Voice
Communication in Business Volume 2
Essays on telecommunications,
1981-2002
It is
easy to take telephone numbers for granted. However, if one
understands something about numbering plans, the operation of
switching systems may be clarified. With this background, we can go
on to PBX numbering plans in the next chapter. I wrote this for the
May, 1993, issue of Business Communications Review as sort of
a general review of the numbering plan problem as it faced the new
world of more or less unregulated telephony.
The
explosion of the Internet a couple of years later, combined with
ISDN being completely ignored, made all my predictions seem pretty
silly looking back from 2002.
They've Got Your Number
(Business Communications Review, 1993)
It is a beautiful spring
day in New York in the year 1998, and your pocket dialer has dropped
dead. You shake it, unbelievingly; NOBODY dials numbers by hand any
more. There are just too many of them. A major responsibility of
your secretary is to keep your pocket dialer up to date; it may take
five minutes to put in a name and double check the number, but then
all you have to do is get the name in the little screen and push the
"dial" button. Why do you have to dial 212 and a seven digit number
to get the drug store downstairs? Wasn't it possible, just a little
while ago, to make that call with only the seven digits? When did
Brooklyn become 718? And how can the people in the office upstairs
have a telephone number that starts with 917 instead of 212? It just
isn't fair! How can you make your calls in all this madness...
***
The first time I heard
of a shortage of telephone numbers must have been prior to 1958 in
an internal Bell Labs memo. As I recall, the author predicted the
then-new numbering plan would find its numbers exhausted by the late
1980s or early 1990s. He was right, even though the numbering plan
adopted some 40 or more years ago wasn't half bad. For Direct
Distance Dialing (DDD) to work, a telephone number had to be
independent of the location of the caller; that is, it had to be a
"destination code" from which automatic switches could select trunks
in the proper route; it couldn't be a "route code" such as operators
had used prior to DDD, and Strowger Step by Step (SXS) systems had
used until quite recently in private tie-trunk networks.
The translator in one of
the smart crossbar systems of the day would look at the dialed
number, select an outgoing trunk in the proper group, and then
outpulse the called number to the next switch which could (if
necessary) repeat the process. Translations could be changed easily
to reflect new trunk construction, and provide several routing
options depending on the intensity of traffic. This shielded the
caller from the intricacies of actual call routing, and provided
vastly greater efficiency in the use of toll trunks. Indeed, using
automatic alternate routing to set up connections in real time made
better use of circuits than the manual queuing it replaced.
For DDD, telephone
numbers were standardized with three parts: the area code, three
digits which defined a specific part of the country; the office
code, a second group of three digits which defined the central
office within the area; and the customer number, four digits to
select the particular phone line in the central office. Because 0
was reserved for reaching the operator, and 1 was not used as the
first dialed digit because certain early telephones would generate a
false 1 when taken off hook, the first digit of both the area code
and office code excluded 0 and 1.
To make it easy for
simple translators to tell the difference between an area code and
an office code, the second digit of an area code was always made a 1
or a 0, while the second digit of an office code was never a 1 or 0.
In either case, the third digit, and all the digits of the customer
number, could by any of the ten digits. Thus a telephone number
would be described as N0/1X NNX XXXX (see Sidebar 1).
There were 8*2*10 = 160
area codes (minus the 16 N00 and N11 codes, 8*8*10 = 640 office
codes, and 10,000 numbers in each office. This implied about a
billion telephone numbers which seemed adequate for the US/Canada
population of less than 200 million as of 1960. (Reference 1).
Area codes, unlike zip
codes, were not arranged in an orderly plan corresponding to
geography; rather, they were designed to minimize the number of dial
pulses to be inserted by the customer and then transmitted by
senders, a matter of some importance more than ten years before the
invention of Touch-Tone (DTMF). The most frequently called areas had
numbers with the fewest dial pulses; New York, Los Angeles and
Chicago were thus 212, 213, and 312 respectively. There were a
couple of other guide lines: area codes were not to cross state
lines, for instance, and states with only one area code had 0, not
1, for their middle digit.
