That is true. My comment was(in retrospect, very poorly explained) narrowly focused on an issue I heard a lot of complaining about from people operating reactors in the US(PWRs, if my memory serves): between stray hydrogen from the water in the primary coolant loop and massive neutron flux, a combination of hydrogen embrittlement and neutron damage had a way of pushing even very classy alloys into serious risk of developing cracks; and properly servicing internal parts wasn't something you did lightly, since you'd have to substantially lower output power or take the reactor offline while doing so(and when you've got that much capital equipment sitting idle, team balance sheet is not happy).

None of these stories ended catastrophically, or even dramatically, nothing even approached leaving the containment vessels; but the complaint was that speccing materials for use inside the reactor was even less fun than handling plumbing for chemical plants, refineries, and the like.

An engineering challenge, it's what engineers do; but not good for cost cutting.

The special demands of finding materials that work adequately in enthusiastically radioactive environments don't help. Some are worse than others; but I don't think that there is anything that appreciates prolonged neutron bombardment. Can make for some very expensive repairs inside the reactor assembly.

The tricky question(and the one that I've been bombarded with vehement and competing answers on, which has left me confused) is whether nuclear isn't cheap; but military procurement slush used to make it look that way; or whether nuclear could have become cheap; but military procurement slush made that unnecessary and potentially even directly inhibited it.

It's definitely the case that military purposes kept the money rolling in for R&D, pesky questions about safety and storage largely under wraps, purchases of a lot of equipment that could also make plutonium, and some PR-piece "Look at how fuzzy and peaceful nuclear energy can be!" reactor installs at home and in selected friendly-and-not-too-likely-to-change locations abroad.

It's likely that, at the same time, this left the industry largely in the hands of companies that are very, very, good at government contracting; but perhaps a bit shaky on less lucrative and parasitic forms of economic activity.

Where the optimists and the pessimists part ways is the question of whether nuclear energy is in fact just not terribly economic; and so achieved certain unique capabilities for cost insensitive customers, while largely floundering without them; or whether nuclear energy as an industry was wildly distorted by catering exclusively to select cost insensitive customers with substantially different needs than energy production, and simply needs to develop product lines that reflect current requirements.

Are they just incompetent, or do they do most of their work for the same government that owns a third of their stock and not have a lot of experience with real customers?

Honestly, if we are stuck with the NSA amassing a database of all the phone calls, ever, anywhere; and a policy of using CIA killer robots on people who annoy us; I'd be a great deal happier if we at least got some visible benefit from the whole mess by using these assets to locate and terminate telemarketers. They have to stick out like a sore thumb in call traffic analysis, and I'm pretty sure that 'the corporate veil' is not rated to withstand most contemporary munitions.

It's possible (indeed virtually certain) that insurance types will factor in the expected value (positive or negative) of a given feature. They already evaluate expected costs tied to more nebulous associations between a vehicle and risk (Does model X get stolen disproportionately frequently? Are buyers of model Y, in midlife-crisis-crimson possibly not the most cautious of drivers?); if assisted braking or rear-view cameras, or lucky rabbit's feet, reduce the expected cost of insuring a given vehicle and operator, they'll presumably be folded in. How much of any savings the end users sees may or may not be an exciting number; but that'll be more about relative bargaining power.

I suspect that reductions in legal requirements are less likely. With things like BAC, people are already pretty tepid judges of when they've actually had one too many, and keeping the test equipment and testing environment calibrated, reliable, repeatable, and adequate as evidence is a fairly big pain. Even if we assume that all the relevant laws are 100% about public safety and have absolutely no secondary purposes (which is a matter of some...doubt), those aren't conditions that are going to endear some tiered system of caps based on a the car's feature matrix to anyone. Purely informally, effective input stabilization, assisted braking, and any other tech that keeps your car moving in a nice respectable, not-drunk-looking, way even if you are a bit sloshed will probably reduce your risk of being pulled over and tested, and thus effectively raise the limit a bit (except at the delightful 'sobriety checkpoints'); but if they don't mask the effects well enough to avoid attracting attention, I'd bet that the legal results will be the same.

Ah, I have a slightly different philosophy for catastrophe management, one that I find hedonically satisfactory and wish to recommend to you:

You don't want to be moderately impaired in the case of an emergency. Should the emergency prove relatively minor, slurring and vomiting while making your exit from the damaged vehicle at the crash site will be undignified and uncomfortable. Should the emergency prove catastrophic, you'll be much better off dying while deeply relaxed and pleasantly intoxicated, rather than indulging in theatrical heroics.

Whenever possible, try to either be ready and able to manage the situation to a satisfactory conclusion, or enthusiastically accept that the situation is totally hopeless and apply yourself to be business of dying as pleasantly as possible. Just don't fall between the two, which is the dreadful strategy-chasm that combines as much or more effort than option #1 with as ghastly, or worse, an outcome than option #2.

How many drivers? Probably not that many, unless the safety envelope is very wide indeed. However, if the autonomous car is, in fact, autonomous, the question is also how many non-drivers would like to have access to the road, most of the time, without becoming drivers. Especially if they don't have a choice(visual or other incapacity that precludes driving, alcohol violation, too young, etc.) or their use case is relatively miserable driving(If you are going to have a shit commute in heavy traffic, do you want to 'enjoy' the 'freedom' of being the master of your vehicle, or spend the time doing something that sucks less?), the value of becoming a driver may not exceed the hassle.

