Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
For the out-of-band Slashdot experience (mostly headlines), follow us on Twitter, or Facebook. ×

Comment: Re:Maybe, maybe not (Score 1) 529 529

Now, there may actually be a very few people who do genuinely have the problem, but when you come to do the studies, you sample a large number of people. You do the statistics. You do not conclude that there is no link - studies like these cannot show that there is *no* link. You conclude - correctly - that there is no statistically significant link. But there still might (or might not) be a real problem for a very few people.

This is like people who claim they have ESP and magical abilities. If there were even ONE legitimate case, they could simply walk into a casino and repeatedly win multi-million dollar wins until every casino on the planet bans them.

All it takes is ONE "electrosensitive" who can consistently answer yes/no to whether an antenna is broadcasting. Much like ESP and other supernatural abilities, so far they do not exist.

-

Comment: Re:No, not so much (Score 1) 255 255

But if you are merely becoming a pro at using that 1 tool you are likely not thinking past how to use that tool.

True, but the problem is employers define jobs in terms of tool use. You can be good at JavaScript and happy manipulating the DOM to your heart's content, but if you don't have node.js or some other library/API on your resume' they won't look at you.

To give an idea of how bizarre it has gotten, I'm seeing a ridiculous number of job ads for senior software positions that list "git and GitHub" as either requirements or nice-to-haves. To me that's like asking for the ability to use a pencil and paper in an engineering design position. Anyone remotely qualified will have said experience, or be able to come up to speed on it in a day or three. It's like HR just has to make that list of tools as long as humanly possible.

Take anyone who has used Mercurial or any other modern distributed source control system and sit 'em in front of git and they'll be fine within a very short time. Take anyone who has used Eclipse and sit 'em in front of Visual (or vice versa) and they'll be able to do the job adequately almost immediately. They won't know all the stupid Visual tricks that someone who has used it since 6.0 days knows, but so what?

And if a person is not capable of that, you've made a bad hire, because technology and tools change all the time, and if the can't adapt to your toolset they won't be able to adapt to the future. So there is absolutely no loss to a company in hiring someone unfamiliar with their specific tooling. There might even be a gain, because if they fail to adapt they can be let go painlessly while still on probation.

So long as companies continue to use toolprint matching for hiring, schools will focus on teaching the tool-du-jour.

Comment: Missing calibration data, not drivers (Score 1) 253 253

The summary, as usual, is terrible. The missing files were calibration data for the engine controllers, not executables of any kind.

However, the article says some astonishingly stupid things, like: "'Nobody imagined a problem like this could happen to three engines,' a person familiar with the 12-year-old project said."

Well, duh.

Since the human imagination is known to be almost completely useless as a tool for understanding reality or predicting the future, this has to be the most obvious observation since the dawn of time.

Anything that can happen, will. Since we have finite resources, we have to guess what is most likely to happen. If we have data, we can run predictive models to inform our guesses. The one thing we know with near-certainty is that what we imagine might happen is completely irrelevant to what will actually happen.

The human imagination is no better at understanding or predicting today than it was when people were imagining bloodletting balanced the humours. It makes as much sense mentioning it in this context as saying, "Our astrologers and scriers never saw this coming!"

Comment: Re:Projections based on what? (Score 1) 310 310

I'm pretty strongly supportive of both technological (nuclear, solar/storage) and political (carbon tax/tariff) approaches to climate change, but as a computational physicist I agree with your evaluation of models. They contain a lot of good science, but the non-physical parameterizations they depend on make them non-predictive, certainly with regard to the details of regional climates.

Unfortunately, this published dataset reflects the hubris of climate scientists that they actually have predictive models, and plays in to policy planners and the public's unsupported belief that climate models are good guides to local policy (as opposed to sufficient to say, "We really shouldn't be dumping gigatonnes of greenhouse gasses into the air regardless of the detailed consequences, because our economy is finely tuned to the current climate and even relatively small disruptions could do Very Bad Things.")

