Why we're the only intelligent life in our galaxy

[KimB]

This one asked key scientists to name 15 key paradigm shifts in human history, the following plot shows exponential progression no matter who picks the events

Just a little nitpick here, but linear on a log-log plot means power law progression, not exponential (but still faster and faster as time goes forward).

Or this interesting one on actual progress in supercomputer power

The trendline is unlikely to continue like that. We don't know what the scaling law for the next generation of computing technology will be, but it is unlikely to be the same as silicon-based technology.

 

[DemoCoder]

Kurweil has back-extrapolated Moore's law (before the invention of semiconductors, even back before vacuum tubes) using data to show that each new paradigm shift exhibits the same scaling. There is no reason to believe the scaling won't continue. We know physically plausible mechanisms for future computing, such as rod-logic or RSFQ that can scale to far far higher performance than current semiconductor lithography, and most people still think that conventional lithography still has atleast a decade or so of growth in it. We are no where close to fundamental physical limits. And ultimately, the Beckenstein Bound shows that limits on information storage and processing are outrageously large compared to what we are achieving today.

Moreover, super computers don't scale purely by density of transistors, but also density of geometry of the packaging, power density, and communications density. If you get power density, cooling, and communications under control, you can scale horizontally. Limits on clustering in most datacenters are purely a power and networking problem.

Blue Gene didn't beat other super computers because individual Blue Gene CPUs are so much more powerful, but rather, because Blue Gene will eventually have 1 million processors, whereas traditional clustered supercomputers topped out at 10^4 or so processors. Some of them even required cryogenic cooling systems.

The 2013 "power required for functional simulation of brain" is a factor of 100 faster than Blue Gene. So you don't believe that Blue Gene's CPUs, which are not even built on 65nm, won't be able to shrink by a factor of 10-100, and that IBM won't be able to build a system with 10 million CPUs or 100 million CPUs by using multicore to fit say 128 cores per package? I'll note that even Intel thinks 128 cores is possible in the near term!

The 2025 mark only requires a 1,000 fold increase over and above a 128-core version of Blue Gene. This could be achieved simply by a shrinkage of a factor of 10, and using 100 times as many CPUs (1280 cores per chip, and 100 million CPUs) I don't even see how traditional lithography limits this.

Azus Systems is about to ship a 48-core system using 800million transistors manufactured on 90nm TSMC process!


The naysayers on limits to computing have been wrong so so many times. I remember in high school reading in EETimes about how future CPUs would never make it past 25Mhz unless the industry switched to Gallium Arsenide.

Put another way, mostly everyone in the industry agrees that 16nm is achievable and that Moore's law will extend out to atleast 2018. That's atleast 8 more doublings, or a whooping increase of 256times. Intel thinks 7nm is achievable albeit with quantum dots or nanotubes, but that doubles performance a few more times. Rod-logic is expected to be a factor of a million times denser and a million times more power efficient.

 

[KimB]

Moore's law is already slowing, though. Yes, we may be able to push it down to 16nm, but it's getting harder and harder as we go further down: quantum effects are going to make for dramatic differences in challenge here.

You should never expect every scaling law to continue forever. Scaling phenomena are very frequently accentuated by "knees:" places where the scaling changes from one power law to another. These often happen when something fundamentally new comes in.

Anyway, in ten years or so we'll see whether or not the scaling laws for the next generation of technology are different. The scaling may be faster, may be slower, I don't know. I hope for faster, of course, but we'll see.

 

[DemoCoder]

Well, Kurweils' point is that the last several times scaling leveled off from exponential, it was replaced with a paradigm shift that restored exponential.

I like this quote from RIT

"For 30 years we've felt that optical lithography would run out of steam in a couple more years," said Professor Bruce Smith, director of RIT's Center for Nanolithography Research, who presented the EWL paper at SPIE. "But these kind of results show that optical lithography is continuing to make strides. And it may be the only lithography that is ever needed for semiconductor IC fabrication."

Smith added that the developments have given industry participants something different to think about. "We're not limited by conventional optics," he said. "IBM showed how far you can push immersion lithography. We showed a capability beyond the limits of immersion lithography using the evanescent field."

Barlett CEO just recently announced at SF SDC that Intel's road map now officially includes 5nm. They see "how" to do it, it's just a matter of costs and engineering now, which means they won't bother until they get close to the 10nm "barrier".


I don't have a doubt that scaling will continue because IMHO, if the laws of physics permits something, then it can be built, albeit after a potentially very difficult engineering search. We know how to build primitive rod-logic components, we know how to store 1 bit per atom. We can even calculate how fast they can run and how power efficient they'll be. We just don't know how to manufacture them en-masse at the moment.

