Hacker News Reader

Interestingly, this seems to be one of the ways their satellite based emergency services system works - using a relay station to receive messages from phones. This would suggest they already have infrastructure they could use for these features.

I believe there is established precedent in some countries that you can't connect alarm auto dialers straight to the emergency phone number (precisely because of false call-outs), and therefore need to go through an alarm monitoring company. Apple seems to be doing a direct link to emergency services here, in a world where circuit switched telco precedents seem to be missed in the packet switched world.

My guess is they think they can win in the court of public opinion because it's for "people like you" using a phone that you have, rather than something for "other people".

It would be interesting to see how Apple could be forced to implement proper quality assurance and scenario-based negative test validation of their products before they are allowed to do these kinds of things.

> My guess is they think they can win in the court of public opinion because it's for "people like you" using a phone that you have, rather than something for "other people".

This stuff just really upsets me because, if it happens like that, it just shows the corporate capture of the U.S. and Apple's nearly limitless market power.

> It would be interesting to see how Apple could be forced to implement proper quality assurance and scenario-based negative test validation of their products before they are allowed to do these kinds of things.

It seems to me that the bar should be incredibly high, which would mean that it's basically impossible to hook up automated systems to the emergency services line. Because I don't see a way that they could rule out false positives, especially when they're actively pushing their Apple Watch to the active and extreme sports crowds.

Ruling out false positives most definitely should not be the highest priority when it's about saving people's lives.

Note that even the emergency personnel directly involved are not arguing that the feature should not exist or only be allowed if it's perfect.

Because they would much rather respond to five false alarms than be too late for a real one.

What happens when they can only respond to N calls at a time, but have 2N routinely coming in, 2N-1 of which are false positives? They can’t respond to them all. At some point systems like this will unwittingly DOS the 911/first responder situation.

Yes, of course the false positives can become so frequent that it's not tolerable. But that absolutely does not mean that allowing none at all is a sensible goal, because that invariably means more false negatives.

And the article does not sound like the false positives are at the intolerable level, in the opinion of the people actually affected.

But your original comment doesn’t acknowledge the limits on ability to respond. Yes of course in the abstract they would rather respond to all false alarms so as to not miss true alarms, but there’s always a limit. This is why fire departments charge for false alarms. When I worked at my university after graduating, we had a string of false alarms from the data center, and at some point the city allegedly started charging for them. I see no reason not to bill Apple if a jurisdiction has a similar policy already in place.

> But your original comment doesn’t acknowledge the limits on ability to respond.

It also doesn't say that false positives are irrelevant, just that eliminating them should not be the highest priority.

Because the comment I was replying to stated that "the bar should be incredibly high, which would mean that it's basically impossible to hook up automated systems to the emergency services line."

Well, it seems like the problem could be solved by treating “normal” 911 calls normally, and attending to automated calls on a “best effort” basis.

> they would much rather respond to five false alarms than be too late for a real one.

Not in my experience. False calls are a nuisance and inappropriately waste resources.

Many municipalties require yearly application with fee for automated burglar alarms. And fees for excess calls.

https://ci.carson.ca.us/finance/burglaralarm.aspx

> Not in my experience.

Absolutely every article I have seen on this topic has the emergency services people stress that the last thing they want is for people to be afraid of calling them. Because they already get cases where people "don't want to be a bother" and end up dead from a heart attack or permanently disabled from a stroke.

They most definitely do not want to eliminate false positives entirely.

> False calls are a nuisance and inappropriately waste resources.

But as I wrote: trying to eliminate them completely would be far worse. The only real waste are false calls that are done deliberately.

Now this case with the Apple devices is a very unusual one that really needs to have its specific circumstances taken into account. And those specific circumstances are that it does not, in general produce a lot of false positives. But for a specific activity it does, and unfortunately that activity is geographically concentrated so that those false positives are also concentrated. That should be addressed, but it absolutely does not mean the feature does more bad than good overall.

Apple shouldn’t be using emergency services as worker bees to finish an incomplete feature that results in false positives more often than not.

