Yes, I think it's very interesting when you look at Figure 4 here. They almost look like they could be an artist's interpretation of the things they're supposed to be, or a similarity that a person might pick up on subconsciously. The ones that look like static may just be the AI "being stupid", but I think the comparison to human optical illusions is an interesting one. We see faces because we have a bias to see them. Faces are very important to participating in social activities, since they give many cues to another person's emotions and intentions. It's a whole form of communication. A lot of other sensory biases and reactions are related to things like finding food, avoiding predators, and understanding potentially dangerous obstacles (e.g. if I step here, am I going to fall down?).

So if these are optical illusions for computers, what are the computer's biases based on? The computer isn't trying to find food or avoid predators, so what is it "trying to do" when it "sees"?

Even if national borders were to become meaningless, I'm not sure that jurisdiction stops being relevant. Which law enforcement agency will pursue the crime, and which court will it be tried in?

Computer learns to pick out salient features to identify images. Then we are shocked that when trained with no supervision the salient features aren’t what we would have chosen.

I see this as a great ah-ha moment. Humans also have visual systems that can be tricked by optical illusions. The patterns presented while seemingly incomprehensible to us make sense to computers for the same reason our optical illusions do to us -- taking short cuts in visual processing that would fire on patterns not often or ever seen in the real world. Which BTW means even as is, this type of visual identification is still useful, since the random images generating false hits aren’t just any random images, but ones that have visual features similar to the targets identified, even if we humans can’t see the similarities or even if they look like white noise.

Now that we know what computers are picking out as salient features, we can modify the algorithms to add additional constraints on what additional salient features must or must not be in an object identified, such that it would correspond more closely to how humans would classify objects. Baseball’s must have curvature for instance not just zig-zag red lines on white.

It sounds like the plan is for this ship to be the first of several, so the question is how much of that $20 billion investment is for upfront costs (design, shipyard upgrades, construction equipment) that will not be duplicated in subsequent ships. As it is, the first ship looks to probably at least break even or even make a decent profit (provided it works as expected) with bigger profits hopefully to follow. I am sure these numbers have been gone over very carefully. You don't make an investment this large on a whim.

have very low-cost consumer goods

BZZT. As the cost of labor approaches zero, the cost of goods approach the raw materials plus operating costs plus the CEO's pay plus the stockholders' dividends.

Versus the equipment to actually perform the desalination?

California has some pretty big wind farms and one of the issues with wind power is its availability when the grid can't accept the power. I wonder how much capacity goes unused and whether it would make sense to direct that power to a desalination facility that could provide a working load for the power in a scalable way that could be quickly and granularly spun up and down inversely to grid demand for that power.

The power the wind farms can generate but isn't absorbable into the grid is kind of free energy in a way and it would seem to make sense to do useful work with it like desal.

This sounds suspiciously like, "If you have nothing to hide, then you have nothing to fear." I'm not on board with that idea.

Just so I've said it, my comment wasn't intended to be in favor of law enforcement being able to search anything without a warrant, or without proper procedure. I'm just pointing out that the issue, as it's stated in the summary, is a legal argument about whether data lies in a particular jurisdiction. I meant to point out that, with the ephemeral quality of data and the ease with which it can move through countries-- even unintentionally, it might not be the best policy to make it as simple as "data is under the jurisdiction of wherever the physical hardware is that happens to hold it at the moment."

However, it might need to be that rule, because I'm not sure there's a workable alternative. All data is automatically under the jurisdiction of the location where it was created? Under the jurisdiction of the creator's primary place of residence? Under the jurisdiction of the primary location where it is most often accessed? I'm not sure I see how any of those policies would be enforceable.

Whichever jurisdiction it is decided to be under, law enforcement should be required to follow the laws of that jurisdiction.

Thank you, I'm quite flattered by the comparison.

Since California is near an ocean, there is a technological fix for that. Only problem is it would cost a lot of money : roughly between 22 to 44 billion dollars (between 579$ to 1157$ per person living in California).

On the upside, you won't have to deal with licensing issues for the operating system. I'm running into this with XP - Microsoft will not license new copies of XP and you can't downgrade a current version of Windows either. This makes replacing hardware difficult - even if the new hardware would work, if you can't get XP on it and the control software requires XP, you're kind of stuck.

Your examples are all crimes. Not knowing where your friend's house is located is not a crime.

