In the 1950s, the rate of commercial plane crashes caused by human error was 71 percent. By the 1990s, with the introduction of new automated flight systems, the rate had decreased to 52 percent. [0]
Yes, there are horror stories about "dumb" transportation software that linger in the forefront of our memories. But ask yourself, What is the burden of proof for this technology? Do self-driving cars and aircraft automation systems have to work 100 percent of the time before we accept their usefulness?
If that's the burden of proof, we will remain stuck with human drivers who are much more "dumb," by a measure of the car accidents and plane crashes they cause.
The author's argument against automation in these life-saving contexts is weak. His article would have been stronger if it stuck with points about watches and similar consumer products.
The stastic you quoted doesn't actually tell us there are less crashes due to human error, though. We could have just as many crashes due to human error, plus more automation-related ones, and that'd still reduce the percentage of crashes that are due to human errorm
"The author's argument against automation in these life-saving contexts is weak."
I think you missed the point of the article. Its not that existence of automation is bad, its the implementation of it that (in a lot of cases) is wanting, useless, distracting, or even detrimental to the items purpose.
"The notion that we can have automated or semi-automated cars as long as the driver is watching over them is a dangerous myth."
If we accept the author's argument that a human driver can't watch over an automated car, the question becomes fairly black-and-white: do we rule out the driver or rule out the automated vehicle?
Within this rigid self-imposed framework, the author seems to choose Option B: ruling out vehicle automation. Throughout this post (in the aircraft analogy, for instance), he notes the importance of human supervision.
To this, I say, Where is the evidence that human-supervised vehicles are safer than fully automated cars? Where is the evidence that humans are better equipped than good software to react to split-second emergencies?
The only point in favor of human drivers is that we have more stats to describe them, but the stats are horrific. By contrast, Google's autonomous cars have logged over 300,000 miles in various conditions without an accident (minus getting rear-ended by a human driver).
The current fatal crash rate for human drivers is about 1 per 100,000,000 miles, and the injury crash rate is around 1 per 1,200,000 miles. 300,000 miles with 1 accident doesn't prove much.
That's not what the author is saying. Their saying it's bad that the automated system didn't warn the pilot when something went wrong. Or that self-driving cars can't reasonably depend on humans to monitor them at all times. He's not saying that all automation is inherently bad.
Taking one argument for horribly bad designed "smart devices" (the current generation of smart watches) and applying it to a different field altogether (airplanes and cars) is absolute utter bollocks.
First of all, most of all civil aviation disasters in history have been caused by human error[0] (or, in at least 4 incredibly famous examples, human intention). Yes, humans have lost their lives to technology failing, but the assertion that the pilots always are able to do something if they had just been more aware and given control of the dozens of subsystems that a commercial jet controls behind the scenes is rather ridiculous. The "Shit Happens" factor exists for any level of technology.
Or the idea that human beings can react faster than a computer can in the one or two seconds of a car accident. The same car that is able to detect the fact you're trying to leave your lane when you accidentally fall asleep at the wheel and try to wake you up or keep you in lane, the same car that is able to fire airbags in the milliseconds after a major deceleration or impact event. The same car that prevents your brakes from locking up and tries to prevent your car from hydroplaning when you decide to speed over that puddle of water because you're late and speed limits are just a government ploy or some other bogus Libertarian rationale.
People will die driving in self-driving cars. This will happen some time in the future - it's an inevitability. But it is incredibly difficult to believe that figure will be anywhere near the scale of the ~35,000 automotive deaths per year in the United States we average now.
If we take the obviously huge leap to compare air travel's legendary reliability thanks in a big part to autopiloting, equipment redundancy, regular inspections and maintenance, and so forth to self-driving cars, we'd believe the deaths to be on the order of 35 or less a year. Cars may never be that reliable due to other factors (like all of the vehicles being on the same plane of travel), but it's still far more likely to be an overall win.
Yes, progress in technology is a horrible, horrible thing. Let's just put our heads back into the dirt and pretend it isn't happening.
