In 1998, Kevin Warwick implants a trivial RF resonator in his arm (the sticky plastic strip that warns Walmart that someone is stealing a pair of socks). He contacts the press, calls himself a cyborg, and gets tenure at Reading U.
In 2001, he replaces his implant with an RFID tag, and calls the press again, and says, "Look at me, look at me! Now I'm an ACTIVE cyborg!" And he becomes a full professor at Reading U.
And now Warwick gets 300,000 neurons to produce a simple binary response (go straight vs turn). He calls the press again, and says, "Look at what I can do just by waving my arm".
Jackass. Worse, the media can't tell the difference between poseurs like him and real scientists like Sebastian Thrun or Rod Brooks.
"It is not impossible to build a human brain and we can do it in 10 years."
I'm relieved to see that tradition of excessive and unjustified optimism by yesterday's AI crowd is still alive and kicking. It's just too bad that a self-aggrandizer like Markram is using it to give false hope to sick people.
Announcing the arrival of a man-made human brain in 10 years, without first having built anything but simulations is so irresponsible and ultimately hope-despairing that there damn well ought to be a law.
Better yet, ask Markram to sign his promise in blood -- if he fails to deliver, he donates his own brain to the cause. His ex-corpus cerebrum may do little medical good, but at least it'll do no more harm.
I've lived in MI (Midland, Kazoo, E. Lansing, Ann Arbor) for half of my life (50 years), and have a love/hate relationship with her. Other than the six months of gray skies (Nov-Apr), it's the most livable place I know (vs. PhillyPA, WashDC, SouthernVA, ChampaignIL, & HoustonTX). I'd return to A2 in a minute if I could find interesting work there.
One more bennie Michigan has that the OP forgot to mention:
9) SLASHDOT
Randy
Poor intro but good book
on
Wired for War
·
· Score: 1
I've read the first few chapters and other than the lame introduction, the book is quite good. I haven't read any discussion yet of the philosophical implications of automated warfare, though I hope there will be (and not just from professional 'ethicists'). Mostly the author discusses the pragmatics of building ever smarter weapons and intelligence systems, which is mighty relevant to the present, and most assuredly the future of war.
Even when the author doesn't use the word "robot", the trend he describes is clear. Tomorrow's war will involve ever increasing amounts of separation between finger and trigger. Is this a bad thing? Obviously I want our side to win and I don't want any US or civilian casualties. But I also don't want war ever to seem as easy as Bush, Rumsfield, Cheney, and Wolfowitz assumed. Automatic War is a sure recipe for an infinite series of mindless incursions and body counts. This book should serve as a wake up call to the many aspects of war-by-wire, from technology to policy to ethics to cost. Going to war must never devolve to the pressing of one red button.
I'm 50, earned my MS in CS 20 years ago, and have worked in CS R&D for the past 20 years.
Here's why a MS in CS is a good thing:
1) You can take 10 more courses and learn 10 more subjects that you didn't have time for during your BS. Examples: intelligent agents, robotics, machine learning, compiler optimization, virtual machines, advanced graphics, image processing, signal processing, parallel computing, quantum computing, microarchitecture, computer security, computational biology, advanced networks, mobile computing, pure math, applied math, advanced statistics. For many employers, the presence of the right combination of courses in your resume can be very important, and very likely will make the difference in preferring you to another job candidate.
2) Employers value credentials. When it comes time to promote the next staffer into management, they will consider your education. In the first decade of your career, a MS will definitely help you get the more interesting assignments or step up into management.
3) You are less likely to be at a competitive disadvantage to your coworkers. For the first 5 years of your career, almost everyone in the company will have more work experience than you. If you also know less than they do academically, then you lose.
4) Everyone values a MS more than a BS. Knowing more is always better than knowing less.
5) If you want to work in R&D, don't. Not without a PhD. That said, the (optional) research component of a MS may have more value to a R&D shop than would the alternative -- two additional courses. The thesis shows your ability to do (some) research. That said, most non-R&D employers would prefer the courses anyway.
If it's a pipe dream, then why do so many researchers and physicians regard iPSCs as a holy grail? Probably it's because stem cell therapy has *already* repaired damaged tissue and restored function to a variety of tissues in mammals and humans, including the spinal cord.
In case your tragic state of perplexity becomes too much to live with:
"Directed differentiation of human induced pluripotent stem cells generates active motor neurons"
S Karumbayaram, BG Novitch, M Patterson, JA Umbach, L Richter, A Lindgren, AE Conway, AT Clark, SA Goldman, K Plath, M Wiedau-Pazos, HI Kornblum, WE Lowry
"The potential for directed differentiation of human induced pluripotent stem (iPS) cells to functional post-mitotic neuronal phenotypes is unknown. Following methods shown to be effective at generating motor neurons from human embryonic stem cells (hESCs), we found that once specified to a neural lineage, human iPS cells could be differentiated to form motor neurons with a similar efficiency as hESCs. Human iPS-derived cells appeared to follow a normal developmental progression associated with motor neuron formation and possessed prototypical electrophysiological properties. This is the first demonstration that human iPS-derived cells are able to generate electrically active motor neurons. These findings demonstrate the feasibility of using iPS-derived motor neuron progenitors and motor neurons in regenerative medicine applications and in vitro modeling of motor neuron diseases."
