Google failed to warn 10M of Turkey earthquake

https://www.bbc.com/news/articles/c77v2kx304go

by piotrkaminski • 12 months ago

41 25 12 months ago

drdec• 12 months ago

Can't say I like the tone of this article. Google is trying to do something good it is not in any way obligated to do. Painting this like they were responsible just encourages people not to try such things.

esafak• 12 months ago

I agree. At least they improved the algorithm afterwards.

xoa• 12 months ago

Yes, the article does sort of half-heartedly acknowledge that, though via Google stating the obvious vs any journalism on the BBC's part:

"Google says the system is supposed to be supplementary and is not a replacement for national systems."

Before immediately going into the classic "but SOME people say otherwise!"

That said, I think the more basic problem I have with this article in terms of education of the reading public is the failure to touch on the false-postive vs false-negative tradeoff that is so core to tuning warning algorithms. It's easy to cherry pick with 20/20 hindsight after the fact and it feels clear when a warning that might have saved lives isn't sent. But in reality, constant warnings can be as bad as no warning because they have the same basic effect of people "not getting the message". The mechanism (people tuning warnings out, boy-who-cried-wolf syndrome) of course differs, but the end result is the same. And outcomes should be the focus when it comes to public health and security. I think this is a super important thing for reporting to acknowledge, because without at least a somewhat educated public understanding there is a systemic incentive by a lot of governments and organizations to lean into false-positives due to being better at ass covering and blame reversal. They can say "well, we warned you, and if we were wrong other times that was just being cautious, and it's YOUR fault that YOU turned off the warnings!" If Google had set off the alert for that quake, but had also set off the alert for lots of smaller quakes previously, even if the end result was the same number of deaths I suspect this article would not have been written. Or, as you say, simply not trying in the first place and keeping their virtual heads down.

Walking the fine line of balance in mass scale public safety is just very, very hard. And to achieve a systemic approach to safety that really maximizes good outcomes takes a good faith, no fault sort of environment (short of actual malice) vs finger pointing. We can see that play out in areas of life like commercial aviation which has been incredibly successful at reducing fatal accidents over the decades.

protocolture• 12 months ago

Big problem IMHO with peoples expectations.

Like the internet is the merest speck on Human history and yet it is considered so ridiculously critical that people will make death threats when they have minor issues accessing it.

manquer• 12 months ago

Google(and others) marketed these alerts as features of their product, as one of the reasons to upgrade a smartphone from a feature phone.

Median price of phone went from couple of hundred in Nokia era to a thousand or more today, justified by all the features including this one , meaning they certainly made money of providing all these value added services.

It’s no different from a fire or burglar alarm . Yes they are not substitutes for the police and fire departments .

However the alarm company cannot turn around and say you shouldn’t depend on us. The government shouldn’t depend/expect them to do anything for emergencies yes but buyers of their product can.

IF a company sold a product and advertised features for it, then it is a problem when said features don’t work especially when said features are designed around emergencies.

ayhanfuat• 12 months ago

It is actually worse than that. The product manager of this feature claimed that the system worked on both earthquakes and when he was asked why people say they didn’t get this alert he basically said “people probably didn’t notice the alerts due to the severity of the event” (https://youtu.be/z-KjVQJ7XKE?si=xwHuholLJV9xjBov). I know Google would be pretty down the list of those responsible for the effects of these earthquakes but they did misinform the public.

readthenotes1• 12 months ago

I had no idea.

https://blog.google/products/android/earthquake-detection-an...

Completely changed my perspective

iqandjoke• 12 months ago

True. That's why Apple does not attempt to have earthquake detection on China. https://iphonewired.com/news/684225/

jamesliudotcc• 12 months ago

It's worse than that.

After the earthquake Google's researchers changed the algorithm, and simulated the first earthquake again.
This time, the system generated 10 million Take Action alerts to those at most risk – and a further 67 million Be Aware alerts to those living further away from the epicentre.

So, basically, Google learned from the data, improved the algorithm, and they are blamed for not having done that before the earthquake.