> " ... but the assertion that the pilots always are able to do something if they had just been more aware and given control of the dozens of subsystems that a commercial jet controls behind the scenes is rather ridiculous"
That's not his assertion. His assertion is that the pilots had no idea that anything was wrong (or being compensated for) up until the plane basically said 'fuck it, you deal with it'.
> "Or the idea that human beings can react faster than a computer can in the one or two seconds of a car accident"
Where is this claim being made? Certainly not in the article. As for the rest of your paragraph, those seem like individual, single-purpose systems (for the most part). When you're talking about larger systems where things within the car must communicate with each-other a lot more (probably written by different groups), I expect the scope for bugs is greater.
> "People will die driving in self-driving cars ... it's an inevitability"
I completely agree. The really interesting thing will be who is considered (legally) liable at that point. The manufacturer? The software authors? The driver? It's not really that clear cut and I bet the insurance companies are already thinking about this.
> "Yes, progress in technology is a horrible, horrible thing. Let's just put our heads back into the dirt and pretend it isn't happening."
Where did this negativity/sarcasm come from? The author of the piece isn't claiming progress is horrible so why put these words in his mouth? I read his basic point to be "design better stuff, there's plenty of research you can already draw on". That sounds pretty reasonable to me.
I think the main point illustrates that they are designing self driving systems under the assumption a human can correct for error. That assumption appears to be flawed when looking at other domains. People pay less attention the 10% of time they are needed when they spend 90% of the time when they aren't. You need to have that in mind when designing these systems and they should be as Antifragile as possible.
It may actually be true that drivers can correct for errors now because you've got generations of drivers that have grown up... driving. Take that away from them, and... assuming that a driver can 'correct for error' during the time when it's most likely to be problematic, and when the person hasn't had 20 years of day to day driving experience... it'll be a mess.
Most humans have a hard time to cope in an unusual situation without training even if the situation is just a bit off from the normal. Just think of the new snow in a place which only have snow once a year. Cars in ditches all over the place. So yep. It will be a mess.
>Yes, progress in technology is a horrible, horrible thing. Let's just put our heads back into the dirt and pretend it isn't happening.
Yes, sometimes it is horrible. New technology is not always "just a tool" (like a knife, which you can use to spread butter or kill someone), it frequently has inherent uses and can push towards certain societal outcomes. Not always good ones.
Technology is not some king we have to obey and adopt wholesale. It's something intelligent beings create at their and for their convenience.
>If we take the obviously huge leap to compare air travel's legendary reliability
Not that much reliability over cars, that's just the impression bad statistic give.
If you add the total hours of car driving and compare to the total hours of flying, it more or less comes as a wash for deaths per hours travelling with each kind of vehicle.
A great parable for how we try to make technology more human-friendly with unintended consequences.
I went on a road trip recently and I've been thinking about how advances in car technology change how we view cars. I came to a similar conclusion as Don - specifically the change from manual to automatic gearboxes has meant that people have to concentrate less on driving and can devote their attention to other things.
But that's not necessarily a good thing - by getting rid of the idea that driving is an activity that requires full-time attention we are making it more unsafe because people then feel justified in paying it less attention. The reductio ad absurdum argument is the self-driving cars that he mentions at the end of the article, but this is just the end of the spectrum.
I'm not sure what the solution to this problem is.
So far the solution has been to include safety systems that do not depend on the driver. It's worked incredibly well.
The quip about the laws in the article is misguided, none of the laws so far have been about licensing self driving cars for everyday use, they have been to clarify the legality of putting test vehicles on public roads.
There is also cost cutting. Multiple independent 'smart' systems are ok. But pushing cost down means higher integration to save money. Now when engine dies it may take down power steering or even brakes.
Self-driving cars are great example of cost cutting. It would take $ 100 000 000 000 to develop system which can actually drive. Google is using GPS with fraction of cost. But GPS signal is very weak, easy to block or even spoof. If self driven cars will become reality $20 gadget could collapse traffic in entire city for hours/days.