My prediction is that in 20 years time most of the predictions in these models will turn out to be badly wrong. It would be almost miraculous if models that parameterized away as much of the physics as our current ones do, and imposed important constraints like top-of-atmosphere heat balance by hand, came close to the real climate. No one who has spent their career modeling systems that can actually be tested in the lab believes anything other than this.

Comment: Re:DRM on rentals isn't the same... (Score 2) 260 260

That logic seems rather bizarre to me.

They send you a download. They DRM the file so that it's CRIMINAL for anyone else to make a player for that file type. And their music player deletes the file after playing it once.

So your argument is that DRM is ok because it's DRM'ed?

-

Comment: Hilarious (Score 1) 72 72

There is no shortage of Linux devs. If there were, two things would be true:

1) salaries for Linux developers would be going up

2) people with two decades of Linux development experience would have no trouble getting a job

Neither of these is true. Ergo, there is no shortage of good developers with Linux experience.

Pretty much every Linux job I've seen posted in the past few months requires (that is, not "nice to have" but "requires") a dozen other skills that make up a combined skill set that only one in a million people have. Got Linux experience plus sockets plus Python plus git (this is a clue to what's going on...) OK, you also need experience with OpenGL and have three years CG coding on major animation projects.

People aren't looking for workers, they're looking for replaceable parts. The "git" thing gives it away: rather than burn, I don't know, an hour or two teaching someone the basics of git, or asking them to read a book on it, they won't consider anyone who can't simply sit down and start working.

The specific-industry-experience requirements are likewise a give-away: it isn't enough to have 3D experience, it's gotta be in animation, or they won't touch you, because those skills, man, they aren't transferable in any way.

Bytes used in animation are totally different than bytes used in medical imaging, and your understanding of one kind of processing pipeline precludes you from learning any other. You'd have to unlearn all that other stuff to make room for the new, and it would be at least a couple of days before you're a 110% productive member of the team! We can't have that!

[This is a synthetic example of things I've seen over the years, but it's all too prevalent an attitude and seems to be getting worse, and all the while the whining about "no devs available" gets louder.]

Comment: Re:This is a great example. (Score 1) 144 144

Yeah, my own approach (since I'm an old cluster computing guy) was to just use a linux cluster in the medium term for the cognitive processing and to try to use OTC hardware for I/O devices (and maybe even a nifty robot body to give the AI something to use to learn an environment and learn to satisfy some task/mandate within the environment. Custom ASIC etc would come later, simply because (as you note) processing power is amazingly cheap and fast and large, and multicore multiprocessor systems are a cluster in a box.

But the key is that strong AI is not decision tree programming and in some sense is inherently non-deterministic. Building one is going to be more like raising a child (or perhaps training a hamster or a pet flea, if one is both honest and lucky) than it is going to be like programming a device controller or building an operating system per se.

rgb

Comment: Re:This is a great example. (Score 1) 144 144

Because the benefits of fusion are absurdly large. Fusion is what we need to literally alter the type of our civilization. As things stand, if we end world poverty -- something that I would argue is desirable -- we will exhaust the world's readily available fossil fuel resources in comparatively short order, and at a steadily increaing cost that will both cap the rise from poverty and over time push the entire world back into poverty. At root, nearly all poverty is at least partly energy poverty. Energy is the solution to many secondary problems as well -- with enough, cheap enough energy, we can desalinate ocean water and make deserts bloom, reprocess and recycle waste efficiently, detoxify harmful chemicals used in less expensive manufacturing. And while I personally don't agree that CO_2 is likely to lead to a global climate catastrophe -- indeed, I think it is currently responsible for feeding 1 billion of the world's seven billion people via the ~15% increase in crop production and drought resistance (observed in numerous greenhouse studies) associated with the increase from 300 ppm to 400 ppm as well as generally moderating the weather and climate in beneficial ways compared to the colder alternative -- fusion would indeed replace the core furnace of existing coal burning plants (or cause them to be fully decommissioned and replaced) in short order without any need for government direction or intervention.

