Hacker News Reader

Comment by pmoriarty | Hacker News Reader

I'm so disappointed that California doesn't have something like Mexico's Seismic Alert System, which has historically given 60 seconds of warning before an earthquake.[1] Something like that could save so many lives.

"CIRES transmits earthquake early warning alerts through a network of VHF stations similar to the NOAA Weather Radio service in the United States, including use of Specific Area Message Encoding, but with faster lead times in order to issue more timely warnings within two seconds or less. CIRES offers alert systems for buildings and personal use, which also come pre-installed with the official warning sound used for earthquake alerts in Mexico. On its network of transmitters, CIRES issues required weekly tests every three hours to ensure receivers are connected to its network, as well as earthquake warnings when necessary. More than 90,000 users in Mexico City, including almost all public schools, have receivers. The Mexico City Metro additionally receives SASMEX alerts, although not for public dissemination but instead to stop trains or delay departures as necessary."

[1] - https://en.wikipedia.org/wiki/SASMEX

View on HN

It works well for Mexico, but would not be as useful in California.

Earthquake P-waves travel at 5km/s. To get 60 seconds of warning, you must be 300km (186 miles) downstream of the detector.

Mexico has a very unique situation [1]: the filled lake basins that Mexico City sits on have the same resonance frequency as buildings of a certain height. Earthquakes generate that frequency. So a serious earthquake 300km away can get the lake basin rocking, and that will knock down buildings of a certain height.

A detector is nowhere near as useful in California. Consider the Loma Prieta Earthquake. It occurred 90km from the damaged sections of San Francisco. Assuming there is a detector within 10km of Loma Prieta, that is only 16 seconds of warning. Give a few seconds for the alert to get out, and it's barely 10 seconds. That's not enough time to do anything other than hide under a desk -- which you can already do after the earthquake starts. Such a system would be excellent for bringing trains to a halt, but that's pretty much it.

[1] See preface of https://en.wikipedia.org/wiki/1985_Mexico_City_earthquake

Thanks for the explanation. However, it may be nowhere as useful, but still plenty useful. On the top of my head, 10s could be enough so:

* Trains may stop, preventing a literal train wreck.

* People doing delicate tasks (e.g., a surgeon, someone operating heavy machinery, even someone cooking) may lock to a safer position.

* Elevators may stop and open the doors safely, so no one will be stuck inside them.

* Air traffic control may hold take offs and maybe landings.

* All traffic lights going red may slightly reduce car incidents. Seems minor, I admit.

Also, people with little experience (me!) do NOT understand that an earthquake is happening until after the fact. Some blast alert even during the earthquake is very helpful for tourists and even immigrants.

Here's a video of Japan's warning system giving someone 30+ seconds of notice before the 2011 earthquake: https://www.youtube.com/watch?v=n-FMpNBfna8 (and notice how significant the shaking was, even at that distance)

Here's another example of what you can do with ten seconds: https://www.youtube.com/watch?v=gB6-nU2nJqM

30 seconds? you must be joking. I live in Japan and get such alerts on a regular basis. usually you get the alert about 3 to 5 seconds before the shake hits you depending on how far you are from the epicenter. If you get something 10 seconds before it hits then you dont even have to worry because it means you are very far.

Those 10 seconds were staged. She knew that they were coming, she was in a small "room", and she probably took a look at her surrounding before those 10 seconds kicked and planned what to do and in which order in her head.

In real life, the warning such as this would catch people by surprise. That's not to say they wouldn't be useful, but I highly doubt they would be that useful as they were on that video.

Even if I'm in my own home, I imagine it would take me at least a fifth of that time to look around and come up with a plan of what is achievable in such a short time span.

Why do you suppose she props the door open with the chair? Because building shifts could jame it closed via the door frame?

Yes, there have been cases of people getting trapped in rooms/buildings due to doors getting bent or warped, so it’s common advice in Japan to prop open a door to secure yourself an exit.

