The original 911 system was rolled out when many phone switches were electromechanical. Step-by-step and crossbar offices didn't log calls, didn't pass line identification data with the call, and couldn't route based on the source of the call. But they had hard-wired relay logic in the originating registers (the units that receive dial digits) to recognize area codes (second digit 1 or 0 back then) and service numbers (second and third digits 1). Those were forwarded to a toll office, along with automatic line identification info. So 911 was forwarded to a toll office, like a long distance call. "999" was considered as the emergency number; the UK used that. But US switches didn't have a special case for 999, so that was hard to retrofit. Hence 911.
Toll offices, which handled long distance calls, were computerized by then. They could be upgraded to deal with 911 routing. This requires a database to map phone numbers to the appropriate Public Safety Answering Point. The idea of having a computer with a disk drive in the call chain was radical at the time, but it was made to work. The original computers for this were Western Electric 3B machines, set up as duplex pairs.
The toll office 911 system became a problem with the breakup of AT&T. AT&T retained most of the toll switches, while the local exchange carriers got the local offices. (Now, of course, with weak antitrust enforcement, they've all merged back together again.)
My job is upgrading and maintaining 911 and other life safety systems.
The article is correct, advancing technologies are not reflected by additional capabilities by PSAPs, DISA's or ILEC's (The ppl that foward your call to a responder). It was only recently that IP based 911 sending of ANI/ALI (Telephone/Location data) data has been implemented on a large scale. In Canada, it ran on an old packet switched network for ages!
Their are different flavors of 911. E911 (Enhanced), V911 for VoIP phones, and recently the addition of Wifi based calling among others. The original 911 systems was designed when phones were static, they didn't move. Nowadays, with cell, voip and now, wifi devices, theirs no telling where the call (device) originates from. Yes, most send long/lat data, but that is based on triangulation of cell towers and not accurate enough. To further complicate things, long/lat doesn't take into account altitude. In a urban setting, the responders maybe at ground level wondering just which building and floor originated the call (yes, their systems to deal with this but not widely implemented, and, failing that, if the caller is unable to speak, you have a larger issue).
Most responders don't even have the ability to map a long/lat. So at this point, the accuracy is moot.
All that being said. The E911 system works well. Cell/VoIP/Wifi, not so much. The call will generate a response, but it's hardly efficient.
Then there's the idea that they'll go totally internet based, allowing forwarding or conferencing in multiple psaps, translators, etc. Somehow, they're supposed to secure all this, but I'm doubtful. It'd be one of the largest secure federated networks in the world I think. And they aren't exactly tech savvy. Doing one of the first wide-scale VoIP 911 services, we had psaps call us to say they wouldn't take phone calls from VoIP users, even on their landlines, because "they might send us viruses".
NENA folks also wanted the FCC to mandate an IP location system to be available on every Internet connection - what a privacy mess that'd be, when any app with IP access could get your exact location down to the apt number.
I trie experimenting with what I named "Advanced 911". For calls to places where we couldn't deliver location info, we'd prompt the answer to press a key to have us speak the info. But getting adoption for that was super difficult.
Also, the stress involved in those jobs, wow. I thought I could handle stuff, but auditing problem calls where people were dying, panicking, and didn't know where they were - that's traumatic. Also made me really dislike how cavalier 911 is handled by the telecos.
We have dedicated staff whose job it is to audit calls. ALL calls are recorded and we do screen snaps every 5 seconds, or, anytime the screen changes. Were talking terabytes of data. We keep it going back many years.
I'll vouch for the stress of the job. If I f'up. Somebody can die. Same thing for the operators. They get 3 months of training before handling calls. Most wash out before their training is completed. Most, shortly after that. The churn is unbelievable. We hire constantly. All operators max out their sick/vacation days. It's an unpleasant, boring, terrifying job.
This sounds like an interesting technical problem. If you're up for chatting about it further, I'd be interested in hearing more about it--my email's in my profile.
> Most responders don't even have the ability to map a long/lat. So at this point, the accuracy is moot.
This is genuinely surprising to me. The operators having trouble makes sense, because phones are something of a rolling disaster, but responders aren't using GPSes that can do this?
