I would more think of a pack-in (get one code unlocking it and a few other games on steam when you purchase a SteamBox)

because it's not possible to cram one inside an Apple iShiny ?

Nearly "all the rest":

- Intel only has an official open-source driver on Linux (only Windows gets a proprietary one). Their driver works, although it has less features than all the binaries (older version of OpenGL support).

- AMD has very decent *open source* drivers. Althought their *closed source Catalyst* isn't as good as Nvidia's binary, AMD's opensource Gallium3D is very good. It has sometime problems with the latest cards (indeed currently, it has problems with the latest generation of GCN 1.1 Volcanic Island cards), but runs otherwise good, and for older cards is better to the point that AMD drops support for old cards in Linux Catalyst and officially points to the opensource driver for supporting older hardware.

Meanwhile:

- Nvidia might have the best closed source driver (though still not collaborating nicely with standards, but this situation is slowly evolving), with full latest OpenGL support, etc. with decent stability...

- ... the nouveau driver is very bad. Not the authors' fault, as they could only rely on reverse engineering to get it working. So it's good that nvidia is finally making some baby-steps in the direction of helping nouveau.

You, your robot car and anyone else is required to always stay at a safe breaking distance for anything in front. Legally, you should be keeping a safe distance *even in the absence* of metal pipes, so if anyone in front needs to break due to a sudden apparition of a kid, you won't rear-end them.
The robot car is less likely to be tail-gating than the average asshole.

you'd be surprised, but that technology is *already* in production at several constructor (Volvo is equipping it as a standard).
Cars currently on the street are able to spot the kid and apply emergency breaking if needed. (If you ask: yes, that can currently be overridden simply by forcing otherwise on the car's gas pedal). In the next following year, such technology and other similar collision avoidance system are going to be mandatory for new cars in EU.
Now the small killer detail? You can extend the system by putting more cameras. Said volvo has also camera under the side mirror *constantly* monitoring blind sport all the while you drive.
Human driver? No way to graft extra heads/extra pairs of eyes, so only possible to concentrate and whatch one single target. Either whatch in front of the car *OR* whatch either of the side for blid sport *OR* get distracted while fumbling with audio/phone/whatever. Consumer level cars currently on the street are able to monitor all this *continuously, without ever interrupting*.

Actually google's car have managed to total 700'000km without any pot hole-related problems, so one could decude that their solid-state lidars work as intended.

Speed *is* the dominant factor simply due to physical laws like E = 0.5mv^2 each increase of speed squares the amount of energy involved in case of crash, or kinetic energy that you must dump during an avoidance manoeuvre. (See comment above about never tail-gating to avoid rear-ending).

A computer can compute and keep a safe distance from anything in front to be able to break in case of emergencies. And unlike a driver who might get distracted (and hence needs reminders), a computer keep continuously keep an "eye" on that too in addition to watch for everything else. (Some simpler forms of this are already in production in automatic cruise controls on cars currently on the street. The only difference is that these car can only keep safe speeds regarding other cars [= They will slow down not to rear end a car]. They currently can't anticipate safe speeds for upcoming turns on the road, without having signs hinting a top speed that the camera can read [= they aren't able to slow down before a sharp turn]. Whereas google cars have repeatedly demonstrated to successfully adapt speed to road ahead)

...which by it self would *NOT* be a problem if you kept a safe distance from the truck in front of you *AS REQUIRED BY DRIVING LAWS*.
(If you were stupid enough to tail gate a truck, you deserve anything that happens to you next)
(And the aforementioned collision-avoidance system could actually save you under some circumstance - simply due to having much faster reflexes than you and being able to apply emergency breaking before you even start to realise that you're on a collision course)

...again at which point you were probably already violating laws regarding safety distance.
Also, don't forget this nice thing called "inertia". Hint: the truck is currently already moving forward and very heavy. Jackknifing can't realistically be the trailer suddenly accelerating on its own. It's much more likely the truck breaking, but the trailer's breaks failing and the trailer still moving forward at its initial speed. Keyword here is "still moving forward". As the trailer is moving forward, the safest course would be to break and stop before reaching it... provided you have enough room to break (I hope you were keeping a breaking distance, didn't you?). Or hope that your collision-avoiding system would be already applying the breaks for you.
Starting complicated Hollywood manoeuvre around a truck that is about to fail catastrophically doesn't seems very safe...

It also takes time for the human to process its input and it also requires that the human will react in a correct maner (and won't try something stupid).

Actually, the computer would already have started reacting long time ago, when it already did notice the change in relative speed to the truck. As soon as the distance between the two decreases, car with modern automatic-cruise-controls and collision-avoidance are already slowing down.

not "if". Modern cars, already "do" keep safe distance unless you hit the accelerator.
and the definition of 'safe' distance is that distance which allows you to stop in case of emergency (always assume that the vehicle in front of you could be hitting breaks at any time, like trying to avoid the kid from point 2 or trying to avoid a fallen object).
Due to the way inertia works, if you kept your distance you can always break in time if you kept your distance.

(Or you should change hobby, as following trucks that transport exotic matter in- and out- of area 51 doesn't seem a very sage one).