That's just Triangle of Life nonsense.

https://en.wikipedia.org/wiki/Triangle_of_Life

https://www.snopes.com/fact-check/triangle-of-life/

It isn’t about buildings collapsing, it’s about doors getting stuck, so I’m not sure what relevance the Triangle of Life has here.

Imagine being on a tall ladder, painting, when you get the warning. Or working underneath a car on jackstands.

10 seconds would make a huge difference to some people.

Taking exceptional anecdotes under consideration is hardly useful to save lives. In reality such people should always assume that an earthquake can hit if they are in such areas at risk and use appropriate equipment.

* Closing natural gas and other pipeline feeds to prevent fires.

I hate these "it wouldn't help" arguments regarding most things in this world, including this one. Case in point here, even a siren blasting 0 seconds beforehand would be useful to reduce the inevitable delay between when you feel something shaking and when you realize it's really an earthquake. I'm not sure where you're getting the idea that it wouldn't be useful, because as far as I know, the reason we don't have it is lack of funding, not lack of usefulness.

I did not way "it wouldn't help". At all.

I said that it super-helps Mexico, but that because of the specific situations in both regions, we should not anticipate getting anything like their results.

This is enough time for automated systems to have an orderly shutdown. Trains come to a stop so they don't derail. Toll both gates on bridges go down. Fire station doors open up. Elavators stop on the nearest floor and open their doors.

Given the opportunity, I know I'd rather be ten seconds further from a nearby stack of bricks or possibly-unsecured bookcase when an earthquake hits...

> Such a system would be excellent for bringing trains to a halt, but that's pretty much it.

You could also cut off gas and water lines, so it would probably prevent a lot of damage due fire or flooding.

I would bet the average time to turn off the gas line to a random house would be more than 10 minutes. Probably much more.

Automatic shutoff valves at major lines? There would still be smaller leaks where the minor lines would rupture, but the ruptures would not be continuously fed from the upstream pipes.

You could even have an automatic valve on every line.

Perhaps, but that becomes costly, fast: first on install, but especially on inspection and maintenance.

Not if the valves were automatic.

Isn't that enough time to start to slow BART trains?

They're probably already slow :)

Yes. I specifically included trains in my comment.

The USGS has been working on one, the funding for rollout has been short with Trump zeroing out the line item in his budget requests.

http://www.latimes.com/local/lanow/la-me-ln-earthquake-early...

Here's the website for the project: https://www.shakealert.org/

Shouldn't the insurance companies have a huge incentive to see this built?

I imagine that most of the losses in an earthquake are property losses. These don't seem like they'll be significantly mitigated by a one minute warning.

Perhaps life/health insurance? I'm not sure the average costs of injury associated with earthquakes but it could be substantial

I guess the average person doesn't have very much life insurance (<$100k on average per person). Looks like about 100 people have died from earthquakes in the last 30 years in CA, so maybe that gives you $10M in life insurance expenditures.

Health-wise, let's say every injury has cost $30k on average (IMO a very generous estimate). Looks like there have been about ten thousand injuries or so over the last thirty years. So that's $300M.

Let's say for the sake of argument that an early warning system reduces losses by 50%. So that gives us $150M in value over thirty years. Or about $7.5M per year in value. (Less, really, because value saved in the future has to be discounted.)

That doesn't really go very far even toward maintaining an existing system, to say nothing of developing a new one. (Assuming the last thirty years is representative, that my napkin math is good, etc.)

Interesting points, it made me stumble on an article about how much USG regulations generally spend per life saved.

Spoiler: it varies wildly.

Writeup: https://www.thelifeyoucansave.org/blog/id/93/how-much-are-we...

