Stephen Hawking: Off Earth by 2110?

[Temujin]

I'm pretty sure the G Forces are too high for live cargo, unless your physiology is protoplasmic by nature.

[cosmicalstorm]

Well I recently finished some books on the Singularity-theory (Singularity is near, Accelerando), though I'm a bit skeptical there is at least a bit of hope in that train of thoughts.

[Broomstick]

For 'spacegun' type launch systems, would it be possible to them to launch crewed assets into space, or are they only good for unmanned cargos?

It would depend on the g-forces involved, and for how long they are imposed. Typically, anything over 5-6g's in an unprepared human rapidly causes unconsciousness, a g suit means 8-9g's, anything much over 10-15 g's for more than brief moments is going to start killing people. Over 20 g's is frequently fatal though there have been some exceptions to that rule.

From personal experience, things start getting uncomfortable over 2g's for any length of time, and over 3 or 4 are going to be perceived as really unpleasant by just about everyone. While tolerance to g forces can be increased I've never heard of anyone consider them enjoyable. Well, on roller coasters perhaps, but the 3-4 g maximum loads typical of roller coasters are imposed only for seconds. It will take longer than that to reach orbit.

[Temujin]

Well I recently finished some books on the Singularity-theory (Singularity is near, Accelerando), though I'm a bit skeptical there is at least a bit of hope in that train of thoughts.

Any advance that gives us an edge will help, and friendly AI is a pretty big edge.

I've come to the conclusion that the only viability for long term human survival is to become something better than we are now in both body and mind; Human 2.0 if you will. If the technology existed today I would sign up for it in a heartbeat, even if it initially only entailed having a genetically modified peak human body with limited cybernetics.

[Junghalli]

For 'spacegun' type launch systems, would it be possible to them to launch crewed assets into space, or are they only good for unmanned cargos?

You can do manned mass driver capsules but you'd need a barrel hundreds or thousands of kilometers long for the acceleration to be low enough that it's tolerable to a human. I calculate a barrel 815 km long for orbital velocity (8 km/s) with a 4 G acceleration (sustained for 3.4 minutes), neglecting atmospheric and gravitational drag.

Think of the barrel as sort of like a railroad in a giant vacuum chamber.

Even a mass driver not suitable for humans would be an immense help though. Send up the bulk stuff relatively cheaply with mass drivers and reserve rockets for the humans and sensitive components that can't take the G forces.

[Sinewmire]

Has the Space Elevator concept been debunked? Basically a geostationary sattelite with a elevator/weight setup so that the energy needed to haul materials out of the atmosphere can be negated, or at least reduced, by a similar descending weight.

Ok, there are a few niggling problems, like the base of this cable needing to be miles wide or something.

[Stark]

Any advance that gives us an edge will help, and friendly AI is a pretty big edge.

I've come to the conclusion that the only viability for long term human survival is to become something better than we are now in both body and mind; Human 2.0 if you will. If the technology existed today I would sign up for it in a heartbeat, even if it initially only entailed having a genetically modified peak human body with limited cybernetics.

Welcome to science fiction circa 1960s?

[Junghalli]

I'm not aware of any recent developments indicating a space elevator to be unfeasible. IIRC a mass driver would probably be easier and better in many ways though; the structure is smaller, it can be built entirely on the ground, and you're not limited to shipping (IIRC) a small fraction of the mass of the structure at once like with a space elevator. Sikon had a good post on mass drivers vs space elevators a while back, it'll probably show up if you do a search.

[Junghalli]

Space elevators suck on Earth. They require huge amounts of incredibly strong material (the closest we have is carbon nanotubes, which are tiny.. and only strong in one way. I could see wind ripping them right apart) and are just god damn slow to travel; it'd take about a week on the elevator to get to GEO.

That's a pretty small throughput for such a huge investment.

My impression for quite a while has been that the space elevator concept is popular simply because it's well known rather than because it's actually particularly good compared to other less well-known approaches like mass drivers. Would you say that's a fair statement?

[Sinewmire]

My impression for quite a while has been that the space elevator concept is popular simply because it's well known rather than because it's actually particularly good compared to other less well-known approaches like mass drivers. Would you say that's a fair statement?

I couldn't comment on that trend, but it's certainly the case for me. I first read about Space Elevators in The Fountains of Paradise and later in Science of the Discworld.

[eion]

Has the Space Elevator concept been debunked? Basically a geostationary satellite with a elevator/weight setup so that the energy needed to haul materials out of the atmosphere can be negated, or at least reduced, by a similar descending weight.

Ok, there are a few niggling problems, like the base of this cable needing to be miles wide or something.

Des mentioned a few of the problems with geostationary elevators for Earth use, but here are a few more:

Radiation: The trip from the surface to GEO will require passing through the Van Allen Belts, so the cars will have to be shielded for passengers and certain cargos.

Construction difficulties: You have to build a Space Elevator from the top down, which means you need an enormous orbital industrial capacity, which sort of defeats the purpose of building a space elevator in the first place.

Limited Locations: You can only construct a Space Elevator on the equator of a planet.

Now a space fountain skirts around most of those. You build it like a self-erecting crane: the top station is built on the ground and then it lifts itself up to the desired orbital height while assembling the tether en route from materials sent up from the ground, it is an active structure, so the structural load is lessened which reduces the need for super strong materials, since you can built it to any orbital height there's no need for it to pass through radiation belts, you can build it wherever you please, and best of all there are actually near-term applications for closed loop projectile systems like the space fountain. You can build awesomely tall radio antennas, use an arched fountain to build bridges, or just create a tourist observation tower 25 miles up for the heck of it.