Google's self-driving car platform is based on sensor fusion - it does use GPS, but if the GPS network goes down, it still has mountains of lidar, photographic (Street View cars aren't just for the heck of it), and even regular old street maps for dead-reckoning navigation.
Your $20 evil genius GPS blocking gadget is far more likely to cause the regular human beings in their cars to lose track of where they are going than the robot that has been designed with this exact constraint taken into consideration.
Furthermore, as of right now and the near and mid-term future, it's not legal for the car to drive itself without a human at the wheel. Which means the self-driving car still has to function as a human-drivable car. The final nail in your mustache twirling GPS-blocking evildoer's coffin.
Before we go to a complete driverless system, it's likely we'll need lots more sensor assistance, such as "active highways" with mile marker and lane-keeping RFIDs paved into the roads, and other similar technologies. But we're probably at least 30 years off from even having that conversation given the conservative climate and the love for personal driving in this country.
I think it is highly likely that the licensing process for self driving systems will require them to be environmentally aware and function without gps signal. What I mean by environmentally aware is that things like passing through an intersection and lane keeping will be fully guided by immediate sensor data, not from reference data looked up by position.
Maps and dead reckoning fall more under navigation than they do driving and that part of the problem is reasonably solved.
I am just saying that GPS is not reliable and should not be used in high-risk cases. There is no "evil mustache twirling", any teenager will do it as a joke.
I direly hope that we get wireless charging to a point that I could just put my phone down anywhere on my desk and it could get charged.
With something like pebble, I wonder how much energy is necessary? Could some well-placed solar cell on the watch do the trick (think calculators).
If I could get one wish in life, it would be to have something like nuclear batteries we always see in sci-fi (actually, there are probably 12 other things that would be more important in the long run)
> Could some well-placed solar cell on the watch do the trick (think calculators).
My Casio watch has two tiny solar panels built into its face. According to the manual, 5 minutes of sunlight each day (or a few hours of indoor lighting) is enough to compensate for normal use, and the battery will last 6 months even without any sunlight.
I've had the Casio for 9 years now. It still works fine, and I've never had to manually charge or replace the battery. The battery does get a bit low in the winter when the watch spends a lot of time covered by thick clothing, but that problem is easily fixed by leaving the watch in the sunlight for a few hours when spring comes.
Sure, it's not "smart", but it can display the time in several different timezones, act as a stopwatch, and even has a button that lights up the display at night (which will probably drain the battery faster). All running off of two solar panels the size of the nail on your pinky.
The energy requirements are magnitudes away, you could never do that for a "smart" device. By the way, we also have the much older self-winding mechanical watches which harvest body movement; those are pretty reliable and can also run indefinitely without maintenance.
That sounds overly pessimistic. Computing power per watt has improved by several orders of magnitude over the last couple of decades already. Combined with more energy-efficient displays (anyone wants to improve on e-ink?), who knows what we could pull off 5-10 years from now, even without a significant change in battery capacities?
Well, I opened a calculator to prove you wrong, and guess what?
300K * k * 1GHz = ~4pW
Or, in other words, you are right, we can in theory reduce the power consumption of our computers so much that a current watch will last for a lifetime.
Computing power is not the limiting factor. Far more relevant are the display, which has efficiency physically limited by its luminosity -- and we're less than a magnitude away from the limit.
Even disregarding that for e-ink, there's the radio, which is limited by the channel capacity, from which I'd say we're also less than a magnitude away for isotropic radiation.
Better for the watch form-factor would be the kinetic drive. Plus it gives you an incentive to be more active and the watch can annoy you if you're letting it die by being sedentary. But both the Citizen-style "eco drive" and the Seiko-style kinetic drive could be baked into the same watch. That's plenty of energy for the deep low power microcontrollers that exist these days, even a fairly powerful ARM-based one.
But that's still not enough power to drive the screen on most "smartwatches". Nor their higher power cellular/wifi components.
Fact of the matter is, TFT screens eat batteries like no-one's business. No avoiding it. E-ink screens would do better, but have nighttime visibility problems that would be hard to overcome.
