Talk:North American Numbering Plan

Philadelphia earlier than Atlanta

The Philadelphia region started 10-digit dialing when the 215 area code was split into the 610 area code in 1994. Bill S. (talk) 23:14, 8 June 2020 (UTC)

Ringback number

It might be a good idea to talk about Ringback Numbers. In the old Bell Atlantic days (circa early and mid 1990s), 800-800-6969 was a ringback we would use to identify a line's number during maintenance. Nowadays, 800-800-6969 is a sex line. Derp...

Jeffrey Walton (talk) 21:01, 23 August 2020 (UTC)

Why do countries join the NANP?

What are the reasons behind countries choosing to leave their assigned county code and join NANP? For example, why did Sint Maarten stop using +599? 88.90.14.214 (talk) 18:39, 10 October 2020 (UTC)

I added a paragraph in the international section to address this. kbrose (talk) 21:04, 10 October 2020 (UTC)

The Netherlands Antilles broke up in 2010 as a multi-island "country" following a referendum. So St. Maarten as a new semi-autonomous country (within the Dutch Kingdom) following leaving that bloc looked at all options.[1] They looked at what the other Caribbean islands near them are using. The other Dutch isles are all still using the shared (+559) which at some point will have to be decided on later.[2] Or they are switching to phone numbers of the Netherlands itself.[3].

Sint Maarten's country code (internet-wise) was already switched from the old Antilles one of .an to .sx, they laid the groundwork for their own currency going forward since the bloc currency was going defunct at some point, they created their own regulator for telecom Bureau Telecommunications and Post, and they created their own airport regulator entity too, etc. No big deal. CaribDigita (talk) 08:02, 30 May 2021 (UTC)

Was UK once considered for inclusion in NANP?

Doing so early on would have seemed to make sense, due to Red Line heritage and, again, Canada, Bermuda, and the Caribbean islands. knoodelhed (talk) 06:20, 30 May 2021 (UTC)

The U.S.A. had kicked the British out. Why would Britain put the All Red Line in the USA???? CaribDigita (talk) 21:32, 1 September 2021 (UTC)

Details on the scheme for assigning numbers to areas?

The numbers seem to be shuffled across the individual areas at random. Is there a pattern or scheme for the way they are assigned? Mrsuperboot (talk) 00:22, 31 August 2021 (UTC)

In the US, at least, the first area codes were assigned by population to minimize "clicks" on a rotary dial - 212 (5 clicks) for New York City, 312 for LA and 213 for Chicago (or is the other way around?) etc. - DavidWBrooks (talk) 11:35, 31 August 2021 (UTC)

213 for Los Angeles, and 312 for Chicago. knoodelhed (talk) 01:01, 1 September 2021 (UTC)

No, the klick-theory is nonsense, and is not supported by the contemporary technical Bell System references. The topic of assignment is discussed in Original North American area codes. kbrose (talk) 17:34, 14 November 2021 (UTC)

"Nonesense" might be a bit strong, but I way overstated it above (as I note here: https://granitegeek.concordmonitor.com/2021/11/02/when-it-comes-to-area-codes-i-can-be-very-wrong/) - DavidWBrooks (talk) 18:43, 14 November 2021 (UTC)

@Kbrose: Your source does not support your claim, it only states that no authoritative sources for this method have been identified. In any case, there is no claim that having the fewest pulses for the largest population areas was the only criteria for assigning area codes (subject to the constraint of the middle digit being 0 for single-area code states). And "faster dialing" was a micro-optimization, one which was much more compelling in the case that the middle digit was 1 (e.g. 414 has about twice as many clicks as 212, but 404 has only about 30% more clicks than 202. Even for the case of a middle digit of 1, once you get past the lower "number of clicks" area codes, the percentage difference is reduced substantially, e.g. 616 (near to the worst case for middle digit of 1) is only about 50% than 414, though it's nearly 3 times that of 212.

I haven't seen anything suggesting this was a highly formalized process. Obviously the populations of the largest cities were known, but the population (or more accurately, perhaps, the estimated call volume) wasn't the only consideration. There was some perception about potential for confusion of different area codes (e.g. one might avoid assigning "similar' area codes to nearby areas), and I don't think we can rule out the possibility of some bias related to the location of Bell System corporate locations. Anyway, once you were dialing a 0 as the second digit, that substantially limited the benefit of this strategy.