Subscription to Wiley Interscience required for more...
What fraction of this 2/3 already has access to broadband but has rejected it? If the fraction is high, then the 2/3 statistic is obvious pretty much a tautology. The 2/3 includes everyone who *could* subscribe to broadband but rejected it. Meaningless.
The real question is, what fraction of people who CANNOT sign up for broadband also do not want it. I doubt most rural folks don't want better communications technology. I lived in central Illinois for a while and the dial-up Internet community was thriving. It gets pretty damned lonely out on the prairie.
Trust the press to pass along an ambiguity as if it were news...
After 25 years working in software, I discovered a few years back that programming the same old thing in yet another language just didn't do it for me any more. I need to work on new and different kinds of problems. After a few early years in commercial programming, I wandered into R&D where I've worked on AI systems for the military, then supercomputing problems for the government and then for scientists and engineers in academia. Lately, I've found myself analyzing medical images as part of drug design. In addition, I've become intrigued by vision systems on robots as well as computational biology, perhaps elaborating gene networks or modeling disease. Maybe that's where I'll go next.
However, I'd recommend a different path than the one I took (part-time graduate study ending in a MS in CS). To work independently in R&D, you'll want at *least* a MS, preferably a PhD, with a full complement of math courses through diff eq, statistics, and possibly PDEs. To work in engineering areas like communication or imaging, courses like signal processing, image processing, and perhaps 3D graphics will also help to open doors.
Here are a few more applications for computing that sound like fun:
- graphics, game design/implementation - networking, new services from cable companies/telecoms, new wifi products - secure systems, computer forensics - embedded systems (real-time O/S, low-level software or high-level hardware) - robotics, semi-autonomous systems - hand held computing (IMHO, soon to be much bigger) - data mining, social network analysis, machine learning the datasphere - searching through or indexing of media or large content (e.g. web, communication streams)
Personally, I'd stay away from building low-level software like compilers and O/S's. This kind of work is likely to be outsourced by the heavy hitters like Intel/M$, since the needed skills are readily available in low-wage countries, and most of the infrastructure is open source (VCs seem to believe that its hard to make a profit by attaching your IP to free software or competing directly w/ the tech oligopolies).
I agree, passion for producing a better solution which is smaller, faster, better focused, solves problems you didn't anticipate, or does more than anyone ever hoped is *very* much a trait I want in the people I work with. Without it, you're surrounded by 9-to-5-ers.
However, passion for a single technique/tool/philosophy is not really 'passion', but 'dispassion' for alternatives. And that's bad, especially since in computing, there are so many ways to skin a cat, and you need to work as a team whenever possible. Often that means adopting someone else's passion and making it work.
I grant you, the current grading system is clueless and it overemphasis demerit over merit (you feel more like you've lost points than earned them). But the Pittsburgh Model just shifts the grade curve downward:
A is 80%,
B is 60%,
C is 40%,
D is 20%,
F is spelling your name wrong.
.
A far better grading system is perhaps the simplest: rank the class. Then there can be no debate about grades because there grades don't exist.
Rank is immutable. It eliminates politics and grade inflation altogether.
In my opinion, easily the best way to advance *any* technology is to draw a line in the sand and then challenge others to cross it, either through competition, or better yet, by surpassing the status quo in ways that are quantifiably 'better' than before.
DARPA has achieved some of its best bang for the buck in its recent robotic grand challenges, significantly advancing the field of mobile robotics and doing it with great fanfare. The fact that the progress is visible and practical also makes a world of difference. Look at the popularity of battling (mindless) robots.
Therefore, if DARPA wants to advance other forms of AI, I suggest two action items:
1) Create other grand challenge competitions. Robotics is a nice dynamic medium for demonstrating AI advances and overcoming real-world obstacles. Thus other forms of robotic competition might also be interesting, for example: mobile agent coordination, as in a strategic-driven variant of Robocup, or coordination of ground/air/sea forces, perhaps at battalion level -- a fairly natural application of Future Combat Systems development).
Of course, nonsituated agents also have potential. Competition, conversation, and collaboration among software bots would allow many folks to compete who don't have the resources of a CMU or Stanford. And SoftBots would also better explore other forms of AI, like speech/NLP.
2) Define some standard performance benchmarks (speed and/or accuracy) so that professionals can have a target to shoot for. Each of these should measure a different form of AI like speech & NLP, planning, vision, pattern recognition (like USPS character recognition), etc. These might also include metrics derived from the competitions in action item #1 (e.g. sporting events, like slaloms or gymnastics routines).
I suggest that DARPA launch a Request For Information to the AI research community for suggestions on AI benchmarks and competitions.
