Discussion Topic |
|
This thread has been locked |
Ed Hartouni
Trad climber
Livermore, CA
|
|
|
Jul 21, 2014 - 10:49am PT
|
what's with HFCS? he seems very confident in evangelizing science, though seemingly based on his reading of popular books on the subject. Undoubtedly he gets the impression that there is little that is subtle about our understanding. So he proclaims that this understanding is "the truth."
Throughout my career cosmology alone has changed, radically, nearly every decade. This is about to happen again if the BICEP result holds, and even if it does not, the result has challenged details of the current cosmology. The BICEP result has the potential to be the most important result of the 21st century, now only 14 years old. It is that important.
It seems that HFCS needs the surety of "truth" and is very uncomfortable dealing with uncertainty, something I think scientists, and certainly physicists, have to be very comfortable with, the reason being that our lack of understanding on certain issues has to be set aside as we apply our understanding on related subjects. I relayed the Newton story above to indicate that even at a time when physics was creating "immutable laws of nature" there were aspects that we didn't know anything about, Newton knew that gravity acted at a distance, he didn't know how and he didn't speculate "I make no hypothesis" [hypothesis non fingo]. His theory of gravity is still with us, and as such, doesn't need a hypothesis regarding "action at a distance" to be applied. A field theory describing Newtonian gravity comes to us as the "weak field limit" of General Relativity, Einstein's theory of gravity.
There are still major challenges to General Relativity, expected to hold in some domain, but physicists know that when the field gets very large the effects will be felt at the atomic scale, and that we will need to reconcile gravity with quantum mechanics. Still something not convincingly done. BICEP gives us a window on this scene...
But something as trivial as Ohm's law has seemingly unhinged him... I'm not at all sure why. It is an empirical "law" developed by Ohm in 1827, Kirchhoff reformulated it as a part of his "circuit laws" in 1845. These "laws" are used today, in particular the widely used circuit simulation program SPICE has them embedded as it's foundation, along with the circuit element descriptions. In general the description has to have some form of model that describes the "I vs. V" response of the elements, I being the current, and V the voltage potential across that element. For "ohmic" devices, the I vs. V curve is a line.
Ohm couldn't anticipate the various devices that would eventually be found to not "obey" his law. But in some ways his law defines the behavior of a lot of material. Purcell shows freshman physics majors how to approach not only Ohm's law, but other empirical laws too. He builds a model based on the "underlying" physical principles, in this case of the movement of electrons in matter. He doesn't make an ab initio calculation, i.e. starting from quantum mechanics, but instead starts with transport theory, from statistical mechanics, and keeps the problem in the "classical" regime. He proceeds to show how Ohm's law can be "derived."
But the utility of the model is much greater, and isn't meant to be a "proof" of a physical "truth." That utility is to show the domains of its validity. Beyond that domain, some other physics is happening, physics which is identified by the model. Purcell shows that the properties of matter that start to require a quantum mechanical description (the tunneling of a semiconductor junction, the ionization of a atoms important to describe both the high field behavior like sparks, and the much lower field behavior of electron emission) are outside of the rather large domain encompassed by Ohm's law.
These domains, violations of the "law," lead to important new understanding and eventually to the development of technologies which effect our lives every day (no transistor = no STForum). And our understanding leads to the application of these technologies to other parts of science (like building a transistor junction out of a biological lipid membrane to investigate how various ion channels work, a post-doc poster I recently saw at the lab's post-doc research celebration, I didn't understand the biology in detail, but I understood the device).
It might be "picky" to insist that the details are right and our descriptions are precise. But so much of science is exactly that, and even Ohm's original experiments depended on precision to make the collection of observations into a "law." It wouldn't have been called such if there were significant non-linear affects in normal materials.
Instead of "shaking the foundations" of physics, the existence of non-linear devices, the "true" foundations of physics are shown to be much more sturdy by the understanding of this physical phenomenon.