In fact electronics can help you in more circumstance. Like if you aren't actually following the knife jacking truck, but driving in the opposite direction in another lane. As a human, you probably aren't constantly monitoring all incoming truck just for the chance that it might be knifejacking. You probably will notice only once the incident is well underway. And there's a big chance that you'll collide (in a very bad manner, again due to inertia, due to the relative speed as you're heading toward each other, and that this relative speeds counts squared), while on the other hand, as a human you might find some creative way to attempt to avoid (if you are reactive enough).
Meanwhile, a modern car will detect it much earlier (As soon as the truck steps a bit out of its lane, into yours, moden car will simply treat this as a standard "object on the lane, currently on a collision course), the car will automatically slow down, sound a warning and prepare to be able to do an emergency breaking if the need comes. And you'll be informed much earlier, and will have time to react (like changing lane, which will also be easier as the cars was continuously monitoring your blind spot anyway).
And that's only a modern car. For google cars, google has already shown their on board software to be able to consider obstacles on the road and re-plan a different route to avoid them. By the time the truck's trailer is realy sideways, the google car would have moved away, on farthest (rightest if you're in a right-driving country).

But there are a few other things:

- do not underestimate how far we've already reached *right now, as of today, with currently street-legal cars*. The fact that you couldn't imagine how this would be done on an Apple][ two decades ago, doesn't mean that it's not routinely deployed using modern hardware.

- a human can get stressed. There would be a difference between a well seasoned stock-car pilot who'll know to keep calm and has experience nowing how to react. And the terrified random six pack joe, who might start a stupid manoeuvre in a panic (like ramming his car into the one next to him while trying to avoid something in front, whereas simply keeping distance and doing an emergency break would have been the correct course of action). A machine won't. A machine can be programmed to react as the best stock car pilot.

- we are speaking long term progress. wheel-less cars isn't something that is going to happen tomorrow or next week. It's also not only going to be produced by google alone (BMW, Volvo, MB, Tesla, ... all have plans toward automation). By the time wheel-less cars become common, there would be lots of other cars with varying degree of automation. In case of complex problems, the faulty vehicle (Truck whatever) might as well have bordcasted a warning around and all the autonomous cars might as wall have decided to follow the hint and move away to safety. When speaking about extreme far-out development of automation, think networks, not single individual. Otherwise you might sound as someone complaining that the loud motors of sports cars will scare all the horses pullings the carts around it.

- blackboxes can still give valuable information, programs in an autonomous car can still be updated. Even in the tragic case of an accident, all the remaining autonomous cars can be made better.
By the time wheel-less cars become popular, very probably it would be much safer to let computer handle everything than putting behind commands someone who might be over-confident, inexperienced, or plain stupid/drunk.

You know you're on /. when you need to check half of the words on wikipedia, just to be able to understand a 10 words sentence. :-)

LibreOffice and OpenOffice.org do exactly the same.

And I'm sure that at least a few /.ers have taken git and applied it as form of "edit undo history" stored on non volatile support.

Well, git beats it, on the ground that you local clone still has the same roll-back history, but without polluting everything with tons of silly name, specially after one version has gone through an e-mail trip.

No, I think "Power outage" are these moments where the lights in all rooms suddenly go dark, and you grab the screen you were looking at and use it as a make-shift light source to find your way.

It boils down to the fact that you need a killing procedure which is in a way a single point in time.
You do action X and the victim is dead. Or you don't do it and the victim is still alive.

You don't want a protracted procedure, that takes 2 hours to kill someone, and might result in heavy brain dammage?

If the governor phones in to give pardon, while you are in the middle of a lengthy slow killing protocol, what do you do ?
"Oh sorry, you're a bit late, at the step we've reached, it's better to continue with the protocol and have him dead in 1 hour"
or
"Okay, we abord everything, but if he wakes up, he might live the rest of his live as a vegetable".

And in addition to that, the US companies might not have the patent to produce said substance.
Very likely, the patents are in the hand of the same EU companies, and they are as likely to accept selling the patent for execution-related application as they are likely to sell the substance it self.

I second that advice.
As she still has higher functions, and she's still able to communicate (throuhg blinking), the most important thing is to ask her what she wants, and respect her choice.

If she asks the plug to be pulled, respect her choice.

On the other hand, as she had recently a child, chance are high that she'll want to fight to be able to see the child grow and know that the child won't stay motherless.
So aim at all the modern brain-computer interfaces.

They both server different goals.

Server encryption, helps securing the service.
But it doesn't address privacy. (the channel is only secured between 2 servers, or between a client and a server).

End-to-End encryption (like OTR) is for privacy.
It make sure that, no matter what, the message will stay encrypted during the whole transit between one user to the other user.
Even during the time spent on servers, an OTR-encrypted message is still useless and not eavesdropable.

It is, it's only "not a real problem" for user sending one-to-one e-mails.
As soon as you send one-to-many e-mails (newsletter, mailing-list, announcement, or just corresponding with lots of friends) this starts to be a problem, as you need to recalculate a new hash for all mail recipient.

Not a hold-up, at all.
No true spammer does still mail all his/her spam from home using a single mail server (the spam will be immediately detected and blacklisted).
Spammer do routinely use botnets. As each single bot doesn't send much SPAM itself alone, each single bot won't have much problem with hashcash.

Remember, the definite characteristic of SPAM isn't that it's addressed to multiple recipient, it's that it is unsolicited.
It happens that it used to be sent simultaneously to several recipient at a time, because back then, it was best done so.

But hashcash (like many other proposed systems) are not detector for "unwantedness" of e-mail, they don't address the fundamental problem with SPAM, they only address minute detail of the way it was done in the past.

You haven't been using Facebook Messaging, recently ?
The only reason it's not considered such by all is that they still tactfully manage to avoid calling it "E-Mail".
But the set of functionality is very similar to any other webmail system (including attachement, etc.) minus the interoperability.

The problem is that this absolutely doesn't look like the innards of a classic internal combustion engine car.

It would require quite some work to retro-fit it into a classical car... Which could probably be a nice additional for MTV: next to "pimp my ride" and "trick my truck", let's begin the "spark my suv" TV series !!