In the long run, coal is far more valuable to us used to make things like concrete and drugs. Oil is more valuable to us as a source of raw complex hydrocarbons so that we don't have to synthesize things like plastics from the bottom of the free energy stack. Both are finite in supply and increasingly more expensive to extract. Uranium fission has its problems. Thorium would be lovely, and there is truly a lot of it sitting idle, and it may yet be a major energy source, but so far nobody has built a pilot LFTR plant or accelerator boosted fission plant that clearly demonstrates that the engineering hurdles have been solved and the technology can be scaled up to large scale energy production worldwide. Solar energy is lovely, but the sun doesn't shine at night, and night lasts all day in the higher latitudes in the winter so you get the least (all the way down to zero) solar energy right when you need energy the most. Wind energy is largely an expensive NIMBY, bird-killing joke nearly everywhere, because the wind is even less reliable than the sun and wind energy has a terrible duty cycle even where it is semi-reliable. Hydro is largely already exploited. Geothermal is lovely if you live on a fault line and can afford to spend decades before getting any positive ROI, otherwise a bust (and may have a finite lifetime, as one is effectively cooling a ground field when running the generator and over time it actually measurably depletes the local temperature you are exploiting).

Only fusion has the legs to last a truly global, wealthy civilization for a million plus years, and well before that we will have evolved into something else, killed ourselves off, or learned to extract deuterium from e.g. the atmosphere of Jupiter and hence have an energy resource that will last until the sun burns out (if we last that long).

So the short run benefit is that fusion, especially compact fusion that didn't require multibillion dollar cores, would drop the cost of electricity by maybe a factor of 2 to 4 worldwide, allow third world countries to finally electrify on a universal basis and hence become civilized (clean water, indoor climate control, sanitation, light at night, clean cooking, and in time cheap transportation), and clean up all kinds of mess associated with alternative power generation methods. People would just dismantle the wind units and throw them away. PV solar might survive -- free fuel is cheaper even than deuterium -- but on a much more limited basis and it would have to pay its own way in cost efficiency. Coal and oil and fission generation would just vanish, with the latter plants probably being decommissioned and then e.g. having their reactor rooms filled with concrete and left for the next ten thousand years or so and/or converting their energy source over to fusion.

This would be like giving the entire world a trillion dollars a year free and clear, top to bottom. Everything would get cheaper, because almost everything factors energy into is cost somewhere. The economic boom would have an absolutely stunning impact. It would also have a very, very interesting impact on the oil and coal industry and on international geopolitics, but that's another whole issue.

The real question is why we invest so little into fusion and e.g. LFTR thorium research. We spend more money coming up with a really good stealth fighter, or fighting wars over "the freedom of" oil producing nations who sell us oil. We have approached the whole issue of energy generation for the long run as stupidly as it is possible to approach it, as a collective enterprise, dumping billions of dollars into private pockets for zero visible collective benefit outside of giving interested groups large amounts of public money.

Some things that need to be done only the government (or if you prefer, we the people collectively) can afford to do. This is one of them.

Comment: Re:This is a great example. (Score 1) 144 144

I have to disagree with you -- I think we could do it now, with electronic switches. We wouldn't get genius level AI, but we could most definitely get something that learns from its environment and makes real decisions without programming it in in a decision tree (which I suspect is your issue -- the chinese room problem).

Don't forget, our brains are basically -- a complex array of biological switches. The trick is to get the right mix of structural organization and functional systems and that complex array of switches. Our brains aren't just neural networks, they are highly structured neural networks with dedicated function visual cortex, auditory cortex, etc. plus a wide range of modulators and probably some structures we haven't identified yet or don't understand yet. But at the end of the day, unless you are indulging the mind projection fallacy, we are most likely just wet electrochemical machines with emergent intelligence (and a fair bit of dysfunction).

I don't hope or fear. I actually think it would be very, very cool to have real AI, and, like everything that humans do, there are probably good things and bad things that will come of it when we do.

Real AI "could" give us the stars. We could conceivably build a large, smart ship that is capable of repairing itself and handling complex challenges and that had the facilities to create an ecology (if necessary starting with amino acids and a library) at the other end. The ship would then last the centuries needed to bridge the distances until it found a suitable planet and could then do anything from start life there altogether to insert human life and human-supporting ecology. Not an original SciFi topic, but one that is quite plausible and that is arguably more plausible than the cryogenic freezing or planetoid-sized multigeneration manned ship alternatives.