If you'd like, you could do a regression analysis between population served (at the time) by a given area code and the number of pulses. But there seems little doubt that such an exercise would confirm that the correlation is significant. Fabrickator (talk) 18:36, 14 November 2021 (UTC)

Such regression analysis in fact shows that there was no statistical correlation between NPA assignments and population size or density. Technically, the idea also makes no sense. FIrst of all, pulse dialing was already on the research schedule to be eliminated, with the first trials of tone dialing having been conducted in the late 1940s, simultaneously with the design of the area code system. Second, the area codes were not even used in pulse dialing systems until DDD was finally ready in 1951, operators used push-button dialing for MF tones in the Operator Toll Dialing system for the area codes, where it makes no difference. Saving some clicks on one, two, or three digits saves very little time when dialing a ten-digit number. Many callers had to dial even more digits when counting additional prefixes on many systems (up to three extra digits). Only the area code was effected anyhow, and by the theory of large places, it was not even those that had the benefit of dialing "short" digits, it was those in smaller populations that did. The overall cost gain of shorter register holding times does not reduce the load on the switches much since there are fewer people dialing in smaller places. The most time spent for register holding is in completing the connection, not in the dialing step, so it made no sense to optimize that. In the late 1940s it took often still 30-60 seconds or longer for a long-distance connection to complete. But most systems of the time did not yet have registers anyhow, that had to wait for proliferation of crossbars and new adjuncts on direct-control systems. The click-theory is purely based on fiction, promulgated unfortunately by many media articles who simply repeat the pronouncements of people who simply do not know. kbrose (talk) 18:56, 14 November 2021 (UTC)

@Kbrose: I dont know whether the click-based assignment of area codes was expected to improve the operation of some types of telephone equipment or if it was only about the customer experience. However, your claim that

Such regression analysis in fact shows that there was no statistical correlation between NPA assignments and population size or density.

is implausible. A regression analysis would "always" show a correlation (the exception being the case that the variables are completely independent, which is possible but statistically "never" happens except in the smallest of datasets), the question is really the confidence level. Ha ha, if you actually have the results of a regression analysis, then show us your data and your results. If not, then just 'fess up. Fabrickator (talk) 19:19, 14 November 2021 (UTC)

What makes it "plausible" in your opinion? A handful (may be five or six?) out of 86, when in fact those six can already be assigned from prior technical history? I do have the data, but this is considered original research here. You are going to have to repeat the work. If you actually start it you will notice very quickly the glaring "exceptions" that show up very soon in the upper left corner of the table, most famously quoted is the Boston vs. Western MA conflct. kbrose (talk) 19:33, 14 November 2021 (UTC).

The criteria for establishing NPAs are detailed very clearly and detailed in the design publications of the time. Customer experience while dialing was not mentioned in a single Bell System technical publication by the original design engineers at the time. "Customer experience" was certainly a famously well researched and publicized goal of many other design aspect of the telephone system throughout those decades, as Bell constantly sought to improve its standings with the public to justify its monopoly. If this was the case here, they would certainly have published this wonderful detail of the click-theory, to save customer time. But to the contrary, there are indications that a consideration was to simply expand all telephone number dialing to ten digits, but the human factor in saving at least seven-digit dialing for a few decades prevailed. Another contravening fact is that a system of assigning NPAs sequentially across the country, similar to the Zip-code system, was discarded, and more randomness was deemed better, which probably seems less user-friendly as the area code provides no insight as to its location. kbrose (talk) 20:11, 14 November 2021 (UTC)

Please see The First Thirty Years - Historical Review of NPA Assignments. Notice this mentions register holding times as a consideration. So was customer experience more or less important than the performance or economics of the hardware? I don't think it matters, as these were highly correlated. The presence of counter-examples only shows that there were considerations other than population or traffic levels, but that hardly disproves the claim that population and expected traffic levels were significant considerations. Fabrickator (talk) 21:07, 14 November 2021 (UTC)

I am familiar with the document, and Keevers is indeed a known entity among Bell System writers, but this was written thirty years after the fact, and the author obviously didn't have any better references than still available today. This may even have been the origin of the click conspiracy theory, certain it appears to be earliest, but the assertion is backed by none of the cited documents. His accounts of this are simply guesses. Even with automatic dialing by MF, the connect times under operator toll dialing was still projected to be 60 seconds (down from 2 minutes!), during which registers (if there even were any, most systems of the time didn't have them) need to be held. Even with early DDD, connect times were still about 30 seconds. So, saving a fraction of one second for max. three digits did not improve customer experience. Customers spend way more time searching for the next digit on the dial. The idea is idiotic and a FABRICKation. kbrose (talk) 14:05, 15 November 2021 (UTC)

Before getting to the question of the credibility of the claims made in the memo, I want to address your contention that there's no correlation between "pulses" and "population". You are essentially claiming that the assignment of NPAs to an area is essentially random with respect to population, which presumably would serve as a proxy for the expected amount of inound traffic.