Minimizing the dial
pulses transmitted, which shortened call set-up time for the benefit
of the telephone company and customer alike, shows the attention
given to efficient operation. But more important than pulses,
minimizing the number of dialed digits was vital. When New Yorkers
called other New York phones, requiring them to dial 212 before the
seven digit number would have caused enormous delays and increased
the number of registers required to handle call set-up. However,
with hundreds of central offices to choose from, and most local
calls going to some CO other than the caller's, a pattern typical of
large metropolitan areas, it was not unreasonable to require the
caller to dial the office code on all calls.
As a result, local calls
were completed with a seven digit number, and area codes were only
added for calls to other areas. The telephone network used the
office and area codes for routing, while the terminating central
office used the customer number to connect an incoming (or
intra-office) trunk to the called party. Implicit in this addressing
scheme was the assumption that a telephone was served by a
particular CO in a particular area; that is, the area code and
office code identified a specific location.
The Impact Of Reality
Like most things of
theoretical beauty, the national numbering plan of 1955 or so was
somewhat less attractive in the real world. In the first place,
every central office, no matter how small, had to have its own
unique office code; with many small COs in rural areas serving only
a few hundred customers, the number of telephones which the
numbering plan could reach was thus far less than might have been
supposed. Further, because party lines were then quite common, party
access cut down even further the number of lines which could be
supported. (Reference 2).
Second, each area code
had boundaries across which next-door neighbors had to dial ten
digits to call each other. When area code boundaries passed through
the open countryside, this wasn't much of a problem; in the
northeast corridor, however, where area codes boundaries split dense
population concentrations, dialing a ten digit number to call a
neighbor would have caused considerable customer resistance. The
initial solution was not to use office codes from one side of an
area code boundary on the other side; this kept office codes near
boundaries unique, and allowed seven-digit dialing to continue.
Unfortunately, it reduced the number of office codes just where they
were needed the most.
Third, metropolitan
areas existed independent of political boundaries along which area
code boundaries were supposed to run. For example, 202 was the area
code for Washington, DC, but for many years it included,
unofficially, the Virginia and Maryland suburbs as well. This
allowed government officials to call each other with just seven
digits, even if their call crossed one or two state boundaries.
Unfortunately, to call from Alexandria, Virginia, to Roanoke, both
nominally in the 703 area code, you had to dial 703 plus the Roanoke
office code and number. Similarly, outsiders could call a number in
Alexandria by dialing either 703 or 202 as the area code. When I
discovered that Band 1 WATS from South Jersey could reach 202 but
not 703, you can guess what opportunities I saw for ARS when there
was a need to reach the Pentagon.
Finally, when the
national numbering plan was new, there were many SXS and Panel CO
switches in use, none of which had enough translation capability to
deal with DDD. What was usually done was to have subscribers on SXS
switches dial a 1 to escape from their CO, accessing trunks to a
toll switch with enough intelligence to take the number following
the 1 and complete the call anywhere in the country.
Panel switches, designed
for metropolitan areas which required from several dozen to several
hundred switches to serve local callers, had pioneered inter-office
customer dialing starting in the 1920s; they could route on a
limited number of office codes, but not on area codes. They had
originally used an initial 1 to call in an auxiliary register for
toll calls beyond the local area; this practice was expanded for DDD.
With either SXS or Panel, a customer dialed seven digits for a local
call, 1 plus a seven digits for a toll call in the same area code,
and 1 plus ten digits for a toll call to some other area code. Of
course, in areas equipped with 5XBAR and, later, the various ESSs,
all with plenty of translation capability, a caller dialed 7 digits
for a local or toll call within an area code, and ten digits for
calls out of the area (except where high communities of interest
were split by an area code boundary and office codes were not
duplicated as mentioned above).
The Two Initial Ones
Because of its use in
Panel and SXS areas, the initial 1 often signified a toll call.
Indeed, many began to assume, based on limited understanding, that
this was a universal truth. It was not. For instance, I can dial a
seven digit toll call to Atlantic City or Princeton, both in my 609
area, while callers in Manhattan dial 1 plus ten digits for local
calls to Westchester (914), Long Island (516), and Brooklyn (718).
However, some state PUCs have apparently decided that the use of the
initial 1 as a "toll alert," particularly for intra-state and
intra-area-code calls, is a good thing. It is hard to understand how
making the customer tell the CO something it knows but he or she
probably does not is in the customer's interest. Some feel that, if
a toll alert is actually required, the switch should tell the
customer rather than the other way around, but this sort of
reasoning is lost on lawyers.
