The article doesn't make it explicit, but from the facts it presents it seems like the fundamental difference between the numeric keypad and telephone keypad is:

- With the numeric keypad, you want an extra-large 0 at the bottom that can be operated with your thumb, because zeros are so disproportionately common in real-life numbers like prices. And smaller numbers are used more than larger numbers, so you put the smaller numbers closer to the 0 so you have to reach the least, and wind up with 7-8-9 at the top.

- With dialing phone numbers, zeros aren't more frequent -- in fact they're less because phone numbers can't start with them (in the US). For local numbers, all digits 1-9 are used with approximately equal frequency. So the keypad starts with a natural numeric order of 1-2-3 at the top in reading order, and puts 0 at the bottom since it feels weird to start counting with zero (just like QWERTY keyboards start with 1, and puts 0 after 9), and because it has the special function of calling the operator.

So there seems to be an actual logic to it.

A line early on in the article caught my eye:

> they serve the same functional goal — input numbers

Well, yes and no. Same as how, when it comes to data types, it often has to be pointed out to inexperienced developers that a phone "number" isn't a number in the mathematical sense - you can't add or multiply 2 of them together to get anything meaningful. It's an identifying string, that happens to use only digit characters. "123" in a telephone number is three individual unrelated digits, whereas "123" in a calculator represents the number one hundred and twenty-three.

So the functional goal isn't exactly the same. One is entering individual characters, but on the other you're more likely to be thinking "one hundred and twenty-three" as you type its digits.

It may or may not be related to the actual reason for the inversion of layout, but the subtle difference felt like a (possibly minor) error in the initial premise.

It's sounds silly when the Android auto in my car reads a text message. "Message from twenty four thousand, five hundred and thirty nine ..."

I worked for a couple summers as a "relay operator"; in the USA there is (was? give the hateful time I suppose...) a law, "Americans with Disabilities Act" to the effect that people unable to do a thing should be able to do the thing. Roughly it means "people in wheelchairs should be able to access buildings" and "people unable to see should be able to read newspapers" and "people unable to hear should be able to talk on the telephone." and so on.

The "let people unable to hear talk on the phone" accommodation was to provide actual teletype machines to people who can't hear (at the time, many of these devices were some hideous 75 baud 6 bit monsters where there were limited punctuation and only upper case); the phone company would then also run a service where they had operators (I was one, for a couple summers) where people would call this service and that service would act as a bridge (or, "relay") to the other kind of device. So deaf people could order pizza, teenagers could call their friends and talk about teenager stuff, etc.

Specific to this conversation, the "relay operator" setup was a telephone system billing computer (that would also setup the phone call) and a standard terminal that'd interface with the person with the TTY. There were 2 800 numbers; one to connect to a TTY and one to connect to my ears; people would connect and ask to talk to a peer, and I'd enter the billing / call info into the phone computer, then actually do the talking on the terminal.

Each of these systems had a very distinct keyboard (the phone co keyboard had deep wells on the home keys; the terminal had "normal" nubs on the home keys), and I spent a ton of time entering phone numbers on the phone co's billing computer, with my right hand. To this day, my right hand touch-types "phone company" numbers and normal "ten key" (I did a lot of data entry at other points in my life) with my left hand.

[edit]

oh -- these things, though "ttys" were called "TDD" or "TTD" or some silly name to imply they were for deaf people, though they were just ttys; the cooler kids, calling that relay number, had 300 or 1200 or even 2400bps modems; I think that's as fast as the phoneco's relay terminal went, though)

GA

I noticed ATM keypad in different countries use 1-2-3 or 7-8-9, I have yet to figure out if this is based on something, it seems fairly inconsistent with language/history/colonialism

7-8-9 is the “standard” for calculator keypads but Bell Labs (supposedly) did some research and found 1-2-3 was more intuitive for users when designing the touch tone telephone keypads. When ATMs were being designed, manufacturers in the US/Canada/Europe emulated the telephone keypad while manufacturers in Asia emulated the calculator keypad.

These days you can get PIN-capable card-readers with touchscreens. Some of these models randomize the button layout, which can be interesting to those who rely on muscle memory to type their PIN. Especially given that some shops have the readers physically attached or secured, so you have to type the PIN at an odd angle.

Some examples here[1].

[1] https://ux.stackexchange.com/questions/144937/why-do-some-ne...

As I was briefly a beneficiary of SNAP EBT ("Food Stamps") I was subjected to the process of managing that debit card. When checking out, for example from Walmart.com, a popup would appear so that the user can input his PIN.

The popup was served by the SNAP EBT provider, and it would randomize the PIN number pad. So indeed, you couldn't rely on muscle memory to input your PIN because the number pad changed every time input was requested. It seemed that the mouse was also required for this input, rather than the keyboard.

Relatedly, TV remote controls seem to have settled on the telephone layout with 1 in the top left.

I have also used a few kiosks with a keyboard that has its physical keys arranged in alphabetical order, which is just as confusing.

Seems reasonable to have the most frequently used numbers close to the user; I wonder if there might be something of Benfold’s law involved, where lower digits are more frequently used. https://en.wikipedia.org/wiki/Benford%27s_law

That's my intuition. I spent a lot of time entering stacks of checks into 10-key calculators at my family's businesses growing up in the late 80s and mid-90s. Most entry used the bottom two rows of digits (the zero, double-zero, and 1-3)-- a lot of $10, $20, and $30 checks.

Man, reading all this makes my brain itch, I still mess up on ATM keypads if the layout flips. You think people just adapt even if it never makes sense, or does frustration actually change design over time?

> Picture the keypad of a telephone and calculator side by side. Can you see the subtle difference between the two without resorting to your smartphone? Don’t worry if you can’t recall the design.

Pfft, I have both on the table beside me. I live in a different timeline, I suppose.

What subtle difference? On a telephone the numbers are in a circle, whereas on a calculator they're in a square. They're completely different.

It's possible that at this point, a majority of the people who have ever used Excel to calculate anything have done so on a laptop that doesn't even have a numeric keypad. Certainly, that fraction of the cumulative historical Excel user base has been growing.