Citing: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=424523

Here are some illustrative examples:

* Consumer Product Safety Commission's (CPSC's) regulation on childproof lighters = $100k

* Federal Aviation Administration rules on airplane cabin fire protection = $300K

* National Highway Transportation and Safety Administration rules on passive restraints and seat belts = $500K

* CPSC's rules on children's sleepwear flammability = $2.2MM

* EPA's rules on fugitive benzene emissions = $3.7MM

Probably a better way to guesstimate the "value" of a life is to look at the wages that would have been earned if alive. The median income of a worker in the US earns $44k/year, and can probably expect to work another 30 years for retirement (median US age is 37). Multiply those numbers together, and the median death would have earned $1.2M. If you do more rigorous probability distributions, you could probably build a better estimation of what this number would be. Of course, you could also argue for more precise estimates of someone's value than their wages. But the magnitude of the number does appear to gel with regulatory thought: most regulatory agencies seem to do cost-benefit analysis of regulations assuming that saving a life is worth costing at most a few million. $100K per life does seem quite low.

Your estimate is sound, but GP was talking about value (money saved) for health insurance companies.

A hedge fund might have motivation do it, but I don't know how one would reconcile the cross purposes of knowing the info ahead of time for stock trades vs warning the public asap.

One can trade and warn the public simultaneously.

If you're worried about getting claim-jumped by an HFT firm subscribed to the public-warning system, a few millisecond delay might allow for trade execution?

Once you plan for any delay then the ethical/financial discussion then goes to why not 0.1s or 1 second or two? It might be more ethically balanced, and less of a PR nightmare if a hedge fund were satisfied with maybe matched parallel feed and the advantage they get for just having the infrastructure to respond to a richer signal sent at the same time as the public warning. The public warning is a yes or no to send, but the feed might have added data such as uncertainty, or initial locations.. etc.

You mean the ones that don't cover earthquakes?

most people in California don't have earthquake insurance.

During Loma Prieta I got maybe 15 secs of warning ... driving in Berkeley I heard the announcer on KQED (NPR affiliate in SF) say something like "shit earthquake" followed by dead air ... time to stop the car and see it ripple up the street

(normally on that day of the week I would have been on the lower part of the freeway that collapsed, went into work a day early that week because of the possible World Series traffic)

Any alert is better than no alert, but if you're sitting right on top of a fault it won't give you much time to act. I'd be surprised if it's even enough time to get under a desk.

Would any expert reading this like to comment?

> Any alert is better than no alert

For large earthquakes, absolutely. But if you get a situation like the Canterbury earthquakes (https://en.wikipedia.org/wiki/2010_Canterbury_earthquake - 17,000+ over the span of several years) the warning system becomes useless unless it can rapidly detect magnitude and mercalli intensity.

My family lived through the Canterbury earthquakes. If you live somewhere with this hazard PLEASE prepare accordingly.

Depending on your proximity to the epicenter, sometimes you can hear a big earthquake before it hits. It won't help much if it hits in the middle of the night when you're asleep, but a second or two is enough to drop under your desk and avoid the most frequent danger, falling objects.

In countries with decent engineering standards, broken glass and items falling from above such as light fittings and ceiling panels cause the worst injuries. Giving people a minute or so warning could be very helpful to prevent these sorts of injuries.

60s seems unreasonable. Earthquakes move at around 5km/s. 60s warning would be for someone 300km away, who is probably not at risk. You could get 2-10s warning to those nearer the epicenter, which is still enough warning to be valuable. Especially if machinery and self driving cars and such could be auto-safetied.

You’re assuming that an impending earthquake can’t be detected before it’s at full force.

It is my understanding that at the epicenter the time between "start of earthquake" and "full force" is negligible. When the earth slips, it slips. The reason why there feels like there is a ramp-up and delay to full intensity is that some waves travel faster than others, and therefore at moderate distance from the epicenter you start getting one type of wave before the other.

You're assuming that an impending major earthquake can be detected before it happens. I don't think that is possible.

All 'early warning' systems rely on having sensors close to the epicenter that detect the waves before they propagate out to populations further away, but by that time the earthquake is already happening.

I think it should be possible in principle... but I had a professor whose casual academic output basically consisted of publishing rebuttals to other academics who thought they had found an early warning sign (lightning phenomena is a common approach). A particularly amusing one he shared with us was just a case of bad data analysis in Matlab, the author had used 'filter' instead of 'filtfilt' to get their graph and so were trying to derive their results from the distorted phase...

