Why Uber Can Find You but 911 Can’t | Hacker News Reader

Why Uber Can Find You but 911 Can’t

wsj.com

47 pointsrcarmo56 comments

I always wondered why the Phone app didn't have location services on and a big button that when the user presses it would grab the GPS coordinates and reverse geocoded address then have Siri read those over the phone line. That way no 911 center needs to upgrade and no telcos need to change infrastructure.

A simple button right there on the phone that sends the following sound (in best Siri Voice imitation): "This phone is likely located near 1600 Pennsylvania Avenue North West, Washington DC, Latitude North 38° 53.8258', Longitude West 77° 2.1927' admire.admit.hulk" (last words intentional: http://what3words.com/)

Smartphones are nightmarish enough in an actual emergency-- try unlocking your phone and calling 911 when you've got blood all over your hands, or you're actively sinking in a submerged vehicle in a dark and frozen lake. Or if fluids get into the TFT, good luck getting the screen to stay on long enough for you to press anything at all.

There are better ways. Government has the money to arm police with grenade launchers and MRAPs. The telcos have no end of cash from mobile data exploitation.

Nobody is suffering except the people. Spend the money. Fix the goddamn infrastructure.

Sure, I look at my phone then say, “hey Siri, call 911”

I don’t press anything.

Try imagining a solution that works instead of one that doesn’t. Wearables, for example, will be an even better solution:

https://www.google.com/amp/s/h4labs.wordpress.com/2015/07/28...

You don’t need to unlock to call 911. But clearly if the phone isn’t working or you can’t press buttons then it’s a problem.

I have found that touch screens work pretty well when covered in blood.

Using my phone is the last thing I would be doing in a sinking vehicle and preferably outside of the vehicle.

> I have found that touch screens work pretty well when covered in blood.

I haven't. The screen registered keypresses all over the place. It doesn't work reliably with wet hands either, or even in a steamy bathroom.

> Using my phone is the last thing I would be doing in a sinking vehicle and preferably outside of the vehicle.

Every situation is different. Revisit your strategy once the car has already slipped under the ice and started to drift. The cabin is filling with freezing water but you still have some oxygen left. You won't be able to make any calls once you break the window and go underwater, so you better hope you can find/reach the spot where you broke through the ice...in the dark.

I personally knew someone who drowned under these circumstances. (And the saddest part is, even though she accurately reported her location, responders showed up instead where E911 said she was...an entire county away.)

At the very least, you should be able to just call 911 and engage no further. It should be broadcasting your exact GPS coordinates. To the dispatcher, the pinpoint on the map in the middle of a body of water should speak for itself.

ya sure, there's always corner cases where stuff won't work and you need to bring in the specialized equipment. It's hardly an excuse for discarding the benefits the other 99% of use cases that benefit from the tech.

I had never heard of what3words before. It's a super cool and potentially incredibly useful idea.

I wonder how they manage altitude (would a cave underground have the same 3 words as the patch of surface above it?).

I also wonder how this is a monetizable product that warrants a company behind it. It seems like it'd be a perfect fit for open source work.

Thanks for letting me know!

Aside from the fact, already highlighted by other commenters, that this is a proprietary system operated by a company that actively protects their IP, I think it's also a terrible idea.

As you noticed, it does not differentiate based on height, so the ballyhooed deliveries would require two addresses to find an appropriate suite number in an office building, but more importantly, their app is required in real-time to find anything.

Given an address like 123 Main St, I know that I can find Main Street, and then head in one direction or the other to find 123. Given an address that is three arbitrary words tells me absolutely nothing about where that is in relation to me. Zip, zero, zilch. Only an app, or a clumsy proprietary printed map, can help me. Even then, anything other than following arrows on screen is a genuinely ugly experience.

A public, open address of the traditional variety lets one get close, then closer, and zoom in based on the address alone. In the US, for example, you can first read the country, then the state, then the city, then the street, and finally the street number. It's all stored pretty efficiently in a standardized address.