Stating it somewhat informally, you dispute the hypothesis that there is a connection between population and the number of pulses in the NPA assigned to an area. Specifically, for NPAs with a middle digit of 1, you dispute that NPAs with a lower number of pulses correlates to a higher expected amount of inbound traffic. I will also note that, since area code boundaries were not constrained to actual municipal boundaries, the official population figures are of somewhat limited use in such an analysis.

The number of pulses for these NPAs (with a middle digit of 1) ranges from 5 to 19. The average number of pulses is 12; there are 28 potential NPAs with fewer than 12 pulses, 28 with more than 12 pulses, and 8 with exactly 12 pulses. These 8 are apparently "reserved", as none of them were included in the 1947 NPA assignments. Of the 28 potential NPAs with fewer than 12 pulses, all 28 were assigned. Of the 28 potential NPAs with more than 12 pulses, only 18 were assigned.

Thus, there was a clear preference given to usage of NPAs with fewer pulses. In fact, of the 18 NPAs assigned with more than 12 pulses, 17 occupy the slots with the fewest pulses available. (Northern California, assigned to NPA 916, would have had one less pulse had it been assigned to NPA 519.)

Unassigned NPAs were effectively "assigned" to areas with zero population. Excluding the "reserved" NPAs with 12 pulses, and with the exception of the 916 NPA, total pulses of the actually assigned NPAs was the minimum possible.

One thing to note is that NPA boundaries don't generally correspond to municipal boundaries. Nor do they necessarily conform to the population data that would have been available from the Census Bureau. So the information on population of the area covered by an NPA is not necessarily readily available.

In any case, there were potentially 64 NPAs with a middle digit of 1, given that the first and last digits both had to be in the range of 2 to 9. The number of pulses for a given NPA could range from 5 (for 212) to 19 (for 919). The average number of pulses for these potential NPAs was 12. So there were 28 combinations with fewer than 12 pulses, 28 combinations with more than 12 pulses, and 8 combinations with exactly 12 pulses.

There was evidently a decision made to "reserve" codes with 12 pulses, as not one of the NPAs in the 1947 assignment had 12 pulses. Of the 28 codes with fewer than 12 pulses, all 28 were assigned, but of the 28 codes with more than 12 pulses, only 18 were assigned. Furthermore, all but one of those 18 NPAs were assigned to codes with 15 pulses or less. In other words, ignoring the codes with 12 pulses and ignoring the 916 NPA, NPAs were only assigned to codes with the fewest possible pulses. This demonstrates the preference for assigning NPAs to codes with fewer pulses.

As I indicated, the relevant population numbers are not really readily available, at least to my knowledge. But let's consider the 1940 populations as reported for the top 5 cities:

* New York (212, 5 pulses): 7.5 million

* Chicago (312, 6 pulses): 3.4 million

* Philadelphia (215, 8 pulses): 1.9 million

* Detroit (313, 7 pulses): 1.6 million

* Los Angeles (213, 6 pulses): 1.5 million

Inasmuch as area codes didn't necessarily end at the city limits, it's fair to postulate that 213 included a larger population than the area codes for Detroit and Philadelphia.

Beyond these first 5 cities, the census numbers get much closer. Rounding to the nearest 10k population, Cleveland and St. Louis populations are 880k and 820k respectively. Boston is 770k, Pittsburgh 670k, and San Francisco 630k.

The relatively small differences in populations means that ordering these strictly would have fairly limited benefits, especially given the fact that census populations wouldn't necessarily correpond to area code boundaries.