As practically useless as the Turing Test has been, it has garnered a disproportionate amount of attention over the years merely because "it is there". But AI research needs to walk before it tries to run. What better way is there to do that than to compete, not with humans nor even simpler life forms, but with other "AIs"? Perhaps the Turing Test should become Turing Match Play, with the winner being not "which is the man", but "which is the better man". (Or woman.)
But Octave supports little more than the basic syntax of Matlab and omits most of its basic functions and all of its GUI (e.g. no graphing).
As a better alternative to Matlab, I'd suggest SciLAB (http://www.scilab.org/). Not only does it support a much larger fraction of Matlab than Octave (including a lot of functions from the toolboxes), it's actively under development and actually has a GUI.
In fact, SciLAB might be exactly what the Original Poster wants -- a flexible matrix-based programming language with some visualization primitives, that's 98% compatible with a standard programming language to boot.
3) Problems that are often interrupted, that cannot predict future actions or execution paths (e.g. pipelined)
What fraction of computing tasks match these 3 constraints? At least 90% of the work done on desktops.
Speeding up the other 10% will be done by specialized hardware like multicore video or DSP chips. The real potential for parallelism in everyday computing is negligible. I've been parallel programming for years, and all forms of parallelism are next to useless unless, like Google, you have lots of data.
BTW, Google has no use for multicores either. All of their parallelism is embarassingly parallel, which is better served by shared-nothing architectures like many, many, many cluster nodes that are cheap, cheap, cheap.
That's because Derrida is already incoherent. No smiley.
There's no question that philosophy isn't trivial and that some philosophers are more sensible than others. But NO form of literature or social sciences requires anywhere the effort of real analysis or quantum mechanics. Nobody with comparable skills in the two worlds would assert that the effort or homework in the competing curricula are comparable either.
Like proficiency in math, proficiency in writing is not for everyone. I test as being comparably good at both, but I find reading/writing/logic a great deal easier than proofs/derivations/transforms. I'm pretty confident I could produce a quantitative measure of the difference in effort if needed.
I've always thought that there should be a different criterion for entry into Phi Beta Kappa from math-based curricula. I'm convinced the effort expended by physical scientists and engineers cannot be compared equally to degrees based on reading fiction or history.
I once completed 80% of a BA in psychology, and the effort level was nowhere near that of a CS curriculum, much less engineering.
... 6) UNREAL quantities of homework (all math problems) 7) Textbooks that are often badly written/incomprehensible/irrelevant/error-laden 8) 3/4 of professors who speak English poorly 9) 3/4 of professors who drone on and show no enthusiasm for the subject 10) WAY too many equations written illegibly, rapidly, BY HAND on whiteboards 11) Lectures that, way too often, are not available as handouts 12) Handouts that, way too often, are not available before the lecture 13) Material that often feels irrelevant to the real world (esp. the way linear algebra is taught)
As a long time computer scientist who is taking a 400-level course in image processing (my first EE course), I much better appreciate how damned hard engineering is. By contrast, CS is a breeze, and the humanities are about as hard as watching TV. Now I remember why I traded in a BSEE program for a BS in biology.
In order for the rights of the people to be abused three things must exist:
1) The technology enabling the abuse must exist. In this case, one or more satellites are already in place to spy on US. Strike one.
2) A policy must exist which legally enables the owner of the technology to misuse it. So far, this does *not* exist. Ball one.
3) The people in a democracy must allow the policy and the technology to be abused. Therefore, if Orange Crush speaks for the rest of us, this requirement is already in default. Game over.
Despite all the rhetoric to the contrary, peoples governed by a representative government don't lose their rights, like "Orange Crush", they give them away.
The solution is clear. Ignore people like Crush who never wanted freedom in the first place. Be an American and vote these assholes straight to hell.
In school, you need to learn the basics first. Complex/compound/swiss-army-knife languages like Java and C++ should NEVER be first languages. Nor should they be used to teach any other topics in which abstraction would interfere with principles (e.g. systems, algorithms, theory, perhaps even AI).
Ideally, the programming language used to teach a course should support only the core concepts of the subject. It should avoid all else. That's why Pascal and Scheme are probably still the best means to introduce the basics of procedural and functional programming. Sure each language has weaknesses, but the goal in CS is to learn *principles*, not *languages*, and these two perform that role economically and effectively.
Should CS students learn Java or C++? Yes to the former since its today's lingua franca among professional software engineers (and probably no to C++, because it's requires the programmer to overcome far too many interfering semantics when implementing even simple tasks). But Java should still be introduced only when teaching principles of software engineering, not computer science.
One size will never fit all in parallel computing. There's such a diversity of range (hardware), domain (application space), and user need that a plethora of parallel tools and design choices will always be needed.
SMP machines differ significantly from tightly-coupled clusters, or even loosely-coupled clusters (a la Google), so a range of parallel tools are needed. Some parallelism components inevitably come from language vendors and others come from O/S vendors. As such, programmer access to these tools may need to remain at a low level (perhaps for direct device access) or may rise to a higher level as part of a more abstract programming model.
What's more, tasks go parallel for different reasons: greater speed, greater memory, more responsive GUIs, or to make for a software architecture that's better suited for various data structures like workpiles or trees, etc.