This dumb system loses all of that, but they claim it's better for places that lack standardized addresses. Of course, we already have a system for that: latitude and longitude. The advantage of the public, open lat+long system is that, again, you can generalize and approximate at first, and then add more decimal points when it matters. There is no way to do this with this dumb system.

It's an idea that seems great when you first encounter it, but it's proprietary in a way that hampers its utility, and is also impractical in actual operation.

> potentially incredibly useful idea

Do you really want to depend on a startup's ability to keep their service running and well funded to the location needs of emergency services? Unless they donate the grid and word associations to the public domain, it's a bad idea to use proprietary data sources for emergency services especially since there's nothing wrong with longitude and latitude, and humans really don't need to be able to read it, computers do just fine with it.

Indeed, it's entirely proprietary. Being able to get deliveries to your house shouldn't be dependent on someone paying a royalty to a startup.

Also patented WO2014170646 A1, as such is unusable by everyone except dominoes. Hooray. They also actively takedown similar projects.

Whipped this up in a few hours. While I was writing it I realized it's pretty non-trivial to figure out a nearby address using google maps. Maybe there really is a good use case for this.

https://myaddr.io/

That's a very fascinating idea. Also might that same button turn on all tracking and signal pinging. I wonder if it's been considered but has not been implemented due to mistaken buttons pressed

Seems like a great hackathon idea. You could even build it as a webapp and target all your major platforms at once. I'm going to give this a crack while I'm sitting at the hospital.

A bit off topic, but thanks for the link to what3words It's a great idea. Commenting from near concerts.also.could

Great idea but it's easy to mistake stutters.screening.somehow for stutter.screening.somehow, and it gets worse when you consider words that are pronounced similarly but spelled very differently. If the similar words are indeed spread far apart, and not random, that might work OK, but adding a checkword, or more words with fewer different variants, could help here.

It's also a commercial, closed standard (https://what3words.com/pricing/) which is as good as a death sentence for adoption.

It's an interesting concept! Hello from term.dame.hurt (Auckland).

Here is your answer:

> That way no 911 center needs to upgrade and no telcos need to change infrastructure.

If you ever touched upon how government bureaucracy works you will immediately realize the more taxpayers money spent the "better". Of course Im not for it, I'm just playing devil's advocate.

Just before Christmas I spent an hour in a ditch with a broken hip after cycling into a a sheep (I'm from New Zealand and they crowd all our roads and are a frequent hazard). I was in a rural location so I had to spend quite a bit of time on the phone with the emergency dispatcher, while in a not insignificant amount of pain, trying to describe my location in relation to various landmarks. The dispatcher was in a different part of the country and was unfamiliar with local landmarks so this was very frustrating. I found out later that there is a increased probability of complications (avascular necrosis) if there is a delay having surgery following this type of injury (I'll know in a few month if I need to go back for a hip replacement). Not sure if an extra 10 - 15 minutes it took to get a ambo to me would have helped, but I'd prefer the odds my way where possible! My parents were confused by how I wasn't able to send the dispatcher my exactly location over the phone.

Me after being picked up by ambos -> https://imgur.com/a/HJKWo

http://archive.is/l5uLn

If calling 911 can automatically send your location based on your mobile phone, you'd be able to save a lot of time for the operators, but I wonder how many lives would be saved from faster responses.

Even better, if they created some app that would speed up the 911 process.

As countless other countries have been doing for ages. Even Belgium does it:

https://play.google.com/store/apps/details?id=be.Nextel.Emer...

No need for an app even, I read that (some?) german emergency call centers use a service that will send the caller a text with a link to a website that will then use the browser's geolocation API.

John Oliver covered this topic back in 2016: https://www.youtube.com/watch?v=A-XlyB_QQYs

Also unable to read the article no matter what tricks I use, but I'll add this nugget of info for E911 requirements. I believe we are coming up on the deadline years for when telcos must incorporate barometric measurements from barometer-carrying phones into altitude calculations so that 911 emergency response can get an altitude clue from the moment you ring.