To see how this impacts
users, until Sept. 23, 1991, it was possible for a business caller
in an industrial park in the northwest corner of the Philadelphia
local calling area to place local calls anywhere in the city, ten
miles or more distant, without dialing a 1. However, to go one mile
north or west into the suburbs, still within the 215 area code but a
toll call, an initial 1 had to be inserted. At home in those same
suburbs, the caller had to reverse the process and dial 1 for
Philadelphia, but not to call the neighbors. Businesses with smart
PBXs usually resolved the problem by having callers dial a 7 or 10
digit number, and then letting the PBX insert the 1 when necessary,
based on an internal translation table; machines do this sort of
thing far better than people.
For callers served by
common control central offices, there was no need for an initial 1
for either local or long distance calls for the first 20 years of
DDD. One simply dialed a seven or ten digit number and the call went
through. It was too good to last. The shortage of office codes
within heavily used area codes drove the telephone industry, about
1975, to start expanding the number of office codes by including 0
and 1 as their second digit. This made them look just like area
codes, and a new way had to be found to make the distinction.
The grand scheme was to
use the initial 1 for calls out of the area code; thus the user
either dialed a seven digit number not starting with 1 (or 0), or
dialed a 1 followed by ten digits where the first three were an area
code. The office code was, by definition, the next three digits. If
there was no initial 1, the first three digits were an office code,
even if the second digit was 1 or 0. Needless to say, New York City,
212, was one of the first areas to have to adopt this procedure.
One can immediately see
the problem: in Philadelphia, an initial 1 meant toll, even though
the three digits following the 1 were often an office code, perhaps
in Allentown, while in New York, an initial 1 was followed by an
area code which, if it represented Westchester or Long Island, was a
local call.
Philadelphia was brought
into alignment with New York 14 years later, and in 1991, Bell of
Pennsylvania indulged in an advertising campaign of vast proportions
involving all possible puns on the phrase "no one." In the mean
time, the Philadelphia area had outgrown its supply of numbers. In
the near future, 215 will be split, putting most of the old
intra-area-code toll destinations in a new 610 area code; customers
who have responded to the "no one" campaign will now have to put
back the 1 as well as an area code for calls to the suburbs.
This is a triumph in
public relations comparable only to that of the old Bell System when
DDD was first introduced. People in small towns seldom had to dial
more than four digits for a local call, and their dials almost never
had letters stuck in with the numbers. With the coming of DDD, all
these people had to go from 4 to 7 digit dialing, and even worse,
had to identify the first two of the seven digits with letters! This
did not go down well, but just when everybody learned to conform to
AT&T's whim, AT&T suddenly reversed its field, "invented" All Number
Calling (ANC), and replaced letters with the corresponding digits in
all office codes. The uproar which followed exploded to enormous
proportions, and even helped pseudo-semanticist S. I. Hayakawa
obtain a seat in the U. S. Senate.
Today, of course, we
have come full circle and even AT&T and the BOCs use the letters on
the dial or keypad to identify a particular telephone company
(10ATT0 or 10NJB0, for instance), copying the techniques of business
people making 800 numbers easier to recall. We all dial 800-Business
Communications Review-1234, or 800-LIBRARY with a fairly high
frequency, and take our revenge on ANC.
I-N-P-A! Hooray! Hooray!
Although the numbering
plan problems of Philadelphia and New York are interesting, they
cannot hold a candle to what is brewing in the Washington, DC, area.
With the supply of office codes in the District finally running low,
the informal extension of 202 to include the Virginia and Maryland
suburbs has been abandoned. Today, callers going out of their area
code have to dial ten digits, not seven. But these calls are still
local; the "toll alert" factor requires the caller to NOT insert a 1
before the 301 to call Silver Springs from the District or Virginia,
but to do so before calling Baltimore. Just where the border between
1 and "no one" lies is doubtless spelled out in complex tariffs on
file for public viewing at various convenient locations. But that is
not the end to the "toll alert" convenience: to call Roanoke from
Alexandria, "toll alert" now requires that 1 be inserted before
dialing 703. Similar absurdities doubtless exist within Maryland's
301 and 410 areas. Unlike New York or Philadelphia, three PUCs have
jurisdiction in the region around Washington. Getting this mess
untangled is going to approach the impossible.