Finally, consider that all the unassigned NPAs (but for 916, this includes all NPAs with more than 15 pulses) were effectively assigned to areas of zero population. Our data is imperfect, but we don't need a formal statistical test to see that the 1947 assignment of NPAs was not random, but was closely tied to the population of each area. Fabrickator (talk) 06:29, 16 November 2021 (UTC) (fixed some typos 02:05, 1 January 2022 (UTC))

That's along the line of the standard Fabrickation, you have fallen in love with a theory that has no support in original documentation but is just repeated in folklore. Repeating a lie doesn't make it fact in the real world, but repeating it seems to work for the prey. Area codes were not assigned to municipalities, but to numbering plan areas the size of states, but definitively on documented purpose bounded by political borders. And yes, that makes the population data available with sufficient precision for any test you would want. But now you have regressed from the suggestion of regression analysis to accepting heuristic handwaving. You have to face the inconvenient truth. You have clearly not read any of the references on the subject, nor do you show knowledge of the technology involved. It's just clicks and size, not in any of the relevant references. But I grant you that there was a pattern of assignment, that apparently originated from a method by which most engineers or scientists commonly would tabulate data in a grid, also shown in the Keevers essay. In such a grid the low-numbered NPAs naturally are positioned in the upper left corner and it is natural to start placing geographic points in those fields first: start at the top left and fill the grid. The question is which localities to place there first. It makes no sense to place small places (by area or population) first, because they have little or no suitable existing infrastructure for automatic (or any) call routing. Call routing was of very high priority in dividing the country into NPAs. You would first consider those that are important and actually have advanced and centrally organized infrastructure for routing, such as New Jersey (were Bell Labs had many installations), New York (AT&T), D.C., Philly, etc. Bingo...and that is the pattern. New Jersey got the first area code 201, D.C. got the second, 202, New York got the third, 212, etc. All the Regional Centers show up in the top left corner of the table. This makes sense, and this is the correlation, and the only one that fits with very high confidence level. There were only eight Regional Centers prior to 1947, and they can be found in the top left corner. Next come those places that were in the process of being designated RCs.This is in line with the historical documentation going back to the 1920s. kbrose (talk) 16:02, 16 November 2021 (UTC)

So now you you are suggesting that the area code weres assigned in a particular order (it would seem to be some approximation to "sum of the digits" rather than "sum of the pulses"), these were assigned based more or less depending on when that area had became or would be expected to become a regional center (which seems to at least be a fair proxy for amount of traffic or for population).

Aside from your evident misunderstanding of the basics of statistical correlation, This latest explanation you have provided conflicts with well-settled rules pertaining to the 1947 NPA assignment (and I'm not referring to the fact that you dispute the notion of assignment based on number of pulses). I suggest that somebody else ought to take up your cause on this issue, because the gaps in your knowledge preclude you from effectively arguing your case, such as it is. Fabrickator (talk) 05:01, 17 November 2021 (UTC)

Your senseless bullying and distractions are noted, when you are the one who has no sources and no knowledge, but that is exactly why you have to resort to your manner. kbrose (talk) 12:32, 17 November 2021 (UTC)

@Kbrose: You dismiss the source that has been provided because of your confidence that the whole thing about assignment of area codes is bunk. But no matter, this is a plausible source, while you are the one who seems to be out of touch with reality. For instance, it sounds as though you think that on a rotary dial phone, one could dial a 10-digit number in just a few seconds, when in fact, each digit would probably take an average of 3 or 4 seconds, while dialing "zero" might take about 7 seconds (just my personal estinates). As a matter of fact, the Bell System had an ongoing process of continuous improvement of the system, to make it more reliable, more efficient and more convenient, and their findings are published. Although the results are freely available online and are searchable, that doesn't necessarily make it easy to find an article that's specifically on point with the issues you raise. However, I ran across the following text from the March 1955 BSTJ which is on point:

In considering the value of speed in a telephone switching network, first let us take the subscriber's viewpoint. It is obvious that any increase in speed of establishing connections gives the subscriber better service. The actual improvement in service, however, turns out to be very modest. For example, if the subscriber saves a half second on each connection through the switching network, he saves one second over-all on the originating and terminating connections of a local call. With the present method of dialing, dialing time tends to obscure any saving in time due to a faster switching network.

