There's a great West Wing episode which discusses why we should, but somehow I think that wouldn't gain me much here. Discussions of the nature of man, and the establishment of wonder being particularly squishy in hard science terms.

Instead I'd point out that all safety critical systems are engineered around the notion of redundancy. Shit happens, and when it does, things break down. When that unexpected thing happens to our Earth-bound ecology, what, exactly, is our safety strategy? Hide in a hole? For how long? What if it's biological? What happens if someone accidently creates Card's molecular disruption device. We can't reasonably colonize another star system (yet) but we aren't *that* far from being able to establish some very worthwhile planetary redundancy. It's worth it because we are stuck on this rock that I think we should rename 'The Single Point of Failure'.

It seems like the obvious question, from a geek activists point of view, is: Why are the firmware components in our cars not open source? I should be able to compile and validate the loaded firmware so I (or, you know, someone who actually cares) can verify the security, legality, and safety of it's operation. It isn't even required that I be able to re-load the firmware, just that I be able to validate it.

I feel we've reached the point where more pixels doesn't really help. What we need is people to come up with better ways to interact with what's already there. I can't *do* anything with the megapixels I already have because simply interacting with the device obscures 10% of the screen, and 100% of what I wanted to interact with.

For instance: I've been waiting for a good CAD system that works well with a touch interface since my first iPhone's 3" screen. Now I'm at 5" with 20x the pixels and there's still no decent touch capable CAD app. Maybe it's eye tracking, maybe it's rear panel touch, maybe it's voice, I don't know, but something on the UI front must be changed.

We've got a pretty bad record so far at attempting to intercept objects moving at delta Vs approaching orbital velocities (which 10km/s is definitely that). That's using a guided projectile to attempt the intercept, and counting intercept as close enough to damage via explosive charge. Based on that I can't see us firing a tethered harpoon containing no guidance or propulsion engines, and having any expectation at all of hitting something moving that fast. And that's not even considering the question of actually attaching to the object.

While I agree that there's bias in the reporting here, in all cases the driver of the "autonomous" car has not been cited as at fault. Thus they are certainly within the definition to say the software wasn't at fault. I am with you in one respect: All of the data should be made public on every incident, and on any close calls where actions of any outside agent (the driver of the "autonomous car" or one in a nearby car) prevented an incident. Unfortunately it's not in Google's interest to do so, and there are no laws mandating it, so it probably won't happen.

Almost certainly. Though they will bring in several well-respected highway safety engineers to testify that following the flow of traffic is significantly safer than following the posted speed limit. Enough jurisdictions will lose money arguing these cases that there won't be money to be made by writing the tickets. Absent both the financial and safety benefits the police will stop issuing the citations.

Both the 6 and the B keys belong on both sides of a split keyboard. It couldn't possibly cost more than another $1 and we can get back to fighting over real significant, intractable problems, like the Oxford comma.

Mostly I'd agree, but there are a few exceptions:

1) no recursion. Except perhaps forms of tail recursion known to exit in bounded time. But you definitely have to ask yourself why you are doing something that could either be unrolled into a loop or has some kind of exponential growth potential.

2) Local variables are fine so long as the analysis is done to guarantee the maximum stack requirement is pre-committed. I mean, realistically the return pointer is a stack variable, so just calling a function that returns would violate the "no local variables" rule. I wouldn't allow dynamically sized items, because then bugs might cause stack overflows...

3) I agree that malloc and free are forbidden *during real-time operation*. However, in some situations you use dynamic allocation if you can pre-allocate all necessary elements prior to entering operational states. This really depends on whether your system *has* a pre-operational state,

I get this. And as a software engineer I fully agree. However, in practical terms, there shouldn't be any dynamic memory management happening at all.

It's a real-time system. It *must* interact, on time, with all the planes that are in it's domain. That should be a bounded, predictable load, or there's no way to guarantee responsiveness. Given that, an analysis should have been done on the maximum number of elements the system supported. Those elements should have been preallocated (into a pool if you want to treat them "dynamically") before actual operation began. If/When the pool allocator ran out of items it should do two things: allocate (dynamically) more, and scream bloody murder to everyone who would listen regarding the unexpected allocation.

This is (one of) the reason(s) I generally haven't liked garbage collected languages for real time systems. There's rarely ever a way to guard against unexpected allocations, because *every* allocation is blind.

I'm a software engineer who focuses on automating manufacturing equipment. I've been hearing bad stories about Amazon for at least ten years now. About a year ago I got a call from a head-hunter about a job opening dealing with autonomous warehousing and order fulfillment that sounded like a dream job. Five minutes on the phone got me the name of the company (Kiva Systems) and two minutes of google told me they were being purchased by Amazon.

Nope. Hard stop. I don't even need to know how much they are offering. I love what I do, and I won't willing work for someone who risks making me hate it. Go ruin someone else's psyche. I'd be *really* interested to know how many people there are just like me.

As a person squarely in the "my vote doesn't count" category, I feel compelled to say that it isn't that the candidate I'd vote for can't win, it's that no one I'd want vote for could ever possibly win. There is no possible way for me to vote in any way that both A) represents my beliefs and B) adds support to the least bad alternative. It really has nothing to do with diluting my vote. My vote would never be in the top five parties in the USA, much less the top two. I'll never be represented any way other than badly. Mostly that's ok because I just don't care that much.

If the government let them in, they should have a voice. They are subject to the penalties and restrictions but not the benefits? That is treating them as second class humans, which I'm pretty vocally opposed to. Illegals are more complex, but given how little we actually spend enforcing those rules (in 2013 we deported less than 5% of the estimated illegal population) we seem complicit in their presence here. Given that I find it hard to argue they shouldn't be afforded representation on the same grounds. In my mind representative governance is a human right, rather than a right granted by citizenship.

But I'm pretty well not in the majority...so there's that.

I see this complaint every time an autonomous car story comes up. Is there actual evidence somewhere that rain and snow have any significant impact on the vehicles' ability to perceive an environment or on their ability to navigate it? The sensor fusion algorithms already deal with noise exclusion and transitory sensing failures. I can believe heavy snow which obscures the lane markings would be an issue, but even then that doesn't really prevent collision avoidance, or even safe driving ability so long as there's always to option to just go slower. The rules of traffic interactions in the absence of lane markings are pretty well defined, and are mostly just "do what boats do".

They lose some range in extreme cold and hot temperature ranges. In cold weather much of that difference is purely in keeping the temperature of the cabin and battery heated to optimal levels. It is suggested that you preheat your vehicle from grid electricity before driving to maximize range (implying plugging it in while you're at work I guess). It is also recommended that you use seat and steering wheel heaters because they more directly reach the passengers, allowing the cabin air to be maintained at a slightly lower temperature.

You are totally missing out on an opportunity to lord your success over your peers. You should feel free to call them minions or vassals; Also, I whole-heartedly recommend using the salary information in your next meeting. "Well, George, from the look of things I make almost twice what you do. Apparently no one who matters actually cares what you think, so STFU."

Winning friends and influencing people through transparency...

These vehicles use frequency matched LIDAR since it is unaffected by ambient light conditions. I'm not sure what you were following, but the computer certainly didn't stop for a shadow on the road.