Those foundations provide us the "faith" that we can resolve the various paradoxes that confront physics. Since these paradoxes are the result of moving beyond the domain of current models, into the "unknown," our "laws" are of little use in that frontier.
|
|
Marlow
Sport climber
OSLO
|
|
|
Jul 21, 2014 - 10:53am PT
|
Jgill
That's a cool complex pattern.
If this was a Rorschach-test and I put into words some of my... eh... random associations, the ST moderators could bring me in trouble...
|
|
jgill
Boulder climber
Colorado
|
|
|
Jul 21, 2014 - 10:59am PT
|
^^^
Yes, I was a little shocked when that materialized on my computer screen. It does trigger the libido!
;>)
|
|
MH2
climber
|
|
|
Jul 21, 2014 - 11:02am PT
|
Exploration. The urge to find out. The hardship and satisfaction of making use of our brain. In some sense our past has prepared us and motivated us to move ever onward.
Staying at home isn't a bad choice, either.
|
|
BASE104
Social climber
An Oil Field
|
|
|
Jul 21, 2014 - 11:16am PT
|
What's an example of an unpredictable deterministic system?
A river is a good example. It is a complex, dynamic system. We see many of these systems in nature. With these systems, it is impossible to determine the XYZ location of every water molecule with total precision. That is the nature of dynamic, chaotic, systems.
A river is filled with eddies and turbulence. These can't be precisely predicted down to the last molecule, but they can be characterized, or modeled, to a reasonably high degree of confidence. You can even model the turbulence.
Weather, even at a macroscale level, is also a dynamic, turbulent, system. So is a flame.
They can still be understood and even numerically modeled, despite the fact that even if you know the absolute initial conditions of the system, you won't be able to predict the location of a water molecule moving through the Merced River for 20 miles. You can approximate, though.
This is basically chaos theory.
The brain is a very complex and dynamic system. It can be physically examined down to the last molecule in theory, but it cannot be precisely modeled in a computational system. That is why the idea of simply creating a digital copy of a brain is not the best approach to AI. It is an almost impossible task.
Simply counting the number of neurons and then expecting a mind to emerge from a Moore's Law predicted future supercomputer, with the same number of neurons, is not the problem. It is not a hardware problem. It is a software problem. The complexity of software does not march forward along with Moore's Law. (I pilfered that last paragraph from Ray et al.)
The Turing Test is also a fallacy. Right now we have only one example of intelligence to study: human intelligence. There are many AI researchers who think that limiting ourselves to the example of human intelligence is poor imagination. Indeed, possible types of intelligence are currently limited more by our imagination than by hardware.
For more reading on the topic, and how it applies to AI, I suggest reading a couple of papers.
The first one gets good in the bottom half:
http://life.ou.edu/pubs/kurzweil/
The other thing to read is software evolution, by either natural or artificial selection. This was first achieved by the Tierra Project, and I am neighbors with the guy who wrote the first computer programs which evolved. It grew into a whole ecosystem that was unexpected.
http://en.wikipedia.org/wiki/Tierra_(computer_simulation);
Evolutionary computing is already a rich field in which computer programs design other computer programs, and do it better, or cheaper, than humans.
I am curious if directed computational evolution could eventually produce intelligence.
|
|
Tvash
climber
Seattle
|
|
|
Jul 21, 2014 - 11:59am PT
|
jgill's got it about right IMO. Practically speaking, we exercise some free will at the macro level.
Just how much 'some' is - (addiction, compulsion, etc) well, there's a salient policy question for youz.
This isn't too hard a concept to grok - all natural systems operate differently at various levels of the hierarchy once you zoom in enough.
|
|
Ward Trotter
Trad climber
|
|
|
Jul 21, 2014 - 12:09pm PT
|
The complexity of software does not march forward along with Moore's Law.