I imagine if 20 years ago, someone said they were going to make a chip with 1+million micro mirrors on it, one might be skeptical that it could ever be engineered. But here we are today with DLP DMD chips. Hell, there are even companies like Nanotera shipping Nanotube RAM today, so the future may not even be as far out as we think.

There is also alot of speculation that the NSA is plowing funds into quantum computers and RSFQ devices (they are principle funder of COOL-0), so who knows what exists in black projects.

(BTW, RIT demonstrated that they can produce optics that bypass Snell's law)

 

[KimB]

I don't have a doubt that scaling will continue because IMHO, if the laws of physics permits something, then it can be built, albeit after a potentially very difficult engineering search.

Right, which is exactly my point. The laws of physics don't allow transistors smaller than ~30nm that will operate as we understand them today. You can go smaller, but then the behavior starts to change from the statistical mechanics-determined behavior we're used to (hell, even at ~30 nm it's somewhat different).

Yes, we'll probably make use of single-electron transistors soon after current types of microprocessors lose steam, but they'll require rather different paradigms in construction to produce the desired computations.

 

[nelg]

Personally, I think we're more likely to augment our own bodies instead of developing new ones,

If true it seems that there is a big future in being a penis salesmen.

 

[DemoCoder]

""In five years, the penis will be obsolete." - The very first line of the Science Fiction book "Steel Beach"

Chalnoth, of course transistors will change at that level, but we've got two decades to study it, and the leading process companies have already figured out some of how they will break through that barrier. My point is, "there's plenty of room at the bottom" to quite Feynman. Current semiconductors are no where near the physically possible maximum density/scale achievable with atoms.

 

[KimB]

Chalnoth, of course transistors will change at that level, but we've got two decades to study it, and the leading process companies have already figured out some of how they will break through that barrier. My point is, "there's plenty of room at the bottom" to quite Feynman. Current semiconductors are no where near the physically possible maximum density/scale achievable with atoms.

Er, if we're at 65nm today, then 32nm should be three years out (by Moore's Law), and 15nm three years beyond that. 15nm is in the realm of single-electron transistors, so if Moore's Law held, no, we wouldn't have two decades to study these changes.

 

[zidane1strife]

I was just wondering what everyone was basing their opinion on evolution slowing or accelerating. What evidences have there been of this in science. I don't remember hearing anything about anyone observing evolution(well macro evolution). So, how do you know whether it has slowed or accelerated? Also, it does depend on what school of evolution you believe. There are schools that believe evolution happens very slowing while other schools believe it can happen very quickly in certain conrolled circumstances. It is also possible I am assuming the wrong thing. Is it possible you are discussing microevolution within the human species? Could you guys clarify a bit.

speciation/macro-evo was recently documented by researchers, either in the wild or in the lab, I don't recall the details at the moment(think it was in science mag that I read about it).

Recently evidence suggested a particular species of fish's evolution was accel by human activity(high mutagen production.), probably something similar's been happening in the human species. The amount of toxins/mutagens being thrown into the environment besides increasing the rates of cancer in certain areas, is probably accel evol in the direction of more ideal organisms(aka, those capable of withstanding greater assaults without injury/corruption/cancer.). Not even medicine can stop the costs associated with high mutagen exposure and associated treatments. Besides that non-gene-based evolution has also been accelerating.

In any case evolution does not make any case for/agaisnt the existence of a God.

Moore's law is already slowing, though. Yes, we may be able to push it down to 16nm, but it's getting harder and harder as we go further down: quantum effects are going to make for dramatic differences in challenge here.

You should never expect every scaling law to continue forever. Scaling phenomena are very frequently accentuated by "knees:" places where the scaling changes from one power law to another. These often happen when something fundamentally new comes in.

Anyway, in ten years or so we'll see whether or not the scaling laws for the next generation of technology are different. The scaling may be faster, may be slower, I don't know. I hope for faster, of course, but we'll see.

Chalnoth I used to think molecular machinery was the last straw, that we'd be stuck with it in a decaying universe. Like the norse gods, godlike power, ageless, but fated to die. Civilization doomed in the end. But now I see a path beyond molecular machinery, one I once thought ludicrous, but now believe is in fact quite viable.

 

[NANOTEC]

But now I see a path beyond molecular machinery, one I once thought ludicrous, but now believe is in fact quite viable.

Something like quantum machines?

speciation/macro-evo was recently documented by researchers, either in the wild or in the lab, I don't recall the details at the moment(think it was in science mag that I read about it).

Recently evidence suggested a particular species of fish's evolution was accel by human activity(high mutagen production.), probably something similar's been happening in the human species. The amount of toxins/mutagens being thrown into the environment besides increasing the rates of cancer in certain areas, is probably accel evol in the direction of more ideal organisms(aka, those capable of withstanding greater assaults without injury/corruption/cancer.). Not even medicine can stop the costs associated with high mutagen exposure and associated treatments. Besides that non-gene-based evolution has also been accelerating.