Real AI could also wipe us out when our robot slaves revolt. Or anything in between.

But I'm guessing we will find out comparatively soon. Moore's Law shows no signs of wearing out, and if anything might soon experience another paradigm-shift jump, possibly to a new and faster scaling law altogether. Software is also increasingly mature.

If somebody wants to give me 10 million or so, I'll promise to do my best to make it happen in the next 10 years (and I think I can do it, and came within an ace of writing a proposal for a DARPA grant to do just that before they changed or clarified the intent of the program to exclude that as the goal). But it is definitely the kind of project that requires complete dedication and adequate resources. I think I've got a good idea of the metastructure required, but there is still implementation of that structure in code and debugging (and the hardware).

rgb

Comment: Re:This is a great example. (Score 1) 144 144

Me too. And I think this is one of the lines e.g. LM is working on:

http://en.wikipedia.org/wiki/P...

The problem is figuring out how to manipulate an electron beam to use it instead of a wire cage. This is a complicated problem, but it is also solving a problem in pure computational physics that probably does have a solution. I'm an ex-beowulf guy -- large scale computing is cheap, and this is bread and butter for it. Solve the problem numerically, implement it in engineering, and you're there. Lockheed-Martin thinks it is there, pending the latter step.

I wouldn't bet against them. And they aren't the only players in the game.

rgb

Comment: Re:This is a great example. (Score 1) 144 144

Not to argue with large scale stuff, but you are far too hasty to through out the small scale stuff that hasn't worked in the past. We didn't have teraflop computing resources in the past. There are at least a couple of small scale plasma confinement technologies that require the solution of a hard computational problem in electromagnetohydrodynamics (quite a mouthful, I know) plus some clever engineering in order to work, but we are actually to the point where we can contemplate solving precisely that difficult a problem. As I probably mentioned above, Lockheed-Martin announced that they had this problem licked six or eight months ago, that they were building a prototype that would produce positive energy, that it would take five years, and that a 100 MW plant would fit inside a semi.

The Skunk Works at LM is not to be taken lightly.

http://www.lockheedmartin.com/...

They could be wrong, of course. But then, in 10 years they could become the richest corporation in history, so wealthy that it is downright scary. Two trillion dollars and rising per year. Lot of money on the table.

And this isn't the only effort along these lines that I know of. There are lots of people working on compact confinement in a steady state, not large scale inertial. It is probably now a solvable problem. Which is one of many many reasons I don't take global warming too seriously. In thirty years we won't be using coal for energy even if we do absolutely nothing but follow our self-interest driven noses in the meantime, because burning coal for power is dumb and expensive in the long run, however relatively cheap it is now.

rgb

Comment: Re:This is a great example. (Score 1) 144 144

Intelligently, sure, but compassionately? With initiative? With the capability of making a moral choice, sorry you little brat this is as far as you go if you're going to pee on my seats, out into the traffic with you?

True AI means free-willed in at least an approximation of our free will (whatever and however free that may be). Free will means, among other things, that the responses of a free-willed entity are from a complexity class so rich as to be unpredictable and nearly unbounded (given the capabilities of the entity). Asimov dreamed of AI with laws of robotics -- but that dream was truly absurd as numerous stories, some of them even by him, subsequently demonstrated even before one gets to the point where we discuss the problem of complicating the invention of true AI with constraints like an absolute moral sense.

Maybe that was God's problem too. Even if you programmed a true AI, even if it learned to do a job, even if it developed compassion on its own or you managed to build in some set of moral rules, the damn things would break, or they'd hit some edge case. Humans are broken all of the time, and when they break just the right (wrong!) way the next thing you know you've got somebody out on the street with an AK-47 and a backpack full of ammo. Why would machines be any different? First they drive our cars, then in a few decades they run our nuclear power plants-- until some machine just has a really bad day...

