Page 1 of 1

1993: "Fatal Crash Rate"

Posted: Mon May 14, 2018 1:34 pm UTC
by netsplit
Image


Title Text: Fixating on this seems unhealty. But in general, the more likely I think a crash is, the less likely one becomes, which is a strange kind of reverse placebo effect.


There's actually a point where worrying about crashes starts to dramatically increase the chances of crashes, and that's when you get so scared of driving you're too nerved up to up function while driving.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 2:55 pm UTC
by DavidSh
I suppose, while it is yet unknown how possible it will be to modify self-driving cars to serve as unoccupied tools of murder, the number of fatal car-car crashes a car can be involved in is small, so the impact on the fatal crash rate (of such use) would be negligible.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 3:10 pm UTC
by speising
Apparently, fire trucks are too inconspicuous for Teslas.
We've still got some ways to go before we reach that point. Somewhere around 2027, where Randall seems to put it (including the "widely adopted" bit!), looks rather optimistic to me.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 3:11 pm UTC
by cellocgw
"Leave the driving to us" <-- Greyhound crashes are incredibly rare. Not so useful when you want to go down the road to the chemist, I suppose. But in all seriousness, local Assisted Living places have certified drivers to take residents shopping, movies, events, etc. Beats letting your aging parent drive him/herself.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 3:14 pm UTC
by cellocgw
speising wrote:Apparently, fire trucks are too inconspicuous for Teslas.
We've still got some ways to go before we reach that point. Somewhere around 2027, where Randall seems to put it (including the "widely adopted" bit!), looks rather optimistic to me.


Please don't rabble-rouse. We don't know yet whether any auto-pilot or traffic-aware cruise control (TACC) was enabled, or whether AEB was enabled or even available on this car. We don't even know whether it was a pre-2016 MobilEye Model S or a newer Tesla-system sensor model.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 3:23 pm UTC
by Velo Steve
"But in general, the more likely I think a crash is, the less likely one becomes"

Until you start hitting the brakes at every unexpected motion, or driving so slowly that people are dodging around you, at which point the risk goes way up.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 3:29 pm UTC
by Jorpho
It's been difficult to miss the reports of supposed self-driving cars making fatal errors, but have there been many tales of privately-owned electric car cells going up in a spectacular conflagration?

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 3:32 pm UTC
by jozwa

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 4:23 pm UTC
by sonar1313
Velo Steve wrote:"But in general, the more likely I think a crash is, the less likely one becomes"

Until you start hitting the brakes at every unexpected motion, or driving so slowly that people are dodging around you, at which point the risk goes way up.

For everyone else much more than for you, which is a pretty perverse incentive to drive like a fraidy-cat.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 5:00 pm UTC
by commodorejohn
Jorpho wrote:It's been difficult to miss the reports of supposed self-driving cars making fatal errors, but have there been many tales of privately-owned electric car cells going up in a spectacular conflagration?

Are we talking spontaneous, or just as a hazard of the design? I recall there was an accident with an electric car around these parts a few years back where the battery caught fire and torched the vehicle, but if I'm remembering correctly it was a situation where it'd be hard to argue that the results would've been any less fatal with a conventional car (head-on collision driving down the wrong side of the freeway or something like that.)

Anyway, I move we rename xkcd to Randall Doesn't Like Cars: The Comic.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 5:38 pm UTC
by qvxb
Has Randall acquired high levels of feline estrogenic compounds?

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 6:18 pm UTC
by Velo Steve
qvxb wrote:Has Randall acquired high levels of feline estrogenic compounds?


He does seem a little over-worried. Maybe it's generational. When I was 15, everyone wanted to get their hands on a car as quickly as possible, and with as big an engine as possible. Death was for other people. Now lots of young people don't want to bother with driving, and they see it as an unwanted responsibility. Or maybe just an interruption (I hope) to their texting.

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 7:03 pm UTC
by Velo Steve
I wonder if this idea has any merit, or if there's any data to quantify it.

I have the impression that for every auto accident which happens there are many more averted when an alert driver avoids the crash nearly caused by another. You could even say that most accidents take two mistakes, not one. I certainly believe that I have dodged more accidents than I have been in, not because I am an exceptional driver, but because that's just what you do.

Assuming most accidents are started by driver error and some fraction of those are averted by alert drivers, there would be a 2nd power relationship between bad drivers and accidents. Where am I going with this? If automated cars are good at avoiding accidents started by others, and if they start relatively few potential accidents, even a modest percentage of self-drivers could leave us ahead, even if they engage in occasional firetruck-seeking behavior.

By "ahead" I mean in terms of safety. There's no replacing the rush of hurtling down a curvy mountain road knowing that one human error means death!

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 8:31 pm UTC
by Soupspoon
Just earlier this evening, a car rounded a brow-top corner (towards me) with its offside wheels (or the front one, at least) fully over the centreline markings. On a road not narrow enough between the raised side-verges to make this necessary, but nor so wide that I could habitually hug the nearside in the expectations of leaving room enough for the offside mirror of such a vehicle would pass harmlessly by my counterpart one.