Parent seems to be assuming, rather, that earthquakes take time to reach "full force". I weakly suspect that's not the case, but it's a different question than "before it happens".

From the videos I've seen of the 2011 Japan earthquake, the general reaction generally follows a pattern of "meh, just an earthquake... okay, this is a strong earthquake... HOLY SHIT THIS IS A STRONG EARTHQUAKE." I don't know how much of that is particular to the earthquake, though, and it's hard to tell why that kind of reaction seems to occur.

Small earthquakes are common enough in Japan that many people will just brush them off. My personal reaction to anything 4 or less on the intensity scale is just to stand/sit still. Basically, if it’s not strong enough to trigger the early warning system, I don’t really worry.

Even with the early warning, sometimes the earthquake will still be on the weaker end of the spectrum, so you brace yourself but not necessarily panic.

The 2011 earthquake started off moderately, so no one in my office ducked for cover until it really started to shake, which was already about half a minute in.

I didn't know we had the tech to detect earthquakes with high prob (say >99%) something like 60 seconds in advance. Can someone comment? If we can detect, you are right that it is shocking we don't do population wide drills with messaging infrastructure.

The tech here is P-wave detection. P-waves are somewhere around twice as fast as S-waves, so if you're 100-200km away from the hypocenter, it gives you that much advanced warning. Powerful earthquakes are still destructive at that range.

The other advantage is that once the seismic wave is detected, you can transmit that info at lightspeed vs the 10s of km/h that seismic waves propagate.

If they can text everyone about an Amber Alert dozens of miles away, they can text everyone about an incoming earthquake. No sense how this doesn't exist yet.

Latency. With Amber alerts, a latency of tens of seconds (not unusual for mass text message) is irrelevant; with earthquake, the seismic waves can overtake the messaging queue. (Computation itself might be happening near speed of light, high-level communication protocols in general pas through thousands of buffers)

What would you do with 60 seconds warning? Crawl under a table? I am not facetious. It's probably not enough time to run out of a building.

>Crawl under a table? I am not facetious.

Yes actually (or go into a doorway or similar). I don't understand the derision extremely simple and basic response strategies (like the well designed and effective "duck and cover" for nuclear strikes) get when they can be absolutely enormous lifesavers. A huge amount of the human injury potential from a lot of major classes of disaster is from simple stuff like flying/falling glass and falling dense materials, all of which can do terrible things to an unprotected human body but are also absolutely trivial to deflect with another inert object. Shattered non-safety glass will cause tremendous damage to soft tissues, and someone can certainly be badly injured or killed by even a small weight like a brick or other dense objects that fall from just ceiling or high shelf height.

But none of those is going to go through heavy steel or wood. Typical work desks and tables (and of course institutional use furniture in general) are quite sturdy. Door frames tend to exhibit significant structural reinforcement as well and be away from many short range dangers. Putting an inch or two (4-6 cm) equivalent of hardwood, dense composite wood, or metal between a person and debris can be the work of seconds, and the difference between a brick landing on the table vs on someone's skull is not nothing, as is getting into a position protecting the face and vulnerable extremities and organs as much as feasible and being braced.

It's true that better architectural and infrastructure design matters tremendously and is worth continuing to work on, that if a heavy building pancakes then anyone inside is in trouble/toast, etc. But on a population survival/mass casualty reduction basis a little executed well can make a bigger difference then a lot of people give credit. The difference between death/grievous injury in many mass disasters can come down to relatively small differences in where someone is and what actions they take right before, and how practiced they are in taking the best actions promptly. Don't underestimate it.

60 seconds is better than nothing. Similar to japan, that 60 seconds could mean stopping trains and anything else sensitive to the ground moving. Elevators could be stopped and people directed out. Gas and water shut off. Quite a bit more too, really.

Partial list of things that can happen with even 10 seconds of warning:

* Elevators stop at closest floor and doors open, preventing people from being stuck between floors.

* Critical garage doors & gates at fire stations / hospitals will open, allowing emergency vehicles to quickly exit.