I'm sorry, I should clarify. Their are many layers of 'responders', their maybe just one, your call was routed to the local police. It may have been routed to call center in that area, who in turn broker the call, etc.
Imagine a situation where your traveling down an interstate out in the boonies. Theirs a wreck. You call 911 from your cell. The responder will receive an approximation of your location, likely the closet cell tower. From that, and before being able to send a responder, they must determine WHO to call based on your location. Just because your closer to town B than town A doesn't mean town B gets the call. Town B may forward your call to town A. Then, you have to determine where to dispatch too. It's doubtful you know the closest mile marker. Only your direction of travel, and the last area that you can remember.
It may have been routed to call center in that area, who in turn broker the call, etc.
Yep. I was a 911 dispatcher for a while and can confirm this. I worked for Brunswick County (NC) 911, which borders New Hanover County (NC), Columbus County (NC) and Horry County (SC). Cell phone calls were usually sent to the right place, but if you were way out in the boonies on highway 211 near the Brunswick / Columbus line, if was about 50/50 which 911 center would get the call. And we'd occasionally get a call that was dialed in New Hanover County, Horry County, and - rarely - even further away. So on those calls, we had to try and quickly determine exactly where the caller was, so we could relay the information to the other county for dispatch. Sometimes if we weren't exactly sure and the call seemed to be near the line, we'd dispatch units and request the neighboring county to dispatch theirs as well, and then let the responders figure it out when the arrived.
I took a call once that, as far as I could tell, turned out to be several counties over... something like Pender or Duplin County. We didn't even have their contact info logged anywhere, since it was thought (at the time) that there would never be any need for us to contact, say, Duplin County. I had to call New Hanover County, ask them if they had the number for Pender (who they border), then call them, etc. The term "clusterfuck" comes to mind, but luckily that only ever happened once that I can remember.
Note that this was in the mid 90's and I'm not sure if things are better, worse, or the same now.
There's some interesting technical history behind some of the emergency number choices:
999 in the UK (from 1937) was chosen because public payphone could be easily modified to make it a free call.
111 in New Zealand (from 1958) was chosen because the system was implemented using British Post Office equipment that already supported 999. But New Zealand phones pulsed in reverse so 111 on a New Zealand phone produced the same pulse as 999 on a British phone.
000 in Australia (from 1961) was chosen because 0 was already used for trunk access. On an automated rural exchange, 0 would connect you to a main centre. In remote communities it was 00. This meant that dialing 000 through an existing remote exchange would at least connect you to an operator in a main centre.
I once worked at a bourgeoning hosting provider who had one of the higher floors in a building that was condemned, but couldn't be torn down because it was home to a phone switching room that had been a part of the original 911, and the designers of technology then simply did not even imagine their systems being EOLed, so it turns out, no living or dead being knows how. I'm curious to know if that's still standing.
The freight elevator shaft had a shattered wooden freight elevator carriage at the bottom of it. We took the stairs.
Those kind of buildings are all around the country and serve various purposes - some of which specifically manage 911 calls. I can't imagine ever decommissioning one. Sometimes even gaining access takes something just shy of an act of god. While AT&T has re-consolidated in a major way, there are still a lot of external concerns to deal with.
Somebody would have to be very motivated to push through turning an interconnection point like that off. The politics involved in doing so would be extraordinary - even if there was common agreement from the start that it was time to let it go. I don't even know that you really could initiate a project like that from within a Telco.
It is a bit surprising that it was common knowledge that part of 911 ran out of there. While it's not "secret" information, the purpose and location of most similar properties are typically not openly advertised. Most people don't even know where their local CO actually is.
"Hotline 911 received 2,475 recorded calls, between 12 a.m. to 7 a.m. Monday.
Out of the total, only 75 calls were legitimate while the 1,119 were dropped and 1,356 were prank calls."