With such a diversity of architectures, multilevel implementations, application domains, and designer needs, it's little wonder that different solutions have emerged. Given the four dimensions and 2-3 variations that I've described for each, I can see justification for at least 2^4 different parallel programming models, tools, or languages.
A final thought. If each of the popular programming languages were extended to support even one form of parallelism (perhaps SMP threading), in addition to the numerous parallel process libraries that exist, how many *more* solutions would then arise? Can we really expect language implementors not to extend their products in order to serve the needs of their current users?
I think it's inevitable: the babel of parallel polyglots is here to stay.
Moore's Law describes a CPU speedup that died at least 3 years ago (all other legalisms aside).
To wit: I bought a laptop in 2003 with a 2.2 GHz 32 bit P4. According to The Law, by 2005 CPUs on comparable laptops should have run at 4.4 GHz, and by today they should zip along at 8.8 GHz. But in fact, no commodity CPU runs at that speed nor even *half* that speed.
And don't you believe the claim that multicore or power throttling compensates for or explains The Law's "failure to thrive". The fact is, the industry is no longer delivering CPUs whose SPECMarks/FLOPs/etc (AKA performance) is rising at the rate that they have for the previous 20 years. I tell you, "Moore's Law is pushin' up daisies. It's a DEAD parrot."
What's puzzling to me is that while this emperor is clearly naked, for some reason, nobody wants to admit it. Why not? Are we afraid that sexy soothsayers like Ray Kurzweil or Rod Brooks will be regarded laughably when they forsee cool stuff like The Singularity or robots possessing human-level cognition, brought to us by the inexorable exponential march of Moore's Law? Or do we simply dread the day when we have to depend entirely on advances in *software* to deliver our next high-tech fix? Perish forbid *that* thought.
Well, we'd better get used to it, the emperor is naked *and* dead. There's a new emperor in town, and Moore's Law 2.0 depicts a future that looks a hell of a lot like the past.
Not at all. It'd be trivial to add a button to future preamps/amps that does the same thing for CDs that the narrow/wide toggle did for FM stereo years ago -- namely compress/expand the dynamic range.
Of course, compression could be introduced when burning the MP3/AAC, which is the better solution anyway. Compression makes less sense for music that is played around the house (e.g. CDs). It makes more sense when the listener is on the move -- in the car, ipodding, etc.
If two versions of each song were to be distributed, I'd prefer they be built into the media: disc vs. audio file, where only the MP3/AAC gets compressed by default.
A final thought. It may be that if CDs are to be compressed by default, the practice may only increase the demand for new music format variants (like the various bit rate versions of a song available now from iTunes). Perhaps this would add yet another MP3/AAC/AIFF format -- perhaps an uncompressed SACD-quality binary. Now that's a compromise that an audioprig like me could live with.
"Academic" vs "professional" is a fatuous distinction. By those terms, any discipline that is useful outside of the university is less deserving. Should we respect such degrees less, or charge more for them (irrespective of their instructional cost) because they have commercial value?
There is nothing inherently less challenging, enlightening, or "worthy" about purposeful studies than purposeless ones. In fact, the fact that a subject has utility *must* be the single most important criterion by which any field of study should be judged... inside or outside the academy. Otherwise, irrelevancy takes precedence over relevancy, and publishing "A Critique of Humor in The Simpsons" becomes preferable to stopping a war or curing cancer.
It's not only the engineerings and business, but medicine, chemistry, geology, clinical psychology, laboratory-based biology -- all of these studies are both applicable In The World and costly to acquire. The pure science degrees then fall between the cracks, perhaps immediately useless yet grounded in useful principles: physics, mathematics, and perhaps economics and non-laboratory biology. Should we charge more for their tuition, not because they cost more to teach, but merely because employers might value them?
If so, your notion of a worthwhile education certainly differs from mine.
I work for a big pharma, and I very much would like to see our research improved by the use of computers. Not only will drugs make it through development faster and cost less, but fewer animals will be sacrificed during testing (a pet peeve of mine). Clearly we all benefit by making better use of the data generated in the lab.
As to whether in-silico lab work is feasible, they jury is out. It may be that computers will be most useful as a complement or augment to lab techniques. It's indisputable that computers are central the use of most lab instruments today, from calibration to DSP, image processing & analysis to statistical analysis of the data they generate. The next question is whether computers can generate useful information by modeling the real world. That endeavor is in its infancy, but it'll get better, specially with improvements in extracting data from lab instruments, which again, will be accomplished by computers.
Should we draw a hard line between the use of computers on the supply vs the demand side of science -- in collecting and organizing data from instruments, but not exploring the possible interactions of that data using simulation? I don't know why we should. There's a natural feedback loop there, where the data guides further exploration. Computers are central to both sides of that equation.
I want to extract all the information that's out there and then make the best possible use of it. Science exploration is not an either/or proposition. The problems are hard enough that you try *anything* that improves your odds of success. Then you try it again.