The initial systems are going to fail miserably in any environment less constrained than a test track, so they'll be too limited to be of much use to anyone who can't also drive them when the need arises; and there is a strong cultural value associated with gaining access to driving capabilities(in some suburban areas, you basically aren't a real person until you can drive; because if you can't do that you are homebound and dependent); so I suspect that swaying existing drivers will be harder; but convincing future non-drivers to just never bother to head down to the DMV at all, and instead employ an autonomous vehicle either casually(like zipcar; but with autonomy) or for more lasting lease or ownership arrangements might well be easier.

People already put up with stranding risks of various sorts(it's not as though cars breaking down is a thing that requires computers, and mass transit, cabs, car-pools, etc. have their own failure modes), so as long as they skew more toward 'annoying' than 'overtly lethal much of the time', people will definitely risk it if the convenience is sufficient.

Your suggestion about the pace of the transition is highly plausible; but that needn't imply much regulatory complexity: At present, a car with a licensed operator can have a variety of convenience features that involve a level of automation and (very) bounded autonomy without any variation in what type of license you need for what level of features. There may be some hassle on the vendor's end, in convincing the relevant feds that their intended new feature isn't an automated accident generator; but there's nothing on the driver side.

Given that operator handoff is most likely to happen either under relatively hairy conditions, or when some system failure has left the automated systems unable to cope, there isn't an obvious incentive to relax the(already not terribly demanding, at least in the US) requirements placed on licensed drivers until 'self-driving' actually does mean 'self-driving'. If it means 'sometimes self driving, except the hard parts', that may require less operator effort; but not obviously less operator knowledge(if anything, given that drivers usually get somewhat safer with experience, at least until they hit the point where each additional year stops making them less young and stupid and starts making them more old and inept, I'd be particularly worried about the likely performance of somebody whose vehicle is sophisticated enough to coddle him most of the time, then screams and hands him the wheel when the situation is already halfway lost.)

I have no doubt that the laws(or at least the liability litigation and insurance-related contracts, even if carried out under existing law) for damage and death caused by partially-automated vehicles will be an epic nightmare of horrendous proportions; but on the operator licensing end "If you might have to drive it, you need a driver's license; if you won't have to drive it, you don't." really covers a lot of territory. There might be some incremental adjustments, mostly to the format of the test(say, allowing use of a rear-view camera in addition to mirrors and over-the-shoulder during tests of parking); but not too much need to complicate things.

There's also the whole body of meta-analysis, which definitely has its uses; and also tends to substantially increase the number of studies you 'used' without necessarily vastly increasing the number of pages you had to slog through.

It would be my pleasure to see the whole lot of them have the actual text of the law applied to them as though they were a tattooed black guy with multiple priors.

We often have suitable laws; but they just mysteriously never even get brought up, much less by people in a position to do something about them.

It's either bullshit(fairly likely) or the rules need to be changed yesterday(actually, at least a couple of administrations ago).

Aside from the obvious issues with complying with transparency, discoverability, and archiving requirements that are legally imposed on official business even at much lower levels(heck, I've done penny-ante IT minion stuff for small municipalities that was subject to public records laws that would have made doing things over personal email grossly unprofessional at best and illegal at worst, and she's the fucking Secretary of State...), what about security?

Given the delightful creep of the Top Secret National Security Stuff blob to cover ever larger swaths of DC, surely the Secretary of State does some emailing about stuff that is, at least for little people, probably supposed to not leave the SCIF, much less be handled by who-knows-who at some random email provider or a DNC mailserver admin.

I can't speak for TechyImmigrant; but it sounds like he is nervous about things that suggest software development(especially by people with no track record of it) substantially more involved than microcontroller firmware. Which is probably fair, given that even professional developers working for companies that make software have horror stories.

If that's in fact his reasoning, there's probably a distinction between "Using a handy bluetooth module that acts like a serial port because most laptops and no cellphones have RS-232 ports", which is relatively safe, easy, and close to drop-in; and "Adding a nice, easy, drop-in wireless module so that our smartphone app(coming real soon now, we promise!) can communicate with the onboard web interface we've almost finished writing!", which has the potential to go nowhere, slowly.

Also a reasonably handy way(not necessarily optimal; but good enough to get the job done) to organize the equivalent of a group buy. Unless you really love screwing around with toner transfer and ferric chloride, PCBs get markedly cheaper when you go from 1 unit to a few hundred or few thousand units, as do pretty much all the components you'd stuff the PCBs with. DIY isn't impossible or anything; but it'll be a good bit cheaper per unit if the quantity is higher.

What I find a bit surprising about this story is not that it turned out to be a kickstarter disappointment(big surprise there...); but that they managed to survive a legal challenge(expensive, unpredictable, bolt-from-the-blue), and software development going to hell(common enough that you can safely assume it'll probably happen; but still a potentially crippling blow to timelines and budgets); but then got blown to hell by the BoM somehow going from 'sell for $99' to 'can't sell for less than $350'.

Apparently naively, I would have expected BoM to be markedly more predictable, and controllable, than either legal or software costs.