In my experience the concurrency will get you but it's also the lack of memory protection that will drive you nuts. The Linux kernel has everything running in the same address space so a bug in some dippy USB driver can crash the whole system. And that is why you shouldn't let n00bs write kernel code.

But not every driver gets included into the kernel. I wanted to use an open source ISDN driver and it was broken because some yahoo had decided that the kernel logging macros all needed to be renamed.

I did kernel development back before Linus even started on Linux and I avoid Linux internals like the plague because they're in a constant state of flux.

Two out of three of those things are not benefits for the customers.

How about a publicly accessible forum where the SUPPORT STAFF answer questions?

What's wrong with peer-to-peer support? Basically the company is free-riding on the backs of its users.

Maybe it's time for "corporate jail" - the company's operations get suspended for the time it's in "jail" but it's required to continue paying employees. That might finally start getting their attention.

Automatic cars for a taxi service wouldn't have user accessible controls - unless they're a JohnnyCab!

Large enterprise customers tend to buy "enterprise" grade hardware. Which, these days, is mostly just a fancy case with a label stuck on the front and a warranty. Amazon doesn't run expensive hardware, they run cheap hardware, and mark it up enough as a "cloud service" to make a good profit but still undercut the expensive hardware vendors. Essentially, you're paying Amazon for a warranty. If you're spending $100M/year on hosting you should be able to run your own cloud for less than AWS charges you but in order to do that you need to let go of the enterprise vendor security blanket.

Have you ever really looked at what cloud services provide? You would like to believe that if a physical server at Amazon goes down that you're just fine - the fact is that no, you're probably pretty screwed. Even worse if they lose an "availability zone". And those are just hardware failures. Just wait until they f-up a software release. When I spec'd data centers we would typically require at least two different carriers with no common points of failures for internet connectivity. Are you doing that with your cloud providers?

Cloud services are not as robust as they are portrayed to be. In order to reach that level of robustness you need to engineer your applications to fail-over between servers and availability zone and even cloud service providers.

Uber has different levels of service. This appears to be a crackdown on "UberX" which lets anyone drive for extra cash. There's also "Black Car" which uses limousine services (i.e. "Town Cars") which are licensed and insured. That probably remains legal unless there is some problem with them picking up fares anywhere.

We used Uber Black Car and regular taxis in San Francisco recently. San Francisco taxis have really gone to the dogs - we had one driver who did nothing except talk on the phone and swerved in and out of traffic. The limo drivers were much nicer, the cars were nicer and the price was about the same.

I think you're thinking of Single Stage To Orbit (SSTO). DC-X was pretty cool and had a strong possibility of working until they gave it to NASA who promptly broke the prototype and then poured billions of dollars into VentureStar only to give up before ever completing a vehicle.

SSTO is pretty marginal. You do need a high mass fraction to pull it off. The DC-X people thought they could but they never got far enough along to actually have it happen. However, multiple stages is proven technology and the first stage, especially, has usually got some margin.

In any case, we can armchair quarterback all day with no effect. SpaceX is actually DOING which is damned impressive to me.

Salt water is a big problem - the SpaceX plan appears to be to land the booster back at the pad, though, not in the water. No one has ever gotten a booster to fly back after a launch before, so that's a pretty big score for them.

It's easy to say "can't, too expensive, why are you wasting your money?" - the fun thing here is that SpaceX is wasting their own money, not the government's (the government is paying for the launches but not the experimental part). Maybe they'll be right, maybe they'll be wrong. However, they are trying and that's pretty exciting.

Probably not. Datacenters don't employ very many people. It's easy enough to fly in specialists to set them up and debug the really ugly problems.

Montana's total population is just slightly above 1 million. SF Bay Area is more like 7.4 million with a much higher percentage of tech workers. So, no, Montana isn't going to be the next tech hub because there aren't enough workers there. Might be a place for DC's if there's enough bandwidth.

Thank you!

I read a few of those long long ago but don't remember many plot details except maybe for some anti-Force sloths. Were Han, Luke and Leia in those novels and if so, how old were they? That may be part of the problem in adapting them.