To my passenger, the complaint was that I was a bit close to the grass, as I took a more vergeward (but smooth) line around the corner than I might have otherwise taken at a comparatively gentle 40mph on on a default-60mph road, but I knew that I was closer to exchanging plastic shards and silvered glass (at a combined closing speed closer to 100mph though likely no more) than I was to muddying my opposite-side rims and wheel-arches.

Not that I couldn't be sure that, accepting the terms of this partial game of Chicken, my counterpart would not have themselves moved back across the median (or at least roughly averaged) line had I not myself pulled a goodly fraction of a foot off track, but one can't always be sure in the fractional time it takes to do this, so I took the instinctive option and it we both passed, like strangers in the night late-afternoon/early-evening, probably(/hopefully?) never to meet again.

But two aggressive drivers (my replacement straying towards the line, slightly away from the apex, without much tolerance for hugging the inside so close) might have had a problem. A few inches amd it's mirrors, a few more and there's potentially some bodywork scraping and beyond that you get into damaged-wings and perhaps some torsion as the ensemble comes to a halt atop the brow of a hill and blocking the popular (if post-commute) subrural route between townships and some particularly unlucky mechanical failures during the contact (or enough injurious force to either set of occupants) could make the road entirely blocked whilst one or more type of flashing-lighted attendees (blue, if not merely the amber of a recovery truck) resolves the issue accordingly.


This did not happen, though. And there are other fateful things which did not occur (a cyclist on the road, perhaps, closed in upon by me at ~30mph or the other guy at up to 50mph, though I like to think my own care and observation was enough to have retarded my 'reckless' speed for anybody other than a truly invisible cyclist, with or without invisible bike) and one day they might. The odds do get shorter as the timespan is extended…

Re: 1993: Fatal Crash Rate

Posted: Mon May 14, 2018 11:23 pm UTC
by Mikeski
commodorejohn wrote:Anyway, I move we rename xkcd to Randall Doesn't Like Cars: The Comic.


I was wondering the opposite: how many hundreds of thousands has he recently invested in stock in Uber/Waymo/Tesla/etc, planning to hit it big when self-driving cars take over?

Re: 1993: Fatal Crash Rate

Posted: Tue May 15, 2018 3:44 am UTC
by gmalivuk
Velo Steve wrote:Or maybe just an interruption (I hope) to their texting.


This study wrote:Adults are just as likely as teens to have texted while driving and are substantially more likely to have talked on the phone while driving.

Re: 1993: Fatal Crash Rate

Posted: Tue May 15, 2018 4:50 am UTC
by Velo Steve
gmalivuk wrote:
Velo Steve wrote:Or maybe just an interruption (I hope) to their texting.


This study wrote:Adults are just as likely as teens to have texted while driving and are substantially more likely to have talked on the phone while driving.


Perhaps an unwarranted jab at teens on my part, but I stand by my observation that driving is less appealing to young people than it once was.

Re: 1993: Fatal Crash Rate

Posted: Tue May 15, 2018 6:00 am UTC
by pareidolon
Hmm, shouldn't lifetime probability be the sum of the probability for each year, and therefore necessarily increase with age? Unless, obviously, you have a clone waiting to receive your mental imprint upon death and continue in your place.

Re: 1993: Fatal Crash Rate

Posted: Tue May 15, 2018 6:18 am UTC
by Soupspoon
Or the sum probability for each year remaining, thus decreasing?

(Not entirely sure it applies to this, as being in a fatal crash is not itself a guarantee of personal fatality depending upon the exact definition used for the former, just a thing I was wondering about with this life exlectency calculator, as previous ones have suffered from "you should already be dead" miscalculations, whereas only proven survivors (for maybe not much longer more, or many more years yet) are going to be putting details in, so should always have a future of some positive amount.)

Re: 1993: Fatal Crash Rate

Posted: Tue May 15, 2018 11:55 am UTC
by gmalivuk
Why should Randall sum the probability at all in either direction?

It's a plot of the probability of being in a crash that year. If you want cumulative past or future probabilities, you can add them up yourself.

Soupspoon wrote:previous ones have suffered from "you should already be dead" miscalculations, whereas only proven survivors (for maybe not much longer more, or many more years yet) are going to be putting details in, so should always have a future of some positive amount.)

Were those miscalculations, or misinterpretations of what the number was telling you?

Your life expectancy at birth might indeed have been shorter than your current age now, with no miscalculations at all.

Re: 1993: Fatal Crash Rate

Posted: Tue May 15, 2018 12:37 pm UTC
by Soupspoon
(Don't know if was your modly powers or just coincidence of rapid write/read/re-write cycle, but I read your unaugmented message. Then I got an 'orphan' "Notifications [1]" not reflected by any new handy access via New Posts list. Never mind, I knew I could find it.)