I'm with you on this. STEM is a term that's being pushed on us by political/media types for who-knows-what reasons. It reminds me of people talking about "ya". I saw that thrown around a lot as a genre of books, apparently meaning "Young Adult (literature)", and it took me a while to figure out what the hell people were talking about. It's not really even a genre, but a classification of the target audience. It's pretty dumb use of jargon.

Back to STEM. Science, technology, engineering, and math. As though those are the same things. As though astrophysicists and programmers and marine biologists are all doing the same thing, and their expertise is interchangeable. Whoever lumped all that stuff together either has an agenda, or has no idea what they're talking about. In the contexts I see it being used, I assume that the intention is either:

(a) Companies that rely on software developers complaining about the lack of people with "STEM degrees", in an attempt to justify more H-1B visas; or
(b) Dimwitted programmers who want to lump all kinds of people into a subculture of "science people" to make themselves feel important. Like, "I'm a STEM person, just like Einstein, Neil deGrasse Tyson, and Carl Sagan. I'm just like those guys, because we're all STEM, unlike the filthy common people who like reading fiction and looking at art."

I mean, I'm not sure what else I'm supposed to take away from the whole STEM thing. Nobody talked about it 10 years ago. Having a lot of biologists does not help with developing software. Having lots of people capable of making iPhone apps does not push particle physics forward. I really think we need to drop the whole classification of "STEM" as a thing.

And the whole "preferably within the space industry"... what space industry does this guy mean? Does he want to work for NASA designing probes, or Boeing trying to design a space plane? Or is there some other "space industry". It'd be great to know, because it would really help narrow down what he'd need to do to accomplish that goal. But doing something related to space would probably mean, yes, you need to go back to school and get a undergrad in that particular field. Go find out what schools have the best Aerospace Engineering departments, and work your ass off, because that's going to get competitive.

I would say that it's not just the implementation, but choosing which thing to implement in the first place. A lot of these fads, whether it's "big data" or "cloud computing" or "agile development", have become popular because they're extremely useful in some cases. The mistake, sometimes, is in thinking that you've found a single solution to solve all problems, and applying it everywhere will fix everything.

Someone else here used the example of the language "Ruby" as a fad that was useless because Ruby is "awful". That doesn't seem right to me. In my experience, which is admittedly a bit limited (I'm not actually a programmer), it seems like different programming languages have their own strengths and weaknesses, so you may want to choose a specific language for a specific goal. However, realistically, in the projects that I've managed, it always made the more sense to take into account (a) the language any current code is written in; and (b) the languages my team is most comfortable working with. If you have a bunch of PHP programmers who only know PHP well, working to revise a web application written in PHP, then Ruby is probably a terrible choice. But then, Perl and C++ would also be terrible choices. Those aren't bad languages. They're just not the best choice for that particular project.

I don't want to start a shit-storm by talking about languages, since as I said, I'm not a programmer, but I think that example is simple enough. Similarly, "cloud storage" like Dropbox can be great for small teams working from different locations on small office documents. On the other hand, if you're a big company with tons of people working in a central office, editing video files that are multiple gigabytes each, then you're going to want some kind of internal storage. The issue isn't about implementing your Dropbox well, but making an appropriate choice for your needs.

Because you're being belligerent, and frankly, kind of dumb. That's "the reason why you shouldn't put things in the cloud". If it were the reason, then it could be easily fixed by having cloud providers give assurances that your data will be stored in a specific jurisdiction, and then you would have absolutely no reason not to put things "in the cloud".

But really my post is meant to indicate that there are some things about digital storage and transmission that needs to be considered, and the law might need to be revised to address any unclear points. In my mind, it's a bit like having copyright laws saying that you can't make any copies of copyrighted materials, ignoring (a) the possible need to make backups; and (b) the fact that, technically, copies might be made in various caches whenever you transmit or playback the digital file.

And by that comparison, I mean to point out that some things don't simply translate between the digital world and the physical world, so the rules have to be considered carefully. There may be issues that need to be worked out and clarified, even if it's only a common-sense judgement.

If you asked me to give a simple "yes" or "no" answer, right this minute, as to whether Microsoft is on the right side of things here, I'd say "yes". I'm just pointing out that it might not be something that you can decide simply without any unintended legal ramifications. Is that so hard to understand?