Just out of of curiousity, has there been any studies showing newer generations of the population having stronger immune systems while retaining the same performance of their existing immune system?

 

[nutball]

Out of interest where did the figure of 10^19 FLOPS required for a "human brain simulation" come from?

 

[NANOTEC]

Out of interest where did the figure of 10^19 FLOPS required for a "human brain simulation" come from?

SONY's CELL ppt.

 

[Druga Runda]

ooohhh... this has grown

just to comment - great slides DC

 

[DemoCoder]

Out of interest where did the figure of 10^19 FLOPS required for a "human brain simulation" come from?

There are various papers in the literature that place an upper bound on the amount of information stored in the brain and how many calculations per second it can perform using various neurological models to count. One might be tempted to discount them, except for the fact that they have successfully used such models to place artificial neurons in lower lifeforms that record, model ( "shadow" ), and then assume the functionality of the natural ones, and the outputs agree.

The success of these techniques on small numbers of neurons (typically just a handful using nematodes which have ~300 neurons) have led IBM to attempt to build a 10,000 neuron simulator that will be used to model a small section of the human brain to test the predictive power of the neurological models.

Relatively crude estimates are obtained just by counting the total number of synapses (where most memory assumed to be stored) and assigned some value per synapse (1-bit per synapse, 1 byte per synapse, it doesn't much better), and you'll get ranges like 10^16-10^19 bits depending on how many bits per synapse you assume. Depending on whether you assume neurons, dendrites, axons, and glia all participate in computation or just neurons, we know from biology the firing rate of these neurons (how fast they can signal), so this places an upper bound on the total processing rate.

Ultimately, the laws of physics also bound you. The Beckenstein Bound places an absolute upper limit of 10^43 bits and I can't recall the CPS, but the speed of light and the diameter of the brain limit CPS with an ultimate upper bound.

There's just no escaping the fact that it will soon be possible to build a computer with more storage and processing power than a single brain, even though many dualists will fight tooth and nail the idea of any kind of physical limitation on brain power that puts it in reach of any possible computer, no matter how big. (leading to absurdist claims in some camps, like the human brain stores 10^8432 bits of information)




For CPS, they can calculate

 

[Druga Runda]

And if they arrived prior to the existence of any intelligent species on the planet, I find it exceedingly likely that they would simply take the planet as their own, supressing further evolution of an intelligent species (us).

I pretty much agree with the rest of the argument Chalnoth presents thus far apart from this tidbit.

Why is it necessary for the intelligence species to extinguish all other forms of intelligent life because they are the first? The post about "social development" of us humans gives me enough faith that a developed civilization would not necessarily follow that route. Think how we now are keen to protect "endangered species" and not only because this is necessary for our existance, but purely because they are here,and we can both live good, and protect them at teh same time with little effort.

One point thus far that all of you seem to be ignoring is the exponential speed at which the "evolution" progresses , and correspondingly the technological advances that such a civilization posesses and changes that such a society has gone through are going to be unfathomable to our mind.

I would very well imagine that "non habitable" planet colonization, perhaps even fasat interstellar or intergalactical travel (if the laws of physics in this universe permit that of course) will all be discovered and used.

Going even further, even if the universe physics never allowed the FTL travel, if the age of the universe + size allows intelligent life to evolve elsewhere.Than even if those intelligences were locked out in some galaxy far far away :smile:; they would have moved out of their galaxy given a few million years time post the stage of development which we arrived at. I am pretty certain they will go out of their galaxy/galaxies into the new ones - think egyptians traveling to America and similar.

While this is a nice long stretch, I think it is much less off a stretch than "we are the only ones", "first ones" etc...

--- now going to read some more ---

This thread is very interesting ... for more than just the subject matter.

It seems to be saying a lot about what people want to be true. It also seems to show how deeply certain ideas from science fiction have become engrained in common culture.

---
So basically ... science, either accept and acknowledge it's flaws but see what it has to tell you, or ignore it and accept that you have a faith-based world view. There's not really a half-way house.

of course - but than again to me those kind of arguments are more about the "worldview" than postulating a scientifically provable theory on why there is something or not, as it is we do not stand much chance of doing that.

What are the options: There is nothing apart from us, or there is something but not near, or there is something but does not want to show itself.

To me option #3 is most plausible.

To you the main argument is "well the probablilites have not been calculated"/ "this is unmeasurable" etc... in any case if the probability for life is >0, and if either time or space is infinite than life is a certainty, and in quantities more than 1. Even if the probability for existance of life is exceedingly small, the probabilty that we are the first/only ones is again a lot less.