Edit: PDF source https://apps.fcc.gov/edocs_public/attachmatch/FCC-15-9A1.pdf

And edit: This is a tricky situation, because the barometer does not provide real altitude data, and if the math is done incorrectly or there is some interference, the 911 response could be sent to the wrong altitude. Perhaps there are delays with that because extracting the correct altitude from the data is extremely difficult to do with only a single phone to use a source.

https://l.facebook.com/l.php?u=https://www.wsj.com/articles/... will let you read it :)

Good trick, thanks! Reading now. Super interesting.

In Finland we have an official 112 (that’s our 911) app which you can use to make the call and pass location info.

The app will send the phone number and coordinates and then initiate a normal call. Officials can see the calling phone number and connect that to received location.

Not as serious as the article, but I still don't understand why Find My Phone can almost tell me what room in the house each device is, but the (built-in) weather app tells me I'm 2 towns away...

Not sure how far away towns are around you, but around the Bay Area at least, 2 towns don't stretch more than ~5 miles. Weather usually doesn't change that fast so its not a smart power move to thrash your battery querying for the most accurate weather data. Find My Phone, on the other hand, is probably a way more pressing issue and worthy of more resources spent achieving the best accuracy possible.

Just my 2 cents, not a iOS engineer or anything.

There was a tv show called APB about this app/911 idea. http://www.imdb.com/title/tt5542294/

To bypass the paywall, try going to Google and typing in the name of the article ("Why Uber Can Find You but 911 Can’t"). That worked for me.

Oh thank you. They must have check for referer. Because the google's link is exactly the same with HN's link

paywall, so cannot read.

i assume it's because you volunteer your location to uber, whereas you don't to gov't (and something about warrant-less surveillance/wiretapping laws). there should be an easy way built into Android/iOS to transmit your location during a phone call to the other end. it would not just solve the 911 use case but a lot of other non-emergency uses.

EDIT: i'm also very interested why this comment is garnering downvotes.

"Hey folks, didn't read the article! But here are some of my assumptions that the headline reminded me of. Also here's this sweeping thing I think both major mobile platforms should implement for some reason. Now to send the comment and rack up those upvotes!"

I posted an archive.is link that you can use to read it. I only skimmed the article, but, a lot of this sounds like signaling issues.

> transmit your location during a phone call

Transmit how? Send an SMS to 911? Some kind of HTTP/API call? Some kind of encoded audio signal?

Probably using the SS7[1] network. This is mainly how "out of band" signalling is done on telecom networks. The problem is, there are already notorious security issues[2] with SS7 and exposing any more functionality "up the stack" might be deemed to risky.

[1]: https://en.wikipedia.org/wiki/Signalling_System_No._7

[2]: https://www.wired.com/2017/05/fix-ss7-two-factor-authenticat...

Extending SS7 to your phone seems... complex.

The more I think about it, having your phone send an SMS to 911 with your location as soon as you dial 911 seems really simple. SMS to 911, where supported, must already be special-cased. Carriers can each deal with that and communicate that back to the 911 call center in their own way (obviously negotiated with the 911 agencies) and phone manufacturers need only do a single (simple) thing.

There would be a lot of flexibility there too. Your phone could choose to update the location as it changed (moving vehicle) or as precision increased.

Since I wrote the software for a 911 CAD (computer aided dispatch) used by a primary PSAP (public safety access point) -- yes.

But... The 911 operator has to initiate the SMS exchange. Kind of strange. I believe that, in certain jurisdictions, TDD service for the deaf was a revenue stream for the incumbent telcom carrier. So, cell phones are specified to have to be "registered" to allow 911 SMS (this is just my speculation). The 911 operator has to initiate the exchange (in the areas that the system I work on is used).

Incoming 911 data already contains location data (from tower triangulation, etc.). More interesting is that the 911 operator must manually request location updates (secondary reports) and that these cannot be requested more frequently than 35 seconds apart (again, these regulations apply in my area). Again, stinking regulations.

Anyway, the data available (formats) are documented by NENA (National Emergency Number Association). www.nena.org. But... carriers (Bell, Telus here) use supersets of those formats.