Slashdot Mirror
In 1998, Kevin Warwick implants a trivial RF resonator in his arm (the sticky plastic strip that warns Walmart that someone is stealing a pair of socks). He contacts the press, calls himself a cyborg, and gets tenure at Reading U.
In 2001, he replaces his implant with an RFID tag, and calls the press again, and says, "Look at me, look at me! Now I'm an ACTIVE cyborg!" And he becomes a full professor at Reading U.
And now Warwick gets 300,000 neurons to produce a simple binary response (go straight vs turn). He calls the press again, and says, "Look at what I can do just by waving my arm".
Jackass. Worse, the media can't tell the difference between poseurs like him and real scientists like Sebastian Thrun or Rod Brooks.
"It is not impossible to build a human brain and we can do it in 10 years."
I'm relieved to see that tradition of excessive and unjustified optimism by yesterday's AI crowd is still alive and kicking. It's just too bad that a self-aggrandizer like Markram is using it to give false hope to sick people.
Announcing the arrival of a man-made human brain in 10 years, without first having built anything but simulations is so irresponsible and ultimately hope-despairing that there damn well ought to be a law.
Better yet, ask Markram to sign his promise in blood -- if he fails to deliver, he donates his own brain to the cause. His ex-corpus cerebrum may do little medical good, but at least it'll do no more harm.
Absolutely! Burke is right up there with Sagan and Nye as the very best at communicating the wonder and thrill of learning about the world around us.
Randy
Nice, fair minded assessment.
I've lived in MI (Midland, Kazoo, E. Lansing, Ann Arbor) for half of my life (50 years), and have a love/hate relationship with her. Other than the six months of gray skies (Nov-Apr), it's the most livable place I know (vs. PhillyPA, WashDC, SouthernVA, ChampaignIL, & HoustonTX). I'd return to A2 in a minute if I could find interesting work there.
One more bennie Michigan has that the OP forgot to mention:
9) SLASHDOT
Randy
I've read the first few chapters and other than the lame introduction, the book is quite good. I haven't read any discussion yet of the philosophical implications of automated warfare, though I hope there will be (and not just from professional 'ethicists'). Mostly the author discusses the pragmatics of building ever smarter weapons and intelligence systems, which is mighty relevant to the present, and most assuredly the future of war.
Even when the author doesn't use the word "robot", the trend he describes is clear. Tomorrow's war will involve ever increasing amounts of separation between finger and trigger. Is this a bad thing? Obviously I want our side to win and I don't want any US or civilian casualties. But I also don't want war ever to seem as easy as Bush, Rumsfield, Cheney, and Wolfowitz assumed. Automatic War is a sure recipe for an infinite series of mindless incursions and body counts. This book should serve as a wake up call to the many aspects of war-by-wire, from technology to policy to ethics to cost. Going to war must never devolve to the pressing of one red button.
Randy
I'm 50, earned my MS in CS 20 years ago, and have worked in CS R&D for the past 20 years.
Here's why a MS in CS is a good thing:
1) You can take 10 more courses and learn 10 more subjects that you didn't have time for during your BS. Examples: intelligent agents, robotics, machine learning, compiler optimization, virtual machines, advanced graphics, image processing, signal processing, parallel computing, quantum computing, microarchitecture, computer security, computational biology, advanced networks, mobile computing, pure math, applied math, advanced statistics. For many employers, the presence of the right combination of courses in your resume can be very important, and very likely will make the difference in preferring you to another job candidate.
2) Employers value credentials. When it comes time to promote the next staffer into management, they will consider your education. In the first decade of your career, a MS will definitely help you get the more interesting assignments or step up into management.
3) You are less likely to be at a competitive disadvantage to your coworkers. For the first 5 years of your career, almost everyone in the company will have more work experience than you. If you also know less than they do academically, then you lose.
4) Everyone values a MS more than a BS. Knowing more is always better than knowing less.
5) If you want to work in R&D, don't. Not without a PhD. That said, the (optional) research component of a MS may have more value to a R&D shop than would the alternative -- two additional courses. The thesis shows your ability to do (some) research. That said, most non-R&D employers would prefer the courses anyway.
Randy
If it's a pipe dream, then why do so many researchers and physicians regard iPSCs as a holy grail? Probably it's because stem cell therapy has *already* repaired damaged tissue and restored function to a variety of tissues in mammals and humans, including the spinal cord.
In case your tragic state of perplexity becomes too much to live with:
Stem Cell
http://en.wikipedia.org/wiki/Stem_cell
Stem Cell Basics
http://dels.nas.edu/bls/stemcells/booklet.shtml
Induced Pluripotent Stem Cell
http://en.wikipedia.org/wiki/Induced_Pluripotent_Stem_Cell
Video: What Are Induced Pluripotent Stem Cells?
