Hacker News new | ask | show | jobs
by bmitc 1759 days ago
I have personally never liked being blankly told that flying is safer than driving, because I’d like to understand how safe flying is versus driving when something goes wrong.
2 comments

Two seconds of inattention in a car is a lot more likely to kill you than it is in a plane. It’s fairly constant risk - going off the shoulder is just a few dozen feet at most.

They also tend to be less well maintained, and the other participants less trained.

That makes sense at first glance, but it's something I'm in control of. I was even thinking of this before, because the comparison is a little mismatched. I'm 100% likely to be a passenger when I'm in any plane, whereas I'm probably >98% likely to be a driver when I'm in a car.

I'm not sure why I was downvoted, because I'm just stating I'd like to understand (data/studies) instead of people's anecdotes and opinions, including my own.

My intuition is that car travel has a constant level of risk, whereas plane travel has a level of risk much below that when everything is good but (I'm guessing here) that risk goes much higher than the constant risk of car travel when something goes wrong (failure, human mistake, crash, birds, weather, etc.). I've searched for it before and didn't find anything, but it'd be nice to see any studies that confirm or deny this.

Considering that ~half of collisions involve multiple vehicles, I suspect your feeling of being in control is a bit of an illusion.
My father always said that accidents tend to happen when two people make mistakes. Even when fault technically lies with a single party, it's often the case that both parties had a chance to prevent the mistake from becoming a collision.

For example, if a turning vehicle cuts off oncoming traffic, it could get hit. An attentive driver in the oncoming lane might see the problem and hit the brakes, while an inattentive or speeding driver might not react in time. The turning vehicle is at fault, but that doesn't mean they were the only one that could have prevented the collision.

There are sometimes events you cannot control, but a lot of those multi-vehicle collisions were preventable by either party.

When you are in a plane there is a 100% probability that the driver will be a trained professional, while according to your stats there is less than a 2% chance that your driver will be a trained professional when you are in a car.

While I have no hard stats I would also assume that a plane is much safer when something goes wrong. Why?

1) Professional 'driver' who has trained for years to handle risks and failure scenarios and practices these on a regular basis.

2) Redundancy and graceful failure is designed into many of the components and systems on a plane, while cars tend towards the cheapest component or system possible unless mandated by law.

3) Space (both altitude and the fact that the sky is basically empty) provides time to solve some problems before they become catastrophic failures and makes other classes of problems very unlikely.

If you blow a tire, run into a deer, or have a transmission failure while you are cruising down the highway at 80mph you have seconds (at most) to react and respond. If a plane hits birds, has an engine fail, or loses some other system they usually have minutes to troubleshoot the problem and redundant systems that prevent the plane from just falling out of the sky.

>If you blow a tire, run into a deer, or have a transmission failure while you are cruising down the highway at 80mph you have seconds (at most) to react and respond

When they happen all you do it move over and come to a stop. Unless you go Full Redditor(TM) and start adding a bunch of extreme control inputs none of these things are that bad. They don't require much reaction at all even if you are paralyzed by indecision it can still turn out fine since all you have to do is stop.

Blowouts are a non-event, IMO, just a flat tire with an extra audio alert.

Automatic transmissions basically don't fail in unsafe ways. Worst case you'll find some extra neutrals and make an ass out of yourself failing to merge or something. Splitting the case on a manual (like you might do if you go for 4th and somehow find 2nd, or jump a vehicle and don't mash the clutch before hitting the ground) is surprisingly uneventful from a maintaining control perspective.

Deer suck, mostly because they tend to break a bunch of expensive cosmetic bits.

Having a hood fly up on you is worse than any of the above events because you can't see. Brake failure kinda sucks too, especially with an automatic trans.

Things “go wrong” fairly “frequently” on airplanes but they are hardened for this with redundant systems and manual (as well as computer) overrides. You don’t hear about most of them but they are all logged and used for improvements.

The main difference is when a car has a catastrophic failure there’s a good chance the people involved survive.

I suppose by wrong I mean something more than the redundant, robust, and various protection systems can handle. For example, on a car, I wouldn't consider the anti-lock brake system kicking in as something going wrong.

I've had an engine failure in my car while driving it. I simply was able to slow to a crawl until I got home. I don't think engine failure on an airplane is such an anti-climatic event, on average.

> I don't think engine failure on an airplane is such an anti-climatic event, on average.

It often is, actually. They have more than one engine precisely for that scenario, and can fly quite well with one down. Flights over water are also carefully planned based on distance to the nearest airport with an engine out.

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

Even if you lose all four on a 747, there’s surprisingly large amounts of time to troubleshoot if you’re at cruise.

https://en.wikipedia.org/wiki/British_Airways_Flight_9

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

Engine failure on a modern airplane is quite anti-climatic. We train for it regularly, and there are standard procedures to follow after the failure. Basically secure the failed engine, if it's just after takeoff climb to a safe altitude (sometimes following a predetermined route) using the other engine, then select a runway for landing and land almost normally.
Sort of interesting how the edge of catastrophic failure has moved out in both cars and commercial aircraft.