In 1965 there wasn't much international direct dialling yet; mainly to neighbouring countries. The rest was done via an operator. International exchanges tended to buffer the entire called phone number anyway, even back in electromechanical times, which meant they could also translate the pulses if that would have been necessary.
goodness, what a thought ><. Not too far back we called 911 on a cell phone that we didn't know had the microphone shorted when it had gotten wet earlier in the day but otherwise was completely operational. Took several minutes to figure out what was going on. I think we wound up texting a friend that we'd call them and to verify they could hear us or not. We wound up having a headset handy luckily enough.
No. It was a basically stock Verizon edition of Android.
After that experience I switched to iPhone. However, since then, I've experienced the iPhone iCloud Phone.app crash, which also denied me 911 service (although fortunately I noticed the problem and resolved it before needing 911.)
Totally agree. I'd really love to see Japan's ETWS system rolled out on the US west coast.
I was in Japan earlier this year, and happened to be watching TV when the earthquake warning system cut in. It gave about 5-10 seconds of advance warning before an earthquake hit. Enough time to move out of the way of anything heavy. It's quite impressive.
(The earthquake early warning cuts in at 0:15, and a tsunami warning automatically cuts in at 2:34... though most of the tsunami warning is cut off at the end. This is what the full earthquake+tsunami warning sequence looks like: https://www.youtube.com/watch?v=qMNu8Y1jIlU)
Yes I experienced it a number of times while I've been in Japan as well, it is impressive and certainly helpful. It would be a great system to have in any area that experiences a lot of earthquakes, or other, short notice emergencies.
It's a balance of efficiencies. On the one hand, false "1"'s can be generated by wind, etc, on the line. "111" calls can butt-dialled by sagging trees. With the rotary, "9"'s are nice and distinct.
On the other side of the balance, users perceived "9" as taking aaaaages, partly because they're bloody users and never satisfied, and partly because in emergencies perception of time gets tweaked. Anyway, 999 it was, and remains.
> The 9-9-9 format was chosen based on the 'button A' and 'button B' design of pre-payment coin-operated public payphones in wide use (first introduced in 1925) which could be easily modified to allow free use of the 9 digit on the rotary dial in addition to the 0 digit (then used to call the operator), without allowing free use of numbers involving other digits; other combinations of free call 9 and 0 were later used for more purposes, including multiples of 9 (to access exchanges before STD came into use) as a fail-safe for attempted emergency calls, e.g. 9 or 99, reaching at least an operator.
I assume rotary dial phones are now all but extinct. I used one about 20 years ago (to win free tickets from a radio station, and succeeded!) but even then they were rare.
I think you're being a bit overdramatic, or there's a difference between your configuration and mine (FireFox Desktop). I find it quite readable, pleasantly so even.
Which is awesome if you have a monitor with 1000 nits or more, as (0,0,0) to (255,255,255) has a blinding contrast on them.
The basic issue is that CSS colors are defined in terms of relative brightness of the monitor, and not within of a proper colorspace. Most browsers assume sRGB, but many users don’t even have that.
In sRGB, a contrast of (102,102,102) to (255,255,255) is a very pleasant, and perfectly readable contrast.
But many users are on shitty 6-bit color depth 100 nits or less displays.
So you intentionally have your monitor set uncomfortably bright, but it's everyone else's fault? And the solution is for everyone else to spend money to make your site look better? How about no.
It may be a shitty option but at least you have the option. Unless you want your site to only be legible to photo editors with some money, you need it to have lots of contrast.
> Unless you want your site to only be legible to photo editors with some money, you need it to have lots of contrast.
Or we simply fix browsers.
Because if it’s a choice between blinding users on most mobile devices, and most higher end desktop computers, or blinding those who have cheap hardware, most developers will choose by who is actually willing to pay.
Toll offices, which handled long distance calls, were computerized by then. They could be upgraded to deal with 911 routing. This requires a database to map phone numbers to the appropriate Public Safety Answering Point. The idea of having a computer with a disk drive in the call chain was radical at the time, but it was made to work. The original computers for this were Western Electric 3B machines, set up as duplex pairs.
The toll office 911 system became a problem with the breakup of AT&T. AT&T retained most of the toll switches, while the local exchange carriers got the local offices. (Now, of course, with weak antitrust enforcement, they've all merged back together again.)