Something will have to
give, however, because INPAs are the way NANPA (see Sidebars 2 and
3) is going to expand the supply of telephone numbers (Reference 3).
INPA stands for Interchangeable Numbering Plan Area (telephone
people call an area code an NPA, choosing not to differentiate
between the thing itself and the symbol that represents it in what
must be an attempt to strike back at S. I. Hayakawa); what is
implied is that area codes will be assigned, starting in 1995, which
look just like office codes. This seems fair enough; if office codes
can have 0 or 1 as a middle digits, area codes should be able to
have use the digits 2-9, making a 640 new area codes available. But
that requires the initial 1 to be used to identify a ten digit
number and thus differentiate an area code from an office code.
"Toll alert" will have to go.
However, the yearning of
telephone people for a universal numbering plan, where all telephone
numbers are alike, is so strong that there is serious talk of going
to 10 digit dialing for all calls, local and long distance alike.
Already studies are being devised to show how much the customers
will like ten digit dialing, just like ANC. As a major advantage, it
will free up the initial 1 for toll alert if necessary; if one
always dials 10 digits, the system knows that the first three are
the area code, no matter what. Another advantage is allowing the
fourth digit to become an X rather than an N, allowing a number to
be NXX XXX XXXX. This, in effect, produces 200 more office codes in
each area code (using numbers which formerly were reserved for telco
use to reach inward operators, test desks, etc.).
There are other factors
pushing toward 10 digit dialing. The area code splitting that has
been so common of late has had as a direct result a great increase
in the number of neighbors who have to dial 10 digits rather than 7
to call each other. The existence of 640 more area codes says we can
expect a lot more area code splitting. This will continue to reduce
the number of calls we can dial with only 7 digits until 7 digit
dialing withers away.
One final factor is
needed to complete this picture: overlay area codes. The first
overlay area code, 917, was assigned on Jan. 1, 1992, to cover the
present 212 (Manhattan) and 718 (Brooklyn/Bronx), providing a whole
area code of additional office codes and customer numbers where such
numbers already exist. With two area codes covering the same
geographical area, the probability of 7 digit dialing decreases as
the new area code fills.
A major step encouraging
10 digit dialing has already been taken by many telephone companies:
a customer-dialed area code, when dialed within that area, is now
almost universally acceptable. In the past, dialing an area code to
complete a call within that area would block the call; this led to
complications in PBX ARS systems when one wanted to overflow from an
FX line to WATS, for instance. Note that 0+ calls within an area, as
when a caller wants to use a credit card, require the local area
code, and many types of dialing to cellular phones also require ten
digits for a local call.
We may have a universal
numbering plan yet.
Numbers For Others
Going beyond ten digits
is for the future; Bellcore and NANPA are pushing INPAs as the
solution of the moment. But the telephone is not the only game in
town any more. Paging, mobile (cellular) radio, and PCS also need
telephone numbers, as will specialized data services which may come
with or without ISDN. AT&T has even started offering lifetime
personal phone numbers in the 700 series; the N00 service codes
suggest there is plenty of room for other service providers to go
beyond 800 and 900 service. All these new number-needers are not
among the Bell operating companies that own Bellcore. As a result,
there is a movement afoot recruit the FCC as custodian of the
national numbering plan, doubtless in the belief that lawyers are
best qualified to handle this task. Remembering the FCC approach to
transmission (forget 40 years of human factors studies and let the
market decide), they may get results they do not anticipate if this
plan goes through.
The "non-geographical"
services do have their problems, however. At the moment, the
portability of 800 service, decreed by lawyers, is giving those
responsible for its implementation fits. Because 800 service is a
callED party feature, there is nothing that can be associated with
the callers or their serving telephone companies which can identify
the long distance carrier to be used or the telephone number to
which the 800 number must be translated to make the call possible.
When a customer can take an 800 number, in which considerable
investment may have been made in terms of advertising, to some other
service vendor, it is obvious that there is nothing in the number
itself which can identify the particular data base which contains
the translations needed; further, operating companies apparently
have to limit their own data bases to the LATA which they serve,
which suggests they will have to add and maintain an
800-number-to-carrier translation in every LATA for ALL 800 numbers.
How this will all come out remains to be seen, but it will doubtless
be repeated with 900 service and anything else identified by N00.