This is true ,but then again most of the applications currently having a transformative effect on society are based upon "Big Data" carried out by what Jaron Lanier calls "siren servers"
And if these big data trends continue then eventually they will become the predominant model ---perhaps in many AI approaches, even as we speak.
In other words, cybernetic complexity and symmetry may not be the primary aim here as reflected in software development . Instead the unromantic journeyman task of merely acquiring and then crunching big data is the fundamental task. The "bottom-up" approach.
A mundane example might be these numerous email ads by loan companies to give you $1000 or $9000 loans .
Many of these entities are not primarily interested in loaning you money. They are chiefly involved in mining data. Millions of people fill out their applications and actually very few are approved. This is because their raw information is what these entities are after. This information is crunched and is then used to form the basis of evolving strategies for investment and perhaps thousands of other uses. Facebook, Amazon and countless others do the same thing-- via other methods but with the same end in mind.
One day, as a result of Moore's Law , it might be possible to combine the equivalent of thousands of siren servers crunching huge amounts of data in mere milliseconds . You could then put this capacity to work into a fairly handsome robot and have it walk around basing its actions and reactions on the crunching of big data and the acquiring of more big data. LOL
Such a robot might not be all that complex. Certainly not nearly as complex as an organic brain and nervous system.
|
|
Tvash
climber
Seattle
|
|
|
Jul 21, 2014 - 12:15pm PT
|
You can't discuss too many natural systems without chaos theory - and some knowledge of probability distributions. We can predict the future of systems with a chaotic component - but the result is a probability distribution of outcomes that gets fuzzy fast the further in the future one attempts to venture.
Take the trajectory of an asteroid. Apply the simple law of gravitation and you know exactly where the thing is going, right?
Until you add the affects of micro meteor impacts, the solar wind (generated by a hugely chaotic system), the dynamic and gravitational vagaries of every body with any gravitational affect (everything in the universe, really - but mostly big stuff nearby), the dynamic and varied albedo of the asteroid itself, its rotational dynamics...not to mention Bruce Willis.
Even the closest near earth asteroids are not very predictable - close fly by's have windows that are tens of thousands of km wide.
|
|
MH2
climber
|
|
|
Jul 21, 2014 - 12:16pm PT
|
I am curious if directed computational evolution could eventually produce intelligence. (BASE104)
Many things can happen...
eventually.
But even if every proton and electron in the universe turned into a monkey typing it would take them far far far longer than the current age of the universe to produce a Shakespeare play.
Your question might be answered by
Von Neumann's stated problem was evolution: how is the complexity growth and evolvability of biological organisms possible? His machine shows how it is logically possible, by using a universal constructor, but does not show how it is possible in practice.
http://en.wikipedia.org/wiki/Von_Neumann_universal_constructor
Von Neumann talked about a 'complexity barrier' below which machines could only make other machines that were less complex than themselves. Above the barrier, there is still the question of what the upper bound is.
|
|
BASE104
Social climber
An Oil Field
|
|
|
Jul 21, 2014 - 01:34pm PT
|
Well, we already have computer programs that write other computer programs. That is a form of replication right there. Natural selection can be "tweaked" into artificial selection.
I am talking about creating a digital ecosystem, and I am not by any means the first one to consider that. It has been part of AI research for decades.
Part of this is the limited imagination of the humans who create the digital ecosystem. We tend to draw from nature, because it is our only example. Even then, natural selection is a powerful attribute.
To really understand this problem there are a few key words that help:
Ontology in a philosophical sense:
http://en.wikipedia.org/wiki/Ontology
Ontology in a computing sense:
http://en.wikipedia.org/wiki/Ontology_(information_science);
Another word you need to understand is heuristic.