I haven't tried extreme values in this latest BBC one (only myself and relatives I know ages of, all 'reassuringly' alive for at leaat a half dozen more years, with "good health" for most of that¹) but other similar-looking life calculators that ask "age now" rather than "year born" have given results obviously including pre-mortality. And even a "year born" version that uses thr alternate input in its aim to give "when you(/they) will die" advice/estimates/²/guesses/speculation is clearly indicating that it is shaving off the deceased cohorts from the actuarial tables.

If it were a missing persons' cold-case database that was trying to determine the likelihood that a historically missing person is still potentially presently discoverable, then (with additional risks associated with their missing status, not included with this one) expectation of negative forward-lifespans might be a deliberate choice. But 'for fun' (without deliberate attempt to frighten) checks that falsely allow negative-remaining results also give wrongly shortened positive-remaining ones the other side of the line where a morbidly waggish programmer might even create a spaecial page to handle the most special case that simply says "You will die in 3… 2… 1… My condolences to your relatives."


¹ In one case I know, this is erring on optomistic. But nowhere are we asked the current health of the person we're giving the current age for. Specifting a current age does rather indicate non-morbidity, though.

² Refusing to use the non-word "guesstimate"in this list, for obvious reasons.

Re: 1993: Fatal Crash Rate

Posted: Tue May 15, 2018 5:01 pm UTC
by gmalivuk
Soupspoon wrote:I haven't tried extreme values in this latest BBC one
You don't need to try extreme values, or for that matter any values. The first paragraph of text after the input box says
Why does life expectancy change based on your age?

Life expectancy is the number of years on average a person is expected to live based on their age, gender and country. The Global Burden of Disease calculates life expectancy by using a country's mortality rates across age groups.

Life expectancy may vary for people of different ages because it is calculated as the number of years a person is expected to live given they have already reached a certain age.

For example, a girl born in 2016 in Mexico is expected to live to age 79, however the life expectancy of a 65-year-old woman in Mexico in 2016 is 84. Her life expectancy is higher because she has already reached 65 and is therefore more likely to live another 20 years.

Re: 1993: Fatal Crash Rate

Posted: Tue May 15, 2018 5:48 pm UTC
by Soupspoon
Tense error on my part, ironically. I had wondered about the approach when I first saw the link (given the issues as often noted with various other 'estimators'). I then did try (real data) ages and skimmed the results before paying much more attention to the blurb, and was already reassured by the more pessimisticly-expected one not obviously revealing the downward bias I had been half-expecting.

Then I decided I was being morbid and left it alone, until the very next day when in this very thread there was mention of integrating in historic prospects made me think again about the integrating in of future prospects. (Which, as you rightly put "you can do yourself" with the comic data trend, and as I'd deliberately left that uncommented because "I agree" isn't very constructive, I might as well officially note that I agree while I'm back here being unconstructive about everything else. Put it to rest, as it were.)

Re: 1993: Fatal Crash Rate

Posted: Wed May 16, 2018 12:42 am UTC
by Steve the Pocket
Velo Steve wrote:"But in general, the more likely I think a crash is, the less likely one becomes"

Until you start hitting the brakes at every unexpected motion, or driving so slowly that people are dodging around you, at which point the risk goes way up.

The latter being the main reason old people tend to be unsafe drivers, isn't it?

Re: 1993: Fatal Crash Rate

Posted: Wed May 16, 2018 4:38 am UTC
by CharonPDX
My wife had a bad car accident ~20 years ago. She still has some anxiety driving. If she's a passenger, and isn't 100% paying attention to the road (playing a game on her phone, for example,) then looks up, I'd say there's a good 50% chance she'll just scream out suddenly, thinking that a parked car/car in the next lane/oncoming car/something random is about to hit us/we're about to hit. And if that moving thing/thing we're moving toward is actually close to us, that jumps up to 90%.

Those screams, I would say, increase my likelihood of crashing while driving at least 10-fold. It's "every time you do that, it doesn't prevent a crash, it makes a crash far more likely." I know she can't help it, and if I see out my peripheral vision that a car is approaching from the side, and that my wife's head is starting to move up from looking at her phone, I'm ready for the panic-attack, so it doesn't startle me, but it happens often enough that it gets MY adrenaline pumping, and my eyes looking all over to see if there's an accident-about-to-be that I didn't see. I'm expecting that some day when that happens, I'll end up rear-ending the car in front of me because I'm looking for an accident coming from the side when she does that.

Re: 1993: Fatal Crash Rate

Posted: Fri May 18, 2018 9:06 am UTC
by OP Tipping
If it is any consolation, your chances of dying in a car crash is very low, regardless of what happens with driverless cars.

Re: 1993: Fatal Crash Rate

Posted: Sat May 19, 2018 12:17 pm UTC
by PinkShinyRose
gmalivuk wrote:Why should Randall sum the probability at all in either direction?

It's a plot of the probability of being in a crash that year. If you want cumulative past or future probabilities, you can add them up yourself.

I think you're right about what it is, but the title of the graph suggests estimated lifetime probability in relation to time, instead of estimated probability in relation to (life)time. You can't really blame people for the confusion.

But really it's a webcomic, there is nothing Randall should or shouldn't do in it.