http://video.google.com/videoplay?docid=8370692532177471184&hl=en
Stem Cell Therapy
http://en.wikipedia.org/wiki/Stem_cell_therapy
Randy
"Directed differentiation of human induced pluripotent stem cells generates active motor neurons"
S Karumbayaram, BG Novitch, M Patterson, JA Umbach, L Richter, A Lindgren, AE Conway, AT Clark, SA Goldman, K Plath, M Wiedau-Pazos, HI Kornblum, WE Lowry
"The potential for directed differentiation of human induced pluripotent stem (iPS) cells to functional post-mitotic neuronal phenotypes is unknown. Following methods shown to be effective at generating motor neurons from human embryonic stem cells (hESCs), we found that once specified to a neural lineage, human iPS cells could be differentiated to form motor neurons with a similar efficiency as hESCs. Human iPS-derived cells appeared to follow a normal developmental progression associated with motor neuron formation and possessed prototypical electrophysiological properties. This is the first demonstration that human iPS-derived cells are able to generate electrically active motor neurons. These findings demonstrate the feasibility of using iPS-derived motor neuron progenitors and motor neurons in regenerative medicine applications and in vitro modeling of motor neuron diseases."
Subscription to Wiley Interscience required for more...
Randy
What fraction of this 2/3 already has access to broadband but has rejected it? If the fraction is high, then the 2/3 statistic is obvious pretty much a tautology. The 2/3 includes everyone who *could* subscribe to broadband but rejected it. Meaningless.
The real question is, what fraction of people who CANNOT sign up for broadband also do not want it. I doubt most rural folks don't want better communications technology. I lived in central Illinois for a while and the dial-up Internet community was thriving. It gets pretty damned lonely out on the prairie.
Trust the press to pass along an ambiguity as if it were news...
Randy
After 25 years working in software, I discovered a few years back that programming the same old thing in yet another language just didn't do it for me any more. I need to work on new and different kinds of problems. After a few early years in commercial programming, I wandered into R&D where I've worked on AI systems for the military, then supercomputing problems for the government and then for scientists and engineers in academia. Lately, I've found myself analyzing medical images as part of drug design. In addition, I've become intrigued by vision systems on robots as well as computational biology, perhaps elaborating gene networks or modeling disease. Maybe that's where I'll go next.
However, I'd recommend a different path than the one I took (part-time graduate study ending in a MS in CS). To work independently in R&D, you'll want at *least* a MS, preferably a PhD, with a full complement of math courses through diff eq, statistics, and possibly PDEs. To work in engineering areas like communication or imaging, courses like signal processing, image processing, and perhaps 3D graphics will also help to open doors.
Here are a few more applications for computing that sound like fun:
- graphics, game design/implementation
- networking, new services from cable companies/telecoms, new wifi products
- secure systems, computer forensics
- embedded systems (real-time O/S, low-level software or high-level hardware)
- robotics, semi-autonomous systems
- hand held computing (IMHO, soon to be much bigger)
- data mining, social network analysis, machine learning the datasphere
- searching through or indexing of media or large content (e.g. web, communication streams)
Personally, I'd stay away from building low-level software like compilers and O/S's. This kind of work is likely to be outsourced by the heavy hitters like Intel/M$, since the needed skills are readily available in low-wage countries, and most of the infrastructure is open source (VCs seem to believe that its hard to make a profit by attaching your IP to free software or competing directly w/ the tech oligopolies).
Randy
I agree, passion for producing a better solution which is smaller, faster, better focused, solves problems you didn't anticipate, or does more than anyone ever hoped is *very* much a trait I want in the people I work with. Without it, you're surrounded by 9-to-5-ers.
However, passion for a single technique/tool/philosophy is not really 'passion', but 'dispassion' for alternatives. And that's bad, especially since in computing, there are so many ways to skin a cat, and you need to work as a team whenever possible. Often that means adopting someone else's passion and making it work.
Randy
A is 80%, B is 60%, C is 40%, D is 20%, F is spelling your name wrong.
.
A far better grading system is perhaps the simplest: rank the class. Then there can be no debate about grades because there grades don't exist. Rank is immutable. It eliminates politics and grade inflation altogether.
.
Randy
In my opinion, easily the best way to advance *any* technology is to draw a line in the sand and then challenge others to cross it, either through competition, or better yet, by surpassing the status quo in ways that are quantifiably 'better' than before.
DARPA has achieved some of its best bang for the buck in its recent robotic grand challenges, significantly advancing the field of mobile robotics and doing it with great fanfare. The fact that the progress is visible and practical also makes a world of difference. Look at the popularity of battling (mindless) robots.
Therefore, if DARPA wants to advance other forms of AI, I suggest two action items:
1) Create other grand challenge competitions. Robotics is a nice dynamic medium for demonstrating AI advances and overcoming real-world obstacles. Thus other forms of robotic competition might also be interesting, for example: mobile agent coordination, as in a strategic-driven variant of Robocup, or coordination of ground/air/sea forces, perhaps at battalion level -- a fairly natural application of Future Combat Systems development).
Of course, nonsituated agents also have potential. Competition, conversation, and collaboration among software bots would allow many folks to compete who don't have the resources of a CMU or Stanford. And SoftBots would also better explore other forms of AI, like speech/NLP.