Heuristic in a psychological sense:
http://en.wikipedia.org/wiki/Heuristic
Heuristic in a computing sense:
http://en.wikipedia.org/wiki/Heuristic_(computer_science);
This whole argument is deep into the roots of ontology.
|
|
BLUEBLOCR
Social climber
joshua tree
|
|
|
Jul 21, 2014 - 03:41pm PT
|
Such a robot might not be all that complex
Your right! One can tell here it doesn't take much for human interaction. Without emotion we would be robots(Jstan's perfect world). They've already shown a computer(Henry? on Jepordy) to be as efficent as a human regarding facts. Now they are working on the emotional questions. Have you used "Siri" of Apple fame? She has come a long ways in the last few years. She can be funny, and even warm and fuzzy. i doubt if they'll allow her to be condescending or hatefull though.
Did you hear the facebook scandle a few weeks back? The woman VP said they were performing a test on 700,000 people to see if they would get emotional over something written on the internet. All i could think was, a WOMAN needed to ask a question like that? LOL! Facebook prolly has the most recorded emotional responses from humans EVER! Along with Twitter someone could program a robot to be able to sift through all those experiences to be able to interact with emotional responses.
Currently they need to focus on the movement of robots. Until they can get one to move from doggiestyle to missionary in 2 seconds, and do the dishes they aren't going to sell many.
|
|
BASE104
Social climber
An Oil Field
|
|
|
Jul 22, 2014 - 07:33pm PT
|
Remember when IBM built Watson?
Watson was a question answering machine that was programmed to play "Jeopardy!" It handily defeated Ken Jennings and another guy. The two best Jeopardy! winners in history.
You guys need to read the wiki description of how Watson was programmed, and how it "evolved" into a monster answering machine that answered questions posed via natural language and then answered them. Better than any human could do.
http://en.wikipedia.org/wiki/Watson_(computer);
It is really interesting.
|
|
BLUEBLOCR
Social climber
joshua tree
|
|
|
Jul 22, 2014 - 09:49pm PT
|
i just watched the Sam Harris vid on Freewill.
They start out taoughting him as one of The Four Horsemen. The anti-christ or somethin. i won't get into all the facts he gets bent,(unless someone wants to?), i'll go right to the irony. The irony is, he spends 45min talking how psycopathic killers(the holders of extreme hate) are NO more in control of their actions than compared to a man wiyh a tumor in his brain. That his biological actions are a result of cause and effect. just as in the man with a tumor. He was saying everything but what he meant to say is that its all genetic. After he started his conversation with criticising christians for not believing in evolution and insinuating the whole bible is questionable because christians don't believe in evolution, and that man was screwing monkeys for a million years. Sam then jumps on Jesus's side in loving your enemys's, cept Sam puts a twist on it. With HIS love and compassion spew. Then he jumps in with God on the whole Adam and the original sin thing like he didn't remember his last 45min of spew.
i don't get'em.. Very hyprocritical. He must be a jew?
i'd love to go toe-to-toe with him on every sentence!
Anybody here like what he said? Besides High fructose corn stoner.
Where do you stand, JStan?
|
|
BLUEBLOCR
Social climber
joshua tree
|
|
|
Jul 23, 2014 - 08:59am PT
|
Who is Mind?
|
|
Tvash
climber
Seattle
|
|
|
Jul 23, 2014 - 09:02am PT
|
Yup:
"consciousness is a schematic model of one’s state of attention."
"Consciousness cannot be what is sometimes called an epiphenomenon — a floating side-product with no physical consequences — or else I wouldn’t have been able to write this article about it."
I prefer articles that do not require a disclaimer like "WARNING: Heavy Lifting Ahead - And You're a 98 LB Weakling, So Don't Even Think About It".
|
|
Tvash
climber
Seattle
|
|
|
Jul 23, 2014 - 09:10am PT
|
"There is no need for anything to be transmuted into ghost material, thought about, and then transmuted back to the world of cause and effect."
"Subjective experience, in the theory, is something like a myth that the brain tells itself. The brain insists that it has subjective experience because, when it accesses its inner data, it finds that information."
Go...
Go...
Go...
|
|
|
SuperTopo on the Web
|
Let us know!