Existing and proposed
radio services demonstrate a different kind of portability
requirement. Here the customer's number, like the customer, really
is portable, and there is no reason to assume that its area and
office code have anything at all to do with the phone's location.
This is no problem for call originations. Cellular phones are set up
to conserve spectrum by allowing the customer to dial the digits of
a called number into a memory and only send them out when the called
number is complete. This burst of information also includes the
identity of the calling phone, needed primarily for charging. It is
only on terminating calls that the flexible relation between phone
number and geography poses problems.
Normally, cellular
systems page a called number from all cell-sites in the home service
area and narrow down to one cell-site when the called phone
responds. However, when such phones roam into a different service
area, information has to be provided to the system to seek them
there, not at home. If the caller happens to know just where the
called phone has roamed, a call can be placed to the remote service
area, the called number passed forward, and the remote system can
then page the called phone. Various schemes are also used whereby
placing a call in a foreign area informs that system of a rover's
presence so that it can send an appropriate message back to the home
service area's data base; other schemes have an idle but turned-on
portable phone send out its identity periodically to create a
location message even if a call is not placed. Modern signaling
networks such as SS7 make it easy for such information to be moved
to the place where it is needed, but as the number of cellular
phones continues to explode, and PCS enters the picture with even
greater potential, one wonders if SS7 as we know it today will be
able to carry the load.
The problem is being
compounded by area-code splitting. As area codes get smaller and
smaller physically, a person who really needs a mobile phone will be
venturing farther and farther afield even for short trips. I have a
friend who starts his drive to work in the 609 area, passes through
908, and arrives at his office in 201. With 90% of his driving in
the "roving" mode, he doesn't have a car phone. A salesman covering
eastern Pennsylvania, New Jersey, New York City and Western
Connecticut would be an even more extreme case of someone almost
never in a specified home service area, and thus would be a
candidate for overloading the signaling systems trying to find him.
One possible use of overlay area codes would be to have one large
overlay just for mobile phones in congested areas with
geographically small area codes for fixed service.
Using the telephone
network to interconnect terminals on different LANs could be very
much like accessing mobile phones. After all, the address of a given
terminal within a single LAN or several tied together by bridges is
just a bunch of bits in the header which ALL terminals read, just as
all mobile phones listen to a page to see if the call is for them.
For both LANs and cellular radio, the location of a particular
terminal within a given service area is irrelevant; another
similarity is that the caller doesn't even know if the called
terminal is turned on.
It would be no great
problem to let LANs continue to develop their own internal
addressing, using a server on each to interface an ISDN BRI for
calling others in the same confederation. Once one or more phone
connections are established, the server can snag the packets
destined for elsewhere and send them out, or take incoming packets
and put them on the LAN to complete delivery. With all the wonderful
broadband architectures available today, nothing this simple would
be of much interest; but if the telephone network's switched Px64
connections are to be used by anything except teleconferencing,
conformance to the telephone numbering plan will have to become part
of inter-LAN addressing at some level.
Buy The Numbers
Bellcore's hope that
INPAs can stretch the present 10 digit numbering for 20 years or so
is not unreasonable if one considers the expansion of conventional
telephony. However, if mobile and data systems are to be part of the
universal numbering plan, ten digit numbers may not last to the end
of the decade. Further, future address may have to include some sort
of indication of service type (do you connect data to fax for
instance, or voice to video?), and specify the bandwidth required.
Perhaps a six-digit customer number, with the first digit specifying
type of service, will be the way to go; perhaps all the new INPAs
can be arranged to identify separate kinds of service. But sooner or
later, the present ten digits will have to be expanded.
Running out of telephone
numbers is as silly as running out of checks while you have plenty
of money in the bank, but it is beginning to look like we could
possibly achieve this goal in the not too distant future. Perhaps
the solution is to let the FCC take charge and auction off telephone
numbers to the highest bidder. The proceeds could retire the
national debt.
References:
1. Notes on
the Network, 1980, Section 2.
2. Notes on
Distance Dialing, BSP 953.200.02, 1958, Section 2. An early
version of 1 above.
3.
Information Letter-93/01-008: North American Numbering Plan
Administrator's Proposal on the Future of Numbering in WZ1-Second
Edition. Bellcore, 1/8/93
Sidebar 1. Talking
About Numbers
In discussing
numbering plans, it is necessary to talk about digits in general.
