Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Get HideMyAss! VPN, PC Mag's Top 10 VPNs of 2016 for 55% off for a Limited Time ×

Comment Re:I don't believe the claim (Score 1) 265

The problem is that as cars get better and better at doing this - it's going to become almost impossible NOT to fall asleep (or at least zone-out to the point of total inattentiveness to the road). If we can't make these systems sufficiently foolproof to allow people to do that - then we're going to cause accidents that wouldn't have happened had the car not had those safety features. The trick here is to build only those safety features that save more lives than they cost. As a society, we're going to have to accept that "Driver-fell-asleep and car-AI-hit-truck" accidents will increase - but "Reckless driver ends up in the wrong lane" accidents will decrease - and that the net result is an overall improvement in road safety.

The problem with that is that as a society, we're TERRIBLE at statistics. People are frightened of flying in planes - even though they are vastly more likely to be killed in a 30 minute drive to the airport than on the 8 hour flight they take when they get there. People will go to any lengths to prevent terrorists from killing a dozen people per year - but refuse to drive at the speed limit, causing hundreds of thousands of deaths per year. People switch to LED lightbulbs in their homes in an effort to stave off global warming - not understanding that removing 10% of the beef from their diet would have a much bigger effect.

Given that mindset - I suspect that as the number of prominent AI-induced car wrecks (inevitably) increases - public outrage will take over without anyone understanding that human-induced car wrecks are going down more steeply.

Tesla are trying to buck that trend with this press release - which is probably a wise and necessary "big picture" thing to do - but statistically, what they're saying is clearly bunkum.

Comment Re:Never go to be work "as is" (Score 1) 265

Perhaps not embedding stuff in roads - or beside the roads - that's horribly expensive and it requires governments all around the world to invest heavily and to have common standards. In remote areas, getting power to those sensors and computers could get difficult. There are problems when they all get covered in 3 inches of water or a foot of snow.

More likely is to design a distributed system where cars can talk to each other - an "Internet of Cars" if you like.

If the car in front of you can tell you exactly where it is, it's speed and direction from GPS and it's short-term intentions - then your detection of its gets a whole lot easier. If a big-rig truck can tell you about an incipient problem that it's detected but you can't see because it's in the way - then you have more information than a human driver has - and that's a good thing.

Think about the "Kid crossing in front of a parked schoolbus" problem. The schoolbusses sensors can see the kid standing in front of it - moving across the road. The car in the adjacent lane can't see the kid - and radar doesn't help. But if the bus can transmit "Human being at location X,Y,Z, currently moving at dX,dY,dZ" to anyone within 50 feet of it - then that's a massive win.

A distributed network of cars would greatly improve other things too - such as how to decide when to cross an intersection - or when it is safe to change lanes. Don't you just wish you could tell drivers in the adjacent lane that you need to pull off at the next junction and to please form a suitable gap for you to merge into? All you have right now is one blinky red light - and pulling that from an image in a camera is painful.

Comment Re: FUD ....the problem with radar (Score 1) 265

The problem with adding radar is that when 100 cars around you are also using radar. Distinguishing your reflected radar pings from those of another car nearby - or an oncoming car starts to get exceedingly difficult. We know it's possible because an entire swarm of bats can echo-locate together - but that doesn't make it an easy task. Worse still, to do it right requires careful attention to the frequencies and waveshape of the RF chirp you use...that's fine if one company designs all of the autonomous car radar transmitters - but not so good if every car manufacturer develops their own system in secrecy as seems to be happening right now.

If radar is to be a part of the answer - there need to be standards. Ditto for lidar and acoustic techniques.

If it's possible to make this work safely using only a couple of cameras (which is the way humans drive cars) - then I think that's a more robust solution for the longer term when there are many, many more cars on the road with these kinds of features.

Comment Statistics gone wild. (Score 1) 265

So far, it seems that there are reports of one death and one rollover incident - which could easily have become a death if circumstances had been different in ways that were not related to what decisions the car made.

