What would you study ?

How to build automation to replace the other 50%?

There are plenty of community/city/state schools that are dirt cheap (e.g. 2-3k per semester, 4-6k a year). Also, average student debt in the country is ~$25k or so, I'd hardly call that a 'lifetime of debt'... (yes, fresh graduates can't pay it off in a year, but that's why they're not all due in a year). And if you work on the side (yes, tough, many folks had to do this) you can graduate with no debt at all. And then there's financial aid (if you *really* cannot afford it).

Now, if you're getting an acting degree from NYU and end up with $150k in debt and no job... well then, that's just poor decisions right there.

It works the other way with medical degrees. Yes, the degree is "expensive" by `normal' standards, but a doc is expected to earn 2-3x that debt per year... so dollar mounts a skewed.

The cameras wouldn't be there to help in *that* incident. It would be there to help document and train for situations that might happen on future flights---situations that may not be as clear cut at this particular incident. The argument that it would not improve aviation safety is silly...

I can imagine a situation where Lenovo (or Dell, etc.) keep it unlocked for their "business" customers (e.g. Thinkpad line), and lock for everyone else... In a few years, that would pretty much kill off linux for anyone who casually wants to try it out on their then-"old" laptop.

Very likely the reverse.

I'd imagine that most accidents involving automated cars will *provably* (video and telemetry info and all that) be human operator's fault (or the other driver)... suddenly humans will find their insurance go sky high, while insuring a self driving car will be dirt cheap (they'll be harder to steal too).

All your concerns are valid, BUT, do you really think that "average human driver" makes the right decisions that much better than the potentially *random* behavior an automated car will display in all these extreme scenarios? Yes, lets say an automated car runs over a child (and saves the dog)... but do you really think the "average human drive" would do any better???

My guess, automated systems will prove to be several orders of magnitude safer overall than current human operators... there will still be accidents, but they'll be much more rare (and perhaps much more deadly, but if accident rate goes to 1% of what it is now, that would be huge---so much so that human driving may actually be banned on most streets).

Kind of like "most car accidents involving trucks" are *not* caused by the truck driver mistake... the future automated car accidents will probably not be caused by computer error, but by someone being stupid around one.

Imagine everyone staying in lane, maintaining speed, distance, etc. And actually driving the posted speed limit on city streets (even automation wouldn't have much problem slowing down from 25mph for a child (or dog, or tumbleweed) on the road).

Laser Diode Arrays with remaining eye.

In that case, R wouldn't help them either. You really need to understand how you're solving the problem before you throw some R at it and suddenly get an answer.

And the Earth Stood Still.....today.

Is it resistant to gin & tonic?

I think one of the reasons Amazon's phone failed was because it was tightly coupled with the amazon echosystem and not the google echosystem---the same exact phone sold by "google" [e.g. marketed as "nexus" line] (even at the same price) would've done MUCH better in the market. It's not just "uh oh, you're bundling your services with the apps"... it's that people actually *want* those apps and services and often wouldn't buy the device otherwise.

Also, plenty of manufacturers roll their own Android, so what are they complaining about? If you don't like it, recompile your own and convince folks to use it.

It's not 'using', it's quite bit of updating (e.g. greenplum was essentially postgresql updated to work on multiple nodes).

I don't think anyone is raising up the "using" part as being inherently bad---even stuff compiled with GCC is "OK" by most standards... It's taking open source and recompiling it into your closed source product that GPL is objecting to (and it's perfectly fine with some open source licenses... just not GPL).

Try http://en.wikipedia.org/wiki/G...

It's a multi-billion dollar industry out there, built on open source... that's currently completely closed source. (they're not the only one---just about all commercial closed source software is built on top of something).

current depleted uranium ordinances aren't railgun-line-of-sight deals and yet still cause enormous damage... It doesn't have to be going *that* fast to do damage (e.g. take a few pounds of aluminum and send it at a ship at say 2x the speed of sound, and it will do lots of damage even without any explosives---and yet still be traveling in a parabolic curve).

I'm a huge fan of Perl... Perl5 that is.... I'm just hoping they don't screw up the language with 6. Perl5 works great for me---the few bits of Perl6 that I've seen look akward :-/