* Other critical doors & gates will close to prevent people from entering tunnels or other areas that could be blocked off.

* Trains will slow or stop to prevent derailments.

* Oil/Gas pipeline pressures will be reduced and valves will close to lessen impact if line is compromised.

* Hospitals alert to prevent mishaps in surgery or other dangerous areas (like MRIs).

* Airports alert to stop takeoffs/landings in case there are buckles in the runway.

In short, there are a ton of things that you can do to prevent really bad outcomes (many of which are prefaced on doing something before the power cuts out) that are much more difficult and expensive after the shaking has started.

Could also post electronic signs on freeways that warn motorists to stop and pull over immediately, particularly before hazardous areas like bridges, tunnels, viaducts, and onramps. IIRC most of the civilian casualties of the 1989 Loma Prieta & 1994 Northridge earthquakes were from viaduct or bridge collapses that pinned cars underneath. With a minute's warning, you can prevent any more cars from entering the bridge and get most of the existing traffic out.

I can almost guarantee you that a warning like that, broadcast at certain times of the day on all major freeways in LA or the bay area, would kill more people than the events that you mentioned.

That actually makes sense for things like utilities and trains systems. Power plants also probably could use that kind of warning.

I hope it won't rain during an earthquake or landslides may happen.

You dont stop a shinkansen at full speed in 60 seconds. Note that in 2011 no shinkansen derailed despite the intensity of the earthquake.

Still, m*v^2; lower speed means dramatically less kinetic energy to go around in case of a derailment.

Surgeons can stop cutting on someone during a surgery. An earthquake is bad timing if a scalpel is close to a major artery.

It saves more time. Warned humans will also skip the costly "what's going on"-realization phase of the human brain.

Get down off a ladder. Stop your car. Pull boiling water off a stove.

I can go all day....

That's ample time to exit homes and small buildings, which are are the structures likely to be most dangerous in a quake.

Wood-framed single story homes are some of the most seismically safe (by virtue of flexibility) structures.

Unreinforced masonry structures, tilt-up concrete buildings, and the like are far more dangerous. California has a lot of legacy tilt-up concrete structures in industrial and commercial zones.

Residences are probably more vulnerable to the resultant gas leak fires than the quake itself.

A framed house is about the safest building you can be in in an earthquake. They are small enough that they just flex with the earthquake.

It's masonry buildings that are the most dangerous.

Either way, you should not be trying to exit a building during an earthquake because the greatest risk is debris falling on you, not the building collapsing on you.

In fact, framed houses are so safe that the current advice is that if you are in bed when an earthquake occurs, you should stay in bed. You are more likely to get hurt trying to get out of bed and find cover.

Depends on construction and earthquake size. Many frame houses have collapsed in an earthquake and the risks increase dramatically with building height.

In the Loma Prieta earthquake some of my co-workers exited building while dodging ceramic tiles coming falling off the roof. Then gathered under the power lines on the far side of the parking lot. Being a modern wood and steel framed building it was better to stay inside.

Also heard stories in the SF financial district of large glass windows popping out and falling into the street below.

Best time to sort out what to do is before the earthquake.

Are you sure that's good advice? It makes common sense but technically the advice is to stay indoors afaik

Would you rather be in an open field of a place with things above you?

The reason to stay indoors is if you're in a city or next to buildings where glass and shrapnel could fall on you. If you have open space outside, then that is a much better place to be.

You generally don't have enough time to get into an open field, even if you aren't in a city. If you happen to be right next to one, odds are you're in a single-story building with few weights in the attic. When these buildings fall down the doorframes are often left standing. Plus if you're running out while the ground is shaking there's a significant chance that the unsupported ceiling will fall on you, the building wall will fall on you, the windows will be flexing and shattering, cutting you...

Generally, standing in an interior doorway is a far safer move.

I feel like you are disagreeing with me but also reiterating my point.

