or 1) were
exhausted, and now we have the current "10+10+provider" services
annoying us in the states with their TV commercials.

Let's not even get into international dialing :-)

The whole mess is not unlike IP addressing, which is also overloaded.
And to follow-up on a previous posting about assigning numbers to an
individual, that would be a horrendous nightmare above and beyond that
of the current IP management scheme.  When the internet was relatively
small, routing tables were small and keeping paths to networks was
simple.  Now that we have grown, emphasis is being placed on CIDR
(Classless Internnet Domain Routing) and trying to allocate large blocks
of contiguous addresses to various providers so that specific routes to
individual networks are unnecessary, they can be summarized in aggregate
blocks.  IPV6 will layer on top of this with routing topology in mind
ahead of time, and the low-order bits of the IPV6 address become your
MAC address, making IP address assignment and management trivial.

If you subscribed to DS/3000 or early NS/3000 services, you might be
familiar with a program that HP had where you could obtain an official
network for use by your organization directly from them.  HP had
obtained a block of Class C addresses (ours was 192.30.208.0 and is
registered to us, but the 192.30.x.x block is registered to HP).  But
ironically, we can't use this on the internet as the large providers
such as our upstream will not waste their time routing this single class
C to us.

> Hewlett-Packard Company (NETBLK-HP4) HP4 192.30.0.0 - 192.30.255.255
> University of Tennessee at Chattanooga (NET-UTC-SUB5) UTC-SUB5
>    192.30.208.0

But unless you get *real* lucky, you can reach those addresses here :-(
It does come in handy to subnet for point-to-point links and for
internal network management.

But I digress, as usual :-)

Jeff Kell <[log in to unmask]>