Traditionally, N stands for any digit 2 through 9, while X stands
for any digit. In discussing numbering plans, it should be recalled
that numbering, dialing and routing are considered to be three
separate concepts. Numbering includes only the digits of the
telephone number itself. Dialing is what is required of the customer
to enter the telephone number in a form the telephone network can
use; it includes extra symbols such as the initial 1, escape codes
for calling foreign countries, carrier identification codes, end of
dialing digits such as #, etc. Routing is the process in which the
entire dialed number is used by the system to set up the call.
Sidebar 2. Who's
Got Your Number?
When AT&T was
broken up, administration of what was then called "The North
American Numbering Plan" was passed to Bellcore, the R&D facility
created jointly for the former Bell Operating Companies. The plan
name has now been changed to the World Zone 1 (WZ1) Numbering Plan
to reflect its inclusion of Hawaii and various Caribbean islands and
the exclusion of Mexico and other Latin American countries obviously
part of North America but included in World Zone 5; the Bellcore
administration group is still called NANPA, however.
Bellcore
administration worked all right until the "natural monopoly" of the
LECs was challenged by cellular radio, PCS and high speed data
networks, to say nothing of CATV which already had its wiring
running in parallel with telco cables in most exchanges.
Organizations offering both POTS and new services have at least a
potential need for telephone numbers, and it is not logical to them
that an entity owned by their principal competition should have
control of the supply and administration of such a vital part of
telecommunications.
All this became
acute when the last of the 144 possible area codes available under
existing rules was assigned in 1992. The master plan is to allow
area codes to look just like office codes by 1995; when the 0/1
limitation is removed from the middle digit, 640 new area codes will
appear as if by magic. Who is to get these codes and how they are to
be used is, of course, the jackpot question.
In an effort to
establish a suitable approach, Bellcore issued Reference 3 as a
target to illustrate possible future actions, and invited all
interested parties to come to the "Future of Numbering Forum" held
just outside the Beltway surrounding Washington, DC, from March
16-18 to begin a joint discussion of the situation and to establish
an orderly way to proceed while protecting the interests of all.
To have INPAs
(that is, area codes interchangeable with office codes), some method
has to be found to let a common control telephone switch know when
it has gotten all the digits it is going to get when a customer
places a call. Thus one topic of paramount importance was the use of
the initial 1: should it signify a 10 digit number is to follow, or
should it be used to indicate a toll call? Another topic is
"overlay" area codes: in densely telephoned areas, particularly
where area codes have been split already and more numbers are still
needed, adding a third area code to cover the same geographical area
is a possibility.
With split area
codes already common, and the possibility of overlay area codes
looming, dialing the area code for many local calls is already here;
another topic requiring discussion is simply going to a "uniform
numbering plan" requiring ten-digit dialing for all calls, local,
long distance, and out of home area code.
As one more
example, the portability of seven digit numbers without a specific
geographic designation (numbers following service access codes such
as 800 and 900, for instance) when a customer moves from one long
distance company to another, or from one geographic location to
another, offers almost limitless opportunities for different groups
to express different needs.
Unfortunately,
FNF-1 did not get to any of these topics. Rather, the hundred or so
people present spent the entire two and a half days defining
"consensus," rewriting the 27 word "mission statement," and
discussing "process." As a result, FNF-2 will be convened about the
end of June to continue the good work, with an infinite series of
FNFs projected. Let us hope some progress will result; the future of
the industry and all its customers hangs in the balance.
Sidebar 3: Some
Terms You'll Want to Keep In Mind
NANPA: North
American Numbering Plan Administrator (Bellcore).
INPA:
Interchangeable Numbering Plan Area. An area code that looks like an
office code. There is no particular name for an office code that
looks like an area code.
FNFP: Future of
Numbering Forum (The P is a digit to identify a particular forum in
the series). A series of meetings which Bellcore hopes will allow
the industry to decide numbering plan issues without resorting to
violence or the FCC.
FCC: A group of
lawyers specializing in auctions and the design of complex
technological systems.
WZ1: World Zone
One. The part of the Earth covered by what used to be called The
North American Numbering Plan. Includes the US, Canada, Hawaii, and
various Caribbean islands. It does not include Mexico or Cuba.
[
Top ] [ Next ] [
Table of Contents
] |