They didn't need to calculate things out to three decimal places, this was so blatantly obvious that it hardly needed to be mentioned. And having a larger percentage of traffic dialed more quickly, regardless of the area code from which the call originated, it was just a no-brainer that this would be a material improvement as compared to having a higher percentage of calls take longer to dial. Fabrickator (talk) 04:51, 1 January 2022 (UTC)

What kind of punch are you trying to pull now? It's backfiring, as you demonstrate to understand the subject matter less and less. First of all, dialing any ten-digit number on a rotary dial can easily be accomplished in well under 30 seconds, even when all digits are zeros. Your timing of seven seconds for '0' is absurd. The actual time of pulsing out any digit is *always* less or equal to one second (+/- 10%), but the user needs to wind up the dial first, which can be done in half that time to be generous. So far, this sums up to 10 s + 5 s = 15 s.The third component of 'dialing time' is the time the user needs to locate the next digit which may well be the slowest part. It was also the most error-prone and Bell Labs spend considerable effort in research for minimizing dialing errors. This research resulted in the new arrangement in the #7 dial. This is well enough documented and understood. Of course they constantly improved on the technology. But again, it was not to save time for the customer, but to minimize the high cost of misdialed numbers as entire switch trains would be involved in a failure operation. If the user already knows the number to be dialed, it should take no more than another second per digit to find it on the dial, thereby completing dialing in 25 s. That's very reasonable, even for dialing all 0s. But let's take 30 s for argument's sake. "Optimizing" pulses for area codes would be beneficial for only three digits (30%), or 9 seconds of dialing. If that is cut down to, say, three seconds by clever area code assignments, than dialing an entire number is reduced to 24 s, from 30 s. And that only applies to the handful of area codes that appear to fit the conspiracy, not for the eighty or so others, where no correlation to population size is evident. So, you are asserting that this minimal a saving is justification at a time when in fact no area codes where dialed on rotary dials at all, and not for another four years from 1947 for one single location in NJ, and not for a decade for most of the nation. Yet, during this decade many states experienced post-war population shifts, and numbering plan areas were redrawn, adding over 30% more area codes. None of this could have happened in the framework of 'optimizing pulsing'. It's absurd. They could not have foreseen with any certainty the population numbers a decade later, and if they did, they could have designed the system according to this schema from the onset. But they didn't and, in fact, they already knew that they wanted to eliminate pulse dialing altogether, as the experiments were ongoing .... And now to this article that you found... It has nothing to do with dialing, but deals with the post-dialing delay (PDD), the time needed to complete a call within a single exchange. This paper is known for describing the foundational DIAD experiments in the early 1950s that established the feasibility of using reeds for electronic cross-point switching, instead of using electromechanical means as had been practiced in the Step, Panel, Rotary, and Crossbar systems up to that point. This was a monumental departure from established technology, and had been ongoing for years. Together with the contemporary invention of stored program control (SPC) and other technologies, it was part of the foundation for electronic switching systems, as tested in 1960 for the first time in Morris, IL. Dialing in the Morris system was still rotary, but they improved on the dialing speed by running the dials about 80% faster, at 18 pulses per second. The paper's authors explicitly acknowledge that the switching improvements had nothing to do with improving customer experience, but to reduce cost and complexity of the switching plane. The authors clearly lament the then current state of dialing (pulse dialing) when they write: "With the present method of dialing, dialing time tends to obscure any saving in time due to a faster switching network." With this they knew that they were already working on dual-frequency dialing, and that that would bring the real improvements in customer experience by the end of the decade, for which 'optimized' area codes make no sense. It is puzzling why you would even quote this paper for the current subject matter here. You clearly don't understand the implications. The statement about three decimal places is equally puzzling, and has no link to reality of facts underlying these arguments. It appears you are fishing for arguments by any method, in this case googling, rather than studying telecommunication history and developing some level of understanding. kbrose (talk) 16:43, 2 January 2022 (UTC)

Total line capacity?

Is a calculation available for the total capacity of NANP as its rules are presently configured, and comparing that to how many lines are presently subscribed? - knoodelhed (talk) 04:07, 14 November 2021 (UTC)

Does 988 introduction mean mandatory 10-digit dialing everywhere?

"Seven-digit (7D) dialing may be permissible in areas with single area codes." The way I read Ten-Digit Dialing, I think all area codes are required to implement 10-digit dialing. "[T]he transition to 10-digit dialing must be completed by July 15, 2022." It might just be the ones that do not implement 988 as a central office but since 988 is not N11 maybe there is another reason.Mrdvt92 (talk) 21:20, 2 March 2022 (UTC)

Not everywhere. Maine is not transitioning from seven-digit dialing because it doesn't have any 988 exchanges in the state. Nebraska (I think) also is not changing because it renumbered its 988 exchange. There may be other cases. - DavidWBrooks (talk) 21:43, 2 March 2022 (UTC)

References