The article directly mentions that none of the calls have been actual emergencies and that it is absolutely tying up resources.

The feature is fine in spirit, and I am sure it has helped in some situations. But I don’t like Apple failing fast. It’s going to end up costing a life because responders respond unenthusiastically to an automated call or the automated calls tying up resources preventing proper response to a real emergency.

Automation that is not accurate is a real problem. Humans will quickly calibrate to it being wrong most of the time. This has occurred time and time again.

> an incomplete feature that results in false positives more often than not.

It doesn't.

> The article directly mentions that none of the calls have been actual emergencies and that it is absolutely tying up resources.

None of the calls in this small town in the middle of multiple skiing resorts. The feature works pretty well in general and has saved multiple lives. It just unfortunately produces an increased number of false alarms for an activity (skiing) that is done very geographically concentrated, and on top of that impedes responses to confirmation calls (because the phones are kept in the pockets of thick clothes).

> Automation that is not accurate is a real problem. Humans will quickly calibrate to it being wrong most of the time. This has occurred time and time again.

Even if they did that, they would do it only in skiing towns, leaving the feature useful elsewhere.

> It just unfortunately produces an increased number of false alarms for an activity (skiing) that is done very geographically concentrated

Skiing isn't the only industry affected by this. I'm involved in the auto racing industry. At a recent conference, there was some intense discussion about false alarms resulting in emergency services showing up at the track and who's responsible for paying for it.

I'm unfamiliar with how Apple's relay works, but Garmin's relay (IERCC) doesn't do much screening. It wouldn't have screened out these calls, as an activation by someone who doesn't reply to texts is treated as a very badly injured person who isn't capable of communication, forwarded directly to the appropriate authorities, and in most cases in the US if conditions permit those authorities will dispatch a rapid response.

There's a good reason for this. In TFA, for example, the authorities are merely asking Apple to properly calibrate the algorithm to skiing. They like the general concept, and have examples of it helping.

One thing I find interesting is that in this kind of thread there are lots of people who don't work in backcountry search and rescue condemning this sort of technology on behalf of search and rescue. I don't have relevant experience, but plenty of people I listen to who do support this kind of technology. They want to get more help to inexperienced people faster because their top priority is helping people and not punishing potential moral failings.

Do any of the Garmin devices self activate? You can configure mine to automatically drop pins on the map every few minutes/hours but it won't signal for an emergency unless you hit the button and confirm on the prompt. The button is behind a rubber stopper to prevent accidental activation, without being completely careless you can't kick it off by accident.

> without being completely careless you can't kick it off by accident.

I thought so too, but apparently accidental activation is a real problem. Here's a nuanced look at the issue by an experienced Canadian SAR volunteer:

https://blog.oplopanax.ca/2017/09/accidental-triggering-devi...

Edit: But you're right, it's not a perfect analogue.

Companies of a certain size get a lot of leeway.

Governments didn't move fast enough to block Uber and other illegal taxi services, partially because these well-funded startups had the means to tie things up in courts for months, while continuing to operate.

As a thought experiment, if someone saw a skiing accident and called 911 then that person would not get in to any sort of trouble if the person who crashed was basically OK.

If I was in a skiing accident I would prefer to live in a world where people erred on the side of calling emergency services when they assumed there was an emergency.

What exactly is the the problem with only dozens of false positives?

There is difference between you calling 911 once because of bad looking incident and calling ir again and again every time you see someone fall.

The difference is human judgement.

I can forgive if once in a blue moon a person mistakes what turns out to be a minor mishap for a severe incident. But when a computer running an inacurate algorithm starts spamming the system, that's unacceptable. If it was a person initiating all those false alarms we would similarly make them stop crying wolf.

It's a lot more than dozens. 71 false alarms in the course of a weekend is a significant strain when you've only got two 911 operators. And that's just one county.

> What exactly is the the problem with only dozens of false positives?

Significantly stricter restrictions for all visitors, regardless of skill level would be one.

Like when people say "Back in my day, you didn't need a safety warning on a hot cup of coffee", we may need to do the same once the local 911 dispatch centre starts recording higher numbers of calls.