2) Define some standard performance benchmarks (speed and/or accuracy) so that professionals can have a target to shoot for. Each of these should measure a different form of AI like speech & NLP, planning, vision, pattern recognition (like USPS character recognition), etc. These might also include metrics derived from the competitions in action item #1 (e.g. sporting events, like slaloms or gymnastics routines).
I suggest that DARPA launch a Request For Information to the AI research community for suggestions on AI benchmarks and competitions.
As practically useless as the Turing Test has been, it has garnered a disproportionate amount of attention over the years merely because "it is there". But AI research needs to walk before it tries to run. What better way is there to do that than to compete, not with humans nor even simpler life forms, but with other "AIs"? Perhaps the Turing Test should become Turing Match Play, with the winner being not "which is the man", but "which is the better man". (Or woman.)
Randy
But Octave supports little more than the basic syntax of Matlab and omits most of its basic functions and all of its GUI (e.g. no graphing).
As a better alternative to Matlab, I'd suggest SciLAB (http://www.scilab.org/). Not only does it support a much larger fraction of Matlab than Octave (including a lot of functions from the toolboxes), it's actively under development and actually has a GUI.
In fact, SciLAB might be exactly what the Original Poster wants -- a flexible matrix-based programming language with some visualization primitives, that's 98% compatible with a standard programming language to boot.
Randy
Examples of problems that do not parallelize:
1) Problems that contain little data
2) Problems that require sequential processing
3) Problems that are often interrupted, that cannot predict future actions or execution paths (e.g. pipelined)
What fraction of computing tasks match these 3 constraints? At least 90% of the work done on desktops.
Speeding up the other 10% will be done by specialized hardware like multicore video or DSP chips. The real potential for parallelism in everyday computing is negligible. I've been parallel programming for years, and all forms of parallelism are next to useless unless, like Google, you have lots of data.
BTW, Google has no use for multicores either. All of their parallelism is embarassingly parallel, which is better served by shared-nothing architectures like many, many, many cluster nodes that are cheap, cheap, cheap.
Randy
That's because Derrida is already incoherent. No smiley.
There's no question that philosophy isn't trivial and that some philosophers are more sensible than others. But NO form of literature or social sciences requires anywhere the effort of real analysis or quantum mechanics. Nobody with comparable skills in the two worlds would assert that the effort or homework in the competing curricula are comparable either.
Like proficiency in math, proficiency in writing is not for everyone. I test as being comparably good at both, but I find reading/writing/logic a great deal easier than proofs/derivations/transforms. I'm pretty confident I could produce a quantitative measure of the difference in effort if needed.
I've always thought that there should be a different criterion for entry into Phi Beta Kappa from math-based curricula. I'm convinced the effort expended by physical scientists and engineers cannot be compared equally to degrees based on reading fiction or history.
I once completed 80% of a BA in psychology, and the effort level was nowhere near that of a CS curriculum, much less engineering.
Randy
...
6) UNREAL quantities of homework (all math problems)
7) Textbooks that are often badly written/incomprehensible/irrelevant/error-laden
8) 3/4 of professors who speak English poorly
9) 3/4 of professors who drone on and show no enthusiasm for the subject
10) WAY too many equations written illegibly, rapidly, BY HAND on whiteboards
11) Lectures that, way too often, are not available as handouts
12) Handouts that, way too often, are not available before the lecture
13) Material that often feels irrelevant to the real world (esp. the way linear algebra is taught)
As a long time computer scientist who is taking a 400-level course in image processing (my first EE course), I much better appreciate how damned hard engineering is. By contrast, CS is a breeze, and the humanities are about as hard as watching TV. Now I remember why I traded in a BSEE program for a BS in biology.
Randy
1) The technology enabling the abuse must exist. In this case, one or more satellites are already in place to spy on US. Strike one.
2) A policy must exist which legally enables the owner of the technology to misuse it. So far, this does *not* exist. Ball one.
3) The people in a democracy must allow the policy and the technology to be abused. Therefore, if Orange Crush speaks for the rest of us, this requirement is already in default. Game over.
Despite all the rhetoric to the contrary, peoples governed by a representative government don't lose their rights, like "Orange Crush", they give them away.
The solution is clear. Ignore people like Crush who never wanted freedom in the first place. Be an American and vote these assholes straight to hell.
Randy
Ideally, the programming language used to teach a course should support only the core concepts of the subject. It should avoid all else. That's why Pascal and Scheme are probably still the best means to introduce the basics of procedural and functional programming. Sure each language has weaknesses, but the goal in CS is to learn *principles*, not *languages*, and these two perform that role economically and effectively.
Should CS students learn Java or C++? Yes to the former since its today's lingua franca among professional software engineers (and probably no to C++, because it's requires the programmer to overcome far too many interfering semantics when implementing even simple tasks). But Java should still be introduced only when teaching principles of software engineering, not computer science.
Randy
One size will never fit all in parallel computing. There's such a diversity of range (hardware), domain (application space), and user need that a plethora of parallel tools and design choices will always be needed.
SMP machines differ significantly from tightly-coupled clusters, or even loosely-coupled clusters (a la Google), so a range of parallel tools are needed. Some parallelism components inevitably come from language vendors and others come from O/S vendors. As such, programmer access to these tools may need to remain at a low level (perhaps for direct device access) or may rise to a higher level as part of a more abstract programming model.