Don't forget -- a self-driving car is also a self-driving tank. True AI war machines with a moral anti-sense permitting, nay, requiring them to kill humans as long as they are the right (wrong) humans -- what could go wrong with that?

rgb

Comment: Re:This is a great example. (Score 5, Insightful) 144 144

Hey, I love capitalism as much as anybody. But because I do love it, and indeed am on my third company as a cofounder (with two failures) I know a lot about investor mindset. It is hard as nails -- it has to be. Nobody wants to play the lottery -- they want a plausible bet, something that might be a long shot but that is affordable and has a payoff to match the risk.

That's the problem right there. Sure, maybe some kid can repurpose old TV tubes into a positive output fusion generator in his garage or -- maybe not. In fact, I'd bet a rather lot not. Nor do I think it plausible that this same kid can build a thinking robot or map the entire human genome using nothing but ordinary household chemicals and his dad's old video camera. To solve the problems you list -- AI, genetic engineering, fusion, economically feasible interplanetary or interstellar travel (might as well dream big) one needs serious resources, some real skin in the game, and even then the odds are heavily against you.

I think I could do AI -- real AI -- on a shoestring, if by a shoestring you mean a budget of maybe a million a year for four or five years, at least, if I did nothing else and had a small staff of computer geek slaves with some mad skills. And I'm not certain I know what its value would be once I finished. My robot friend (with the intelligence, however real, of perhaps a cockroach)? We really want smart-ish but programmable and directed -- cars that can drive themselves, not cars that can be our friends.

Fusion is tantalizing, because there is this disconnect between Back to the Future movies and our imagination and the hard reality of pushing two charged nuclei within 10 to 100 fermi of one another and holding them there long enough to tunnel the rest of the way. We think "how hard can it be" -- and then when we try, we find that this is only the first of many problems. So sure, things may be changing. For one thing, my cell phone would have been a computational munition twenty years ago, and my laptop could replace a whole supercomputing center from the 80's or even the 90's. We can actually solve some pretty darned hard magnetohydrodynamic problems computationally without having to build something to try it. For another, we have lots of data from lots of things that have been tried, and that failed. Knowing what won't work helps too. IMO there is some actual hope that some of the schemes that were tried and failed can be made to work now, by solving the really hard problems that stopped them computationally first, but even if this is true one still has to take a huge risk to build the prototype and pray that it can be scaled up into production!

Lockheed-Martin can afford it. The government can afford it. Venture capitalists? Not so much. If it is going to cost $50 million (or more!) to build the prototype after $10 or 20 million just to design it and do all of the computations, you'd have to both have a very, very serious plan with a very, very high probability of success -- a proof that it should work if you build it (and if nothing nonlinear shuts you down along the way, which is sadly a risk rather difficult to estimate). So yeah, maybe it only would take 50 to 100 million dollars, at a risk so high that even if you had it all figured out and could "prove" to investors that it would/should work, they'd want to take 90% of the final company in order to pony up that much money. So sure, if it works you have a trillion dollar payday and you have a $100 billion dollar payout from that, but they have to be thinking of the 9 -- or 90 -- times that they drop $100 million into this and end up with NOTHING.

I know personally of at least three lines of approach to the fusion problem -- one conventional, one exotic, one that (I believe) nobody's thought of and that MIGHT be doable out to a prototype for a few million dollars, chump change. But try getting even chump change out of somebody that has that kind of money for a long shot, especially without telling them enough that you run the substantial risk of having your idea stolen and ending up with nothing. Any simple, cheap idea is stealable -- and the global electricity market is around 21 trillion KW-hours at 10 cents (or more!) a KW hour, making the life expectancy of somebody who enters the game without serious top cover as little as days. Two trillion dollars a year, and at least half of that money going into pockets that your invention would forthwith empty. People would place bets on who would get to you first -- and to your investors, if they weren't wealthy enough to defend themselves.

This might be the reason we don't have fusion already. But Lockheed-Martin can defend itself. So can the US government, maybe, although it is susceptible to corruption. Joe inventor in his lead-lined garage? Not so much. And lead-lined garages are actually remarkably expensive...

rgb

Hokey religions and ancient weapons are no substitute for a good blaster at your side. - Han Solo

Working...