Two incidents isn't a valid statistical sample from which to extrapolate to half a million deaths...perhaps just one person got lucky one day - if they'd died then Tesla would only be able to claim that their system is no worse than people driving without assistance. Perhaps the one person who died was exceedingly unlucky and Tesla would save very nearly everyone who used their software.

The fine details of what happened in those two wrecks (and more importantly in the unknown number of very-very-nearly wrecks) is what matters here and that won't be known until Tesla's have been driven under these conditions for ten times as long as they have to date.

What we know from the Tesla data right now is that their system isn't a total disaster (we haven't seen 100 deaths) - but predicting half a million lives saved isn't good statistics. The correct conclusion from what we know is: "We don't know yet".

Comment What if cars can talk to each other? (Score 1) 364

The problems get very difficult when the cars' choice of actions are determined by the interests of the car manufacturer (and possibly with the insurance company - if those are still different entities) rather than by the occupants.

For example, it seems likely that when there are a reasonable percentage of autonomous vehicles out there, they will be able to communicate with each other - that's a handy thing for negotiating who goes first at intersections and for crash-avoidance.

So now there are a whole raft of other moral dilemmas at levels far below "How many people die?":

* Should the car that knows that its owner is late for work go first at intersections?
* Can you pay more for a car that gets preferential treatment at intersections?
* On a freeway, can cars choose to slow down to save gas or speed up to get there faster? How does this work when cars are "drafting" to save gas? In a "road-train", who gets to decide the speed of the train?
* Are cars allowed to lie to other cars?

Then there are issues about low level accidents - where no humans are harmed:

* In the event of a choice between a fender-bender with car A and car B - can your car figure out which one will cost the least to repair? Will this result in more crashes with cheaper cars?
* Will insurance companies insist on cars choosing outcomes that minimize their liability?

And when humans are harmed:

* From an insurance perspective - it can be cheaper to have the occupant of a vehicle die than having lifetime health issues caused by the accident. If volkeswagen will fake emissions figures and indirectly cause a bunch of people to die as a result - who's to say that some car company/insurer won't bias the AI's parameters to save them money?

It's naive to assume that accidents will cease altogether when AI drivers take control - cars are complicated machines - and parts break unexpectedly all the time. If your brake line suddenly ruptures, then the AI's expected stopping distance is shot to hell - and someone can still die. The AI will need to make life and death decisions as well as broken-rib-versus-crushed-ankle and my-insurance-pays-versus-his-insurance-pays choices. Only now we add safety-reputation-of-my-manufacturer versus safety-reputation-of-competitor decisions too.

This is going to get difficult! Lawyer up!

Comment The Internet is a place. (Score 1) 211

In common usage, the Internet is a place. "My wife and I met on the Internet" is no different than "My wife and I met on Earth".

When it's not being used as a place, by all means down-case it, "I had a hard time getting internet access at my hotel" versus "I had a hard time getting onto the Internet from my hotel".

Comment So leave WIkipedia, not the entire universe. (Score 1) 379

Sure, everyone gets pissed at the Wikipedia community from time to time.

The answer lies on the following line of options:

1) Pick a less contentious set of articles to edit...politics is toxic either with or without Wikipedians.
2) Take a "Wiki-break" - simply leave the site for a month and do something else.
3) Delete your Wikipedia account - treat Wikipedia as a read-only medium.
4) Boycott Wikipedia altogether.
5) Disconnect the Internet to your house. ....
27) Go live on a far distant island where the word "Internet" is unknown. ...
999) Kill yourself.

Personally, I'd want to get up through at least the first couple of options before jumping to the last one.

Comment Re:Android implementation is crap anyway (Score 3, Informative) 340

My ancient Moto-G phone has working FM radio - the reason it doesn't work with Bluetooth (well, sorta) is that FM radio's *need* a couple of feet of antenna - and the way the chip is set up is that it uses the headphone wire as the antenna. So if you don't plug in a physical set of wired headphones, it won't work.