You don't need to be in field in Kansas to be safe outside. If you're in the suburbs you have lawns, parking lots, or streets and there aren't windows above you. Obviously if you're in downtown LA or SF don't go outside. But there's a lot of people that don't live in, or aren't going to be in, an area with multi-story buildings completely surrounding them. You can easily get far enough away from a couple story building (parking lot) and be safe. That is much safer than inside said building.

Here's the rule of thumb "If something can fall on you outside, go to a door frame or braced structure." But I'll reiterate, it depends on your initial location.

It does depend on your initial location. But if your initial location is inside, then when going outside you are necessarily passing through dangerous territory. Because for as long as your exiting your building, you're right next to a building, which is a more dangerous place to be than inside a building under a desk.

In the suburbs, I have a lawn, but there's some big trees. Out on the street, there's power lines. If it's big enough that I should be worried about my house falling down around me, I'm also worried about trees and downed power lines.

It depends on the construction material. If the building is unreinforced masonry, the doorframe is the strongest part of the wall (with respect to earthquakes). If it is not, it is no stronger or weaker than the rest of the wall. (See https://earthquake.usgs.gov/learn/topics/megaqk_facts_fantas...).

The most useful place to be in a major earthquake is anywhere where things are not going to swing into or fall onto you. Such as underneath a desk (assuming it's not a cheap construction that will buckle under any significant weight).

But would you not agree that if available, the most safe space is an open field? That's all I'm saying here. People seem to be missing that I'm not suggesting going outside if things can fall on you outside.

Depends how rickety your house is, and where it's located.

If you can get out easily I don't see a reason to stay indoors. Maybe stay away from trees though.

I can see a reason - protection from falling masonry.

I guess my workplace and home don't have surrounding buildings so it's ok to get out. With buildings around you it's probably less of a good idea.

I think sixty seconds would be more than ample to evacuate most of the buildings I find myself in. In fact, the only building I don't think I could get out of within sixty seconds would be a Walmart or a big box home improvement store.

If you can't escape the building, the usual advice is to find a doorway.

You won’t know exactly how much time you have once you hear the warning. You might be better off finding shelter in the immediate vicinity rather than being caught in a potentially dangerous area 15sec into your 60sec evacuation route.

Oh geez, the aisles with shelves of heavy objects stacked 20 feet high at home improvement stores would be terrifying during a big earthquake. Had never considered that.

That's definitely one of the worst places to be :-)

Are you assuming nobody else is trying to exit at the same time? I can get out of my sixth floor apartment in 60 seconds but not if everyone in the building is trying the same thing.

The crowd factor is primarily why I don’t think I could get out of a large store within sixty seconds, yes.

The tricks discussed in CERT classes (which are well worth taking - very interesting material) is to go to back of store because few people think to. Same with concerts and other events - people instinctively run to the entrance in which they entered, rather than the nearest.

Secondly, if you're in the midst of huge shelves with no escape (like a home depot) get under the first shelf. the shelving won't collapse downwards, but if a neighboring shelf falls sideways you'll be protected.

Harder to do in a real emergency, but good ideas, and thinking them through can be enough to execute them when needed.

Anyway - I highly encourage taking a local CERT class if you're in the US, or finding the equivalent wherever you live.

Maybe when you're the only person leaving the building.

I had to duck a few LCDs and drop ceiling tiles in the 2011 Christchurch earthquake, 60 seconds would've been handy to hide under my desk.

Based on the location of the epicentre, there was no way of getting a 60 sec (or any) warning if you were in Christchurch that day.

It would've been useful in the 2016 Kaikoura quake though - watching the models* of how that fault unzipped up the east coast of the South Island should've given Marlborough and Wellington some warning.

* Here's an example: https://www.youtube.com/watch?v=z1HWet9j070

> Crawl under a table?

That's better than nothing.

> It's probably not enough time to run out of a building.

It could be, if the building is less than 3-4 stories tall.

At my workplace you wouldn't get anywhere within 60 seconds especially with everybody else rushing out at the same time.

Well, if nothing else it's a minute more to decide what to do...

Reply on news.ycombinator.com