I have called 911 as a bystander before, usually noticing something on the road or some other situation when driving on the highway. Although I am not afraid to call 911 to notify them, before calling, I think thoroughly about doing so and evaluate the situation appropriately. The operators also ask questions, like is the object on the road, was there anyone in the car in the ditch, they’ve already been notified, etc. Why does Apple get to play by different rules and operate more recklessly?

From the article, none of the automated calls were an emergency situation. Humans don’t react well to false positives, and over time, those calls will be handled with much less priority or attention.

In the particular situation of skiing, they have their own patrols. I’m not aware of endemic situations where people are injured but not discovered or helped.

My issue is that Apple is using emergency services as the last mile implementation rather than handling it themselves.

Maybe they could geofence ski-resorts and pop up a warning that the feature is disabled.

Maybe apple could give users the option to disable this feature while skiing? Or even create ski mode that behaves differently. (I honestly can’t think of the perimeters but I’m sure apple has software engineers who are more clever by half)

I know this sort of user first mentality is becoming more and more alien to the Apple ideology, but there’s a reason we keep coming back to it.

The user knows what they want and if you think you know better you better be damn sure.

The thought process of apple is that every saved person is worth whatever cost it might have especially if you as a company don't get to carry it. And the topic is fairly hard to discuss given that justification. You only need one single case where the feature was helpful to win the argument.

False alarms may cause someone in real distress to not receive the help they need.

This is exactly the point. Tying up phone dispatchers time is annoying and marginal. Sending limited search and rescue teams off on wild goose chases could have fatal consequences.

Dozens of false alarms seems worth even a single saved life so to actively punish Apple seems ludicrous.

  “We are absolutely diverting essential resources away from people who need it toward a feature on a phone,” Dummer said.

The journalist in question did not do any research towards how large of an effect this is. That quote does nothing but service the claim that this is a “problem” worth discussing, and more importantly, a “problem” worth sharing on commercial social media…

Also, from the article:

Grand County Sheriff Brett Schroetlin called the calls are “problematic and time-consuming,” but said they have not impacted dispatch operations.

Journalists are trained at extracting provocative statements that are conspicuously apt for misinterpretation. The above quote is a perfect example. The original conversation could be 3 minutes about how this is a nothingburger but then the Sheriff being badgered into conceding that in isolation the calls are problematic and time-consuming…

What? How is talking to the dispatchers inundated by false alarms failing to do research? That’s literally the people who would know.

The biggest danger of false alarm is alert fatigue. If your false to real ratio is ten to one you’ve seriously messed up. This sounds like exactly the case where false alarms are causing real harm to people in need.

Your stance shows a serious misunderstanding of incident response.

The article is from a local Colorado news outlet. They're supposed to do a nationwide study before publishing this story?

What’s the trade-off though? If an Apple Watch sends an emergency call every time a skier or snowboarder falls, that’s several calls per day. I would think that would be too much and we should hold Apple liable for that level of false alerts.

It’s an extreme example, but I want to make the argument that there is a balance to consider.

Why not hold the apple user responsible if anyone? Fall alert is an optional feature (that I think might be opt-in during setup). Once it detects a fall, you have a bit of time to respond to the watch to tell you are fine. Although this would obviously be difficult when you are fully bundled up.

But the best place to start would probably be the law enforcement officers working with the nearby ski resorts to educate their patrons about the apple watch feature with signage.

It is on by default.

Is it? I upgraded to a Watch 8 recently and it seems to be off by default (it could be that I opted out a million years ago on my old watch and it carried over).

Though I distinctly remember being asked about it when setting up my mom’s watch a few years ago so maybe it’s based on age?

It was an opt in thing on my watch as well. But it looks like it is on by default on the new iphone 14. Personally I think it should be opt-in during setup.

https://support.apple.com/guide/iphone/manage-crash-detectio...

Everyone who's saying this has clearly never had a job that requires an on call rotation. At a certain point, you start treating these things as non-emergencies. Doesn't matter how much you initially care or want to do a good job, you can't respond to everything.