The app I have tests that the headphone is plugged in and won't run without it.

Now, whether you can plug in a fake headphone wire and use BlueTooth for audio transmission - I don't know - it's not something I've tried.

    -- Steve

Comment What I think Kickstarter should do... (Score 5, Interesting) 139

I think they need a rule that first-time project owners are only allowed to collect up to 200% of their "goal" amount. Once the project hits that number, the "PLEDGE NOW" button goes away.

That limits the amount of damage that a first-time project owner can do. It doesn't prevent them getting the money they need to get the business started - plus a healthy "win" for doing a great job. But it would prevent viral projects from dumping so much money into someone's lap that they become intimidated by the magnitude of the task and find it easier to take the money and run than to complete the project.

Once someone has proved themselves and delivered as promised, they can try again without the cap.

Having run 5 successful Kickstarters myself, most of them 400% or more over goal - I understand how daunting it can be. When the project is running, a kind of "red mist" descends and pushing the total higher and higher becomes highly compelling. When the countdown expires and you suddenly realize that you're tens of thousands of dollars better off - it's exhilarating. But the next morning, when you start to realize the magnitude of what you've just signed up to do...it can be very daunting.

It's also very difficult to plan a project when you don't know whether you'll sell 100 widgets or 100,000 widgets. When you go from "Oh - I can just 3D print that component at home - and solder that switch to the circuit board myself!" to "I've got to get a $10,000 mold made by a company in China and I have to fly out there to make sure it's OK - then find a factory that can solder that switch on for me."...suddenly things get much more serious.

It's exceedingly difficult to design, price and schedule production on a product where you have literally ZERO idea how many you'll sell.

So for that first project - make it so I'll know that I'm selling between 100 and 200 widgets.

Comment Re:Legal Recourse? (Score 3, Insightful) 139

So...force him to give the money back - right? The problem is...the money isn't sitting in an account someplace. The guy spent it on building a house.

So take possession of the house and sell it...or allow the guy to take out a mortgage and pay the money back that way? The problem is that - the house is only half-built...it's worth practically nothing in it's present state...he needs a construction loan to finish it.

So - the smart thing to do (seemingly, at the time, after talking to lawyers) was to let the guy finish building the house - then he can get a mortgage - then he'll pay the money back - and everyone walks away a little older and wiser, but otherwise unscathed. But - if you make this all public and get the cops involved - then he'll never get the loan (nobody lends money to admitted embezzlers) - so he can't finish building it and everyone loses.

So - you draw up a legal agreement - that you get your money in stages as the construction loan comes in - and, just for safety's sake - you get the evildoer to confess on camera, so that if he doesn't pay up - you'll go to the cops. ...and since he didn't pay up - the cops are now involved.

Admittedly, this may not have been the smartest course of action...but then letting even a close friend sit on $600,000 without putting it into a shared account...also not so smart.

Comment Re:25 years and nothing to show for it (Score 1) 565

Think of Android as a window manager - like Gnome or KDE (it's actually a little more than that - but it's close). Ask yourself how much exposure to "Linux" you get from KDE? Well, I guess you feel closer to Linux when you open a shell window - but Android has a shell too - there are apps you can download to give you access to it. Run one of those on your phone (you might want to get yourself a bluetooth keyboard and 'root' your phone) and you can run 'ls' and 'cat' and all of the other good stuff you're used to having right there in 'bash' in a standard Android 'app' environment.

The sheer inconvenience of working on a phone or tablet is all that stops you from "doing real work" on an Android device. Get yourself something like the HP Slatebook (http://www.amazon.com/HP-Slatebook-14-Inch-Touchscreen-Android/dp/B00KB3K6G4) - and you have a laptop that runs Android - and which has 'bash' and all of the command-line goodness.

Oh - and ChromeOS...also Linux.

Slashdot Top Deals

As in certain cults it is possible to kill a process if you know its true name. -- Ken Thompson and Dennis M. Ritchie