Hacker News Reader

Comment by polishdude20 | Hacker News Reader

How do you differentiate an earthwuake from common behaviors while the phone is in a pocket such as running, jumping, walking etc.

View on HN

You compare it with other phones in the nearby area.

Right. At any time, most phones will be lying on something like a table. If they all sense something, it's a big external event.

So you compare each and every devices state every X seconds to determine if somethings that statistically never happens is happening?

I can't be the only thinking this is nonsense. Just install a bunch of actual earthquake detectors and make them send notifications via already existing alert systems.

I'm just saying it can be done. Not that it should be the new gold standard.

A less chatty implementation would be if each phone only contacted the server when it sensed a possible quake. If it gets a few correlated hits, it can even send out a "hey did everyone feel that?" question.

Doesn't really seem that difficult. The patterns of picking up a phone, running with it and having it just lay on a vibrating surface are very different. It would be particularly easy to spot panic during a usual running routine.

More, I'd imagine you could look for a signal common to many devices. Unless you are picking up all phones in your household with the same signal, any signal common to them all is likely more global. Could detect if folks are in the same car?

> Doesn't really seem that difficult

Companies lose billions on projects every year because someone uttered these exact same words.

Reply on news.ycombinator.com