The headline makes it sound like Google had the good model before, and something went wrong in the notification system so nobody got the alert.

cbhl• 12 months ago

For context, early 2023 would put the earthquake in question just after the first big layoffs at Google (see their 2023Q2 Earnings Release). And this would have probably been considered "legacy google assistant" stuff, not "new-top-priority AI" stuff.

gedy• 12 months ago

Yes, sounds like something politicians would say to deflect blame from themselves or their government.

jxf• 12 months ago

If anything, this highlights the need for continued investment in emergency infrastructure at the nation-state level.

Countries and the general public shouldn't be relying on private technology firms to provide this service, nor should they be defunding the public agencies that provide it.

freeopinion• 12 months ago

Google was able to warn 500 people 30 seconds before the earthquake?

That's amazing. How incredibly encouraging for the future.

ChrisArchitect• 12 months ago

Android Earthquake Alerts: A global system for early warning

https://news.ycombinator.com/item?id=44651092

sandspar• 12 months ago

There's an old blog article about "PETA problems", maybe? Can't remember the name. The premise is that if you mention a problem or help in a small way, then people blame you for not fixing it completely. Google saves people from earthquakes... BUT NOT EVERYONE!!! How could they???

jihadjihad• 12 months ago

Feb 2023 was a while ago, and maybe I just didn’t pay attention at the time to the aftermath, even though I knew it was bad.

But I wouldn’t have guessed that over 50,000 people were killed by that earthquake.

anon191928• 12 months ago

real number is way above 100K btw

jamesxoa• 12 months ago

No good deed goes unpunished. The challenging sensor data quality (random handheld devices from various vendors placed in random environments) means that the detection algorithm can only be calibrated with real data, aka real earthquakes.

pr337h4m• 12 months ago

Another example of the Copenhagen interpretation of ethics: https://laneless.substack.com/p/the-copenhagen-interpretatio...

Lammy• 12 months ago

As a user, is it possible to totally opt out of this system? I don't want my phone notifying Google about when I move it, at all, ever.

rkagerer• 12 months ago

I don't like this part:

"...sets off a loud alarm on a user's phone - overriding a Do Not Disturb setting - and covering their screen"

I'm already at war with my smartphone to make all the alerts stop buzzing and squelch the intrusive notifications (e.g. apps that think they're important, amber alerts that seem in most cases to be accidentally blanketed due to user-error, test alerts, false alarms, etc).

I'm fine if Google wants to experiment with it, and I even applaud the effort, but at the end of the day I want control of my device, thank you very much. That means this sort of thing should be opt-in, not foisted upon on me.

t4h4• 12 months ago

Science article: https://www.science.org/doi/10.1126/science.ads4779

IMO this is the most interesting part: Postevent analysis revealed several limitations to the detection algorithms that have since been improved. First, the duration of monitoring has increased. At the time of the event, the algorithm only allowed updates to earthquake parameters for 10 s after first detection. The number of allowed updates was limited because there was a trade-off between more updates providing additional information for larger earthquakes and more updates introducing outlying single-epic data causing a large overalert. We now allow updates for 30 s and use other checks on the rate of variation in earthquake parameters before updating an alert. Second, there were a large number of noisy phones in the monitoring pool at the time of the Türkiye earthquakes. These high-noise phones triggered late, particularly after the P wave for the M 7.8 event, which had a slow start and complex rupture (31–34). The AEA system is now more selective about which phones are included in the monitoring pool. Individual phones determine their noise level when they become available for monitoring, and this noise level is factored into the detection algorithm. Third, many phones were receiving a BeAware alert and vibrating, which prevented them from triggering on the earthquake ground motion. The alerts now issued by Android EEW no longer cause phones that are detecting to vibrate.

Khoomei• 12 months ago

Android was successful in detecting and warning during the 23rd April 2025 Istanbul earthquake. Probably they made some fine-tuning on their algorithm.

iJohnDoe• 12 months ago

Why doesn’t Google invest in existing solutions?

Why doesn’t Google invest in relay communications that doesn’t rely on a central server, like some on HN have done? One developer here had a really good implementation.

Not sure if Google is reinventing the wheel here?

Google failed to warn 10M of Turkey earthquake

Related Stories