What's more, tasks go parallel for different reasons: greater speed, greater memory, more responsive GUIs, or to make for a software architecture that's better suited for various data structures like workpiles or trees, etc.
With such a diversity of architectures, multilevel implementations, application domains, and designer needs, it's little wonder that different solutions have emerged. Given the four dimensions and 2-3 variations that I've described for each, I can see justification for at least 2^4 different parallel programming models, tools, or languages.
A final thought. If each of the popular programming languages were extended to support even one form of parallelism (perhaps SMP threading), in addition to the numerous parallel process libraries that exist, how many *more* solutions would then arise? Can we really expect language implementors not to extend their products in order to serve the needs of their current users?
I think it's inevitable: the babel of parallel polyglots is here to stay.
Randy
Moore's Law describes a CPU speedup that died at least 3 years ago (all other legalisms aside).
To wit: I bought a laptop in 2003 with a 2.2 GHz 32 bit P4. According to The Law, by 2005 CPUs on comparable laptops should have run at 4.4 GHz, and by today they should zip along at 8.8 GHz. But in fact, no commodity CPU runs at that speed nor even *half* that speed.
And don't you believe the claim that multicore or power throttling compensates for or explains The Law's "failure to thrive". The fact is, the industry is no longer delivering CPUs whose SPECMarks/FLOPs/etc (AKA performance) is rising at the rate that they have for the previous 20 years. I tell you, "Moore's Law is pushin' up daisies. It's a DEAD parrot."
What's puzzling to me is that while this emperor is clearly naked, for some reason, nobody wants to admit it. Why not? Are we afraid that sexy soothsayers like Ray Kurzweil or Rod Brooks will be regarded laughably when they forsee cool stuff like The Singularity or robots possessing human-level cognition, brought to us by the inexorable exponential march of Moore's Law? Or do we simply dread the day when we have to depend entirely on advances in *software* to deliver our next high-tech fix? Perish forbid *that* thought.
Well, we'd better get used to it, the emperor is naked *and* dead. There's a new emperor in town, and Moore's Law 2.0 depicts a future that looks a hell of a lot like the past.
Randy
Not at all. It'd be trivial to add a button to future preamps/amps that does the same thing for CDs that the narrow/wide toggle did for FM stereo years ago -- namely compress/expand the dynamic range.
Of course, compression could be introduced when burning the MP3/AAC, which is the better solution anyway. Compression makes less sense for music that is played around the house (e.g. CDs). It makes more sense when the listener is on the move -- in the car, ipodding, etc.
If two versions of each song were to be distributed, I'd prefer they be built into the media: disc vs. audio file, where only the MP3/AAC gets compressed by default.
A final thought. It may be that if CDs are to be compressed by default, the practice may only increase the demand for new music format variants (like the various bit rate versions of a song available now from iTunes). Perhaps this would add yet another MP3/AAC/AIFF format -- perhaps an uncompressed SACD-quality binary. Now that's a compromise that an audioprig like me could live with.
Randy
"Academic" vs "professional" is a fatuous distinction. By those terms, any discipline that is useful outside of the university is less deserving. Should we respect such degrees less, or charge more for them (irrespective of their instructional cost) because they have commercial value?
There is nothing inherently less challenging, enlightening, or "worthy" about purposeful studies than purposeless ones. In fact, the fact that a subject has utility *must* be the single most important criterion by which any field of study should be judged... inside or outside the academy. Otherwise, irrelevancy takes precedence over relevancy, and publishing "A Critique of Humor in The Simpsons" becomes preferable to stopping a war or curing cancer.
It's not only the engineerings and business, but medicine, chemistry, geology, clinical psychology, laboratory-based biology -- all of these studies are both applicable In The World and costly to acquire. The pure science degrees then fall between the cracks, perhaps immediately useless yet grounded in useful principles: physics, mathematics, and perhaps economics and non-laboratory biology. Should we charge more for their tuition, not because they cost more to teach, but merely because employers might value them?
If so, your notion of a worthwhile education certainly differs from mine.
Randy
As to whether in-silico lab work is feasible, they jury is out. It may be that computers will be most useful as a complement or augment to lab techniques. It's indisputable that computers are central the use of most lab instruments today, from calibration to DSP, image processing & analysis to statistical analysis of the data they generate. The next question is whether computers can generate useful information by modeling the real world. That endeavor is in its infancy, but it'll get better, specially with improvements in extracting data from lab instruments, which again, will be accomplished by computers.
Should we draw a hard line between the use of computers on the supply vs the demand side of science -- in collecting and organizing data from instruments, but not exploring the possible interactions of that data using simulation? I don't know why we should. There's a natural feedback loop there, where the data guides further exploration. Computers are central to both sides of that equation.
I want to extract all the information that's out there and then make the best possible use of it. Science exploration is not an either/or proposition. The problems are hard enough that you try *anything* that improves your odds of success. Then you try it again.