The problem is not flexibility -- the systems are in place, and must inter-operate. IP networking for these systems is not completely in place (some use serial communications). The cut-over to IP networking is still on-going, and some places use parallel serial and IP networking.

And, of course, 911 operators get ANI/ALI (automatic number identification, automatic location identification) data for home phones (ALI by database, usually), and the mobile needs to be incorporated in such a way as not to confuse the dispatch operators.

I suppose things have changed a lot since I was a 911 dispatcher 20 some odd years ago (man, where do the years go????). Back then cell-phones were still just short of ubiquitous and we had, at the time, no ability to even receive SMS messages in our PSAP. The idea of using SMS for that didn't even occur to me.

I believe the SMS facility is actually built on top of SS7 anyway though, so that actually makes sense. And it wouldn't mean exposing anything new up to the phone level. Not sure what the state of security is though. How hard is it to spoof an SMS message?

> i'm also very interested why this comment is garnering downvotes.

Your comment about the paywall is explicitly off-topic, and then you've commented without reading the article. And now you're posting about being downvoted, which is likely to attract further downvotes.

People are going to complain about paywalls as long as people keep linking to articles with paywalls, and no amount of rules is going to stop that. Either that needs to be accepted or HN needs to stop allowing paywalled articles to hit the main page.

>>Your comment about the paywall is explicitly off-topic

That seems harsh. If there's a paywalled article posted then it's fair of him to say that he can't access it because it's paywalled. If you have an issue with that then you should have (an issue with the HN rules and whether they should or) shouldn't allow paywalled sites/articles to be posted.

> (this comment was a reponse to the initial comment by chrisseaton which has since been removed.)

I haven't removed any comment at all. The quote you're replying to is still in my comment. And I can't have edited after you posted that comment, because HN doesn't apply comments that have been replied to to be edited.

> I can't have edited after you posted that comment, because HN doesn't apply [allow?] comments that have been replied to to be edited.

Is that true? I'm pretty sure I have edited comments after they have been replied to, as long as it was within the two-hour edit window. But I could be remembering wrong.

Want to try editing your comment now that it has one or more replies and see what happens?

Edit: Testing an edit after this comment has been replied to. Indeed it can be edited.

Well whatever I don't edit my comments after they've been replied to. And it wouldn't make sense to have edited because I've repeated the same information several times in replies to my replies.

I wasn't suggesting that you should edit your comments (except for that one as a test). :-)

Your note about HN not allowing edits got me wondering, so I was curious to verify it one way or another.

Since you replied to my previous comment, I did a test edit on it and indeed editing was allowed. Mystery solved!

Maybe I was thinking of the rules for deletion.

my fault, un-edited.

I think it is a bit harsh.

Most of the initial commenters couldn't read the article due to paywall, that seems very on-topic.

http://archive.is/l5uLn works well to paywall-bypass.

> If you have an issue with that then you should have HN shouldn't allow paywalled sites/articles to be posted.

I'm a bit torn about that. These are good articles, they should be posted/discussed.

You could link to things like the paywall bypass link I sent instead, but, there are infinite variations of those so you could end up with infinite threads about the same article which seems like a poor solution.

I'm not sure you read my comment...we are saying the same thing.

I was trying to agree with you :)

> That seems harsh.

I'm not giving my opinion on it - I'm telling they why they've been down voted, and this is it.

> If there's a paywalled article posted then it's fair of him to say that he can't access it because it's paywalled.

But the HN rules say that it isn't ok for them to post a comment saying that.

> If you have an issue with that then you should have HN shouldn't allow paywalled sites/articles to be posted.

No the HN rules say it's ok to post paywalled articles. They then also say it's not ok to post comments complaining about it.

I think you're confusing that I'm telling you my personal opinion. I'm not - I'm quoting the HN rules.

How is what you're saying any more on topic (the original topic)?

Look I haven't downvoted them!

They asked why they may have been downvoted, and I'm explaining why people may have done that so they know how the site works.

We're not supposed to comment about comment voting at all, but if nobody explained that nobody would know!

Reply on news.ycombinator.com