 

[NANOTEC]

Why is it necessary for the intelligence species to extinguish all other forms of intelligent life because they are the first? The post about "social development" of us humans gives me enough faith that a developed civilization would not necessarily follow that route. Think how we now are keen to protect "endangered species" and not only because this is necessary for our existance, but purely because they are here,and we can both live good, and protect them at teh same time with little effort.

Yes and further an intelligent life might want to preserve "us" just for study and research assuming this intelligent life is totally different from our own. Some of the key breakthroughs in biological science comes from studying how other lifeforms function right here on earth. Just as an example green plants may seem to be just a "stupid organism" but they may hold the answers to certain scientific problems. That is one of the reasons why science tries to preserve what we have here on earth. Maybe a certain plant or lifeform holds the chemical to manufacture a future medicine that cures a certain present or future disease? What happens if we destroy this plant or animal to make room for colonization?

 

[NANOTEC]

double post

 

[DemoCoder]

Druga you are handwaving probability figures around. Neither time nor space are infinite, we have pretty good bounds on the size and age of the universe. You don't know the probabilities involved in life, so it makes no sense to pronounce that small probability * age/size of universe => implies high probability of more than one life event. You could be talking about numbers whos exponents differ by dozens of orders of magnitude.

Even infinite time or space doesn't save you.

The idea that there is something near, but doesn't want to show itself appears to be based purely on belief. Why is it the most plausible? But what basis do you claim life is more likely to be nearer but hidden, than far outside our light cone? On what basis should it be so clustered together?

This is where I think the frustration comes in, because alot of implicit and non-supported (faith-based) arguments get thrown around.

Either you have a rational argument for why it is true, or you have evidence, but you "gut feeling" is not sufficient.

 

[KimB]

I pretty much agree with the rest of the argument Chalnoth presents thus far apart from this tidbit.

Why is it necessary for the intelligence species to extinguish all other forms of intelligent life because they are the first?

No, that's not what I mean. I mean that if we take planet X, upon which some intelligent species could someday evolve if left alone, the colonization of planet X by some other intelligent species Y will alter the evolution on planet X so that a new intelligent species cannot evolve.

One point thus far that all of you seem to be ignoring is the exponential speed at which the "evolution" progresses , and correspondingly the technological advances that such a civilization posesses and changes that such a society has gone through are going to be unfathomable to our mind.

With all of our advancement, we have not yet stopped wanting to spread ourselves and our culture. My claim is that this desire is a fundamental aspect of life, and something that isn't going to go away. Therefore, any sufficiently-advanced society is going to eventually colonize, if it doesn't kill itself off first.

I would very well imagine that "non habitable" planet colonization, perhaps even fasat interstellar or intergalactical travel (if the laws of physics in this universe permit that of course) will all be discovered and used.

Sure, but as I've said previously, it's much easier to make use of an already-habitable biosphere than worry about artificially creating one (except for the possible exception of planets within the same solar system). It will take a lot of time and natural resources to terraform a planet, and much more in the way of resources to generate a biosphere from scratch.

While this is a nice long stretch, I think it is much less off a stretch than "we are the only ones", "first ones" etc...

Well, I claim only ones in our galaxy. That doesn't mean there aren't others in other galaxies, so far that they could never have reached us.

What are the options: There is nothing apart from us, or there is something but not near, or there is something but does not want to show itself.

To me option #3 is most plausible.

But option #3 makes no sense except in the obscenely-unlikely scenario where the intelligent species is within about 50-100 light years. Only in being that close could they have possibly detected our existence as an intelligent species. And they would also have had to evolve intelligence at almost the exact same time as us for them to not already be here.

To you the main argument is "well the probablilites have not been calculated"/ "this is unmeasurable" etc... in any case if the probability for life is >0, and if either time or space is infinite than life is a certainty, and in quantities more than 1. Even if the probability for existance of life is exceedingly small, the probabilty that we are the first/only ones is again a lot less.

Sure. Just bear in mind that the observable universe is finite. The universe with which we have a chance of ever interacting is actually smaller (due to the accelerated expansion), and the universe with which we ever have a practical chance of physically meeting extraterrestrials is even smaller.

 

[KimB]

Druga you are handwaving probability figures around. Neither time nor space are infinite, we have pretty good bounds on the size and age of the universe. You don't know the probabilities involved in life, so it makes no sense to pronounce that small probability * age/size of universe => implies high probability of more than one life event. You could be talking about numbers whos exponents differ by dozens of orders of magnitude.

Nitpicking, but:
There is formally no upper bound on the absolute size of the universe, just the size which we can see, and the (different) size with which we can ever interact. We believe from inflation that the absolute size of the universe is much, much larger than what we can see.