Plain and simple, the undoing of the smart watch is the battery. It's an obstacle we may someday overcome, but until that day, it's not likely to catch on and become the big hit everyone wants it to be.
> "... wireless charging to a point that I could just put my phone down anywhere on my desk and it could get charged."
The actual technology for this has been around for long time. I'm aware of at least two startups that tried to commercialise it but they faced the chicken and egg problem of (1) getting manufacturers to make changes to their devices/batteries and (2) getting enough the charging pads in enough places to for it to be worth the everyone's time.
Neither company survived but I suspect this was a market-timing problem. Given the wide proliferation of devices (phones, tablets, etc) some solution will eventually get adopted. If those companies were starting now I think they'd have a better shot.
"If the watch was so smart, why didn't it tell me at 9 PM that it was low on energy and that I should put it on the charger overnight."
Not all people have the same sleeping pattern. While 9PM may be cool enough time to remind you to recharge the battery, it will be a disaster for other people. I think either of following will be a great feature to add: 1) auto-recognizing sleeping pattern of watch wearer (may be hard to implement); 2) just to add an option in system setting your usual sleep time.
Exactly. And there will always be this kind of mismatch between what the human wants and what the device can anticipate as long as the human is smarter than the device. Even among humans it takes years of learning social norms before we are able to anticipate each other with any degree of accuracy.
People always get upset at these "smart" devices, and decry their stupidity, and yet they wouldn't really do that much better of a job were they in the device's shoes, so to speak.
I see so many comments here attacking the premise of the article. He isn't attacking technology and this man has studying these things. He has he smited your God of technology?
If we are to design autonomic systems that are safer, better and as the article says, smarter, we need to put real thought into the unintended consequences.
I had a WinCE device, one of the first ones ever made. It had two fatal flaws
* volatile storage, all notes, memos, calendar entries, everything was in ram
* it would wake up and vibrate and turn on the backlight to tell you it was running out of energy
Those two things in combination, meant if I left for the weekend and didn't put it in the cradle, it would commit suicide. It would be like having a brand new fresh ipod every monday. Lovely.
Confused Article with mixed messages - Surprised that it had so many up-votes.
The Author has problems with a Sony Smart-watch yet there is a picture of a Samsung Gear on the article.
Much of the criticism for the watch is indeed deserved - Smart Watches are more of a novelty at this point. If he had stuck with Smart Watches then this would be an acceptable and deserved criticism.
But he moves on to Cars and then muddies the waters with why long term attention spans inversely co-relate with automation.
What does one have to do with the other? Why go ranting about unrelated points to "round out" the word count?
I agree with many of the sentiments expressed in the article but am totally mystified as to why he was sleeping with the watch on. Maybe it was because he was testing the device and wanted to try different scenarios, but I just can't imagine any normal user scenario that involves getting in bed and going to sleep while wearing a watch. Is this something that people do? If so, that's the first problem right there.
>Automation has now entered the automobile. Alas, the automobile industry refuses to learn the lessons from aviation automation. The automobile engineers believe that they have solved the problems: cars will drive by themselves without any incidents. Humans will monitor the driving and if there ever is a problem, they will simply take over. In fact, the requirement for people always to monitor the self-driving automobiles is now incorporated into the law in some locations.
That's a legal requirement so people can test autonomous systems - which is very important. It's not suggesting that self-driving cars should always rely on a human to monitor them.
properly designing the software costs too much, requires too much time and effort and too much iterative testing. You can see why they just skip to MVP (for watches).
Yes, there are horror stories about "dumb" transportation software that linger in the forefront of our memories. But ask yourself, What is the burden of proof for this technology? Do self-driving cars and aircraft automation systems have to work 100 percent of the time before we accept their usefulness?
If that's the burden of proof, we will remain stuck with human drivers who are much more "dumb," by a measure of the car accidents and plane crashes they cause.
The author's argument against automation in these life-saving contexts is weak. His article would have been stronger if it stuck with points about watches and similar consumer products.
[0] http://www.statisticbrain.com/airplane-crash-statistics/