This would allow you to iron out any engineering issues without spending billions and billions on an tethered satellite that might break.

[Singular Intellect]

Well I recently finished some books on the Singularity-theory (Singularity is near, Accelerando), though I'm a bit skeptical there is at least a bit of hope in that train of thoughts.

According to the most well known projection, we'll have the capability to simulate an artificial human brain by the year 2029. At that point, the singularity will start take off because the artificial brain will be able to tweak itself magnitudes faster and easier than humans can tweak their biological ones, along with it's direct interface with the computing technology of the time. It won't take long for such a brain (if possessing the motivation) to start enhancing it's own intelligence well beyond human standards and start the snow balling effect for an intelligence explosion.

Personally, I'm hedging my hopes/bets on that scenario, because the current majority mass of human minds are just too stupid and flawed to seriously advance our society in positive ways.

[cosmicalstorm]

Yeah that's basically my dream scenario. Maybe even sooner if we're lucky, as with the HUGO project. I'm a bit disappointed that so many futurists and sci-fi writers don't seem to understand the implications of brain emulations and AI, even if it won't happen as fast as some people think.

[Guardsman Bass]

Well I recently finished some books on the Singularity-theory (Singularity is near, Accelerando), though I'm a bit skeptical there is at least a bit of hope in that train of thoughts.

According to the most well known projection, we'll have the capability to simulate an artificial human brain by the year 2029. At that point, the singularity will start take off because the artificial brain will be able to tweak itself magnitudes faster and easier than humans can tweak their biological ones, along with it's direct interface with the computing technology of the time. It won't take long for such a brain (if possessing the motivation) to start enhancing it's own intelligence well beyond human standards and start the snow balling effect for an intelligence explosion.

How exactly does the bolded work again? You act as if there's some clear path to superhuman intelligence once you have human-level intelligence that can engage in self-modification of its programming.

[LionElJonson]

Well I recently finished some books on the Singularity-theory (Singularity is near, Accelerando), though I'm a bit skeptical there is at least a bit of hope in that train of thoughts.

According to the most well known projection, we'll have the capability to simulate an artificial human brain by the year 2029. At that point, the singularity will start take off because the artificial brain will be able to tweak itself magnitudes faster and easier than humans can tweak their biological ones, along with it's direct interface with the computing technology of the time. It won't take long for such a brain (if possessing the motivation) to start enhancing it's own intelligence well beyond human standards and start the snow balling effect for an intelligence explosion.

How exactly does the bolded work again? You act as if there's some clear path to superhuman intelligence once you have human-level intelligence that can engage in self-modification of its programming.

If nothing else, it can design better and better circuit boards, while optimizing its own code. Besides, a properly designed AI will have vastly different thought processes to those a human; we've got a bunch of clunky relics of our evolutionary history distorting our worldview, and an AI wouldn't have those unless we intentionally programmed them in.

[Morilore]

If nothing else, it can design better and better circuit boards, while optimizing its own code. Besides, a properly designed AI will have vastly different thought processes to those a human; we've got a bunch of clunky relics of our evolutionary history distorting our worldview, and an AI wouldn't have those unless we intentionally programmed them in.

And suppose there are hard physical limitations to how sophisticated computer hardware may be made?

[Guardsman Bass]

According to the most well known projection, we'll have the capability to simulate an artificial human brain by the year 2029. At that point, the singularity will start take off because the artificial brain will be able to tweak itself magnitudes faster and easier than humans can tweak their biological ones, along with it's direct interface with the computing technology of the time. It won't take long for such a brain (if possessing the motivation) to start enhancing it's own intelligence well beyond human standards and start the snow balling effect for an intelligence explosion.

How exactly does the bolded work again? You act as if there's some clear path to superhuman intelligence once you have human-level intelligence that can engage in self-modification of its programming.

If nothing else, it can design better and better circuit boards, while optimizing its own code. Besides, a properly designed AI will have vastly different thought processes to those a human; we've got a bunch of clunky relics of our evolutionary history distorting our worldview, and an AI wouldn't have those unless we intentionally programmed them in.

But the initial AI, at least according to the point raised by SI, is based on a simulation of a human brain. How do you separate the cognitive biases from the cognitive process that you're simulating in that initial simulation?

And suppose there are hard physical limitations to how sophisticated computer hardware may be made?

Or processing power.

[Junghalli]

How exactly does the bolded work again? You act as if there's some clear path to superhuman intelligence once you have human-level intelligence that can engage in self-modification of its programming.

The simplest one would be increasing clock speed by simply adding more hardware and power.

[Formless]

How exactly does the bolded work again? You act as if there's some clear path to superhuman intelligence once you have human-level intelligence that can engage in self-modification of its programming.

The simplest one would be increasing clock speed by simply adding more hardware and power.

I don't think its that simple. If the simulation is accurate, than all speeding up the clock will do is make everything seem like it moves in bullet time from the perspective of the poor simulated person. It would NOT make them smarter, and it might even be psychologically damaging. Just think about the effect this would have on the brain's sleep cycle. That alone would be bad.

[Junghalli]

Everything going faster does make them smarter though, since they can run through more options in the same time. Even if the intelligence is nothing more than guess and check, running faster means it can do more guesses in the same time.

Beat me to it.

[Sarevok]

How exactly does the bolded work again? You act as if there's some clear path to superhuman intelligence once you have human-level intelligence that can engage in self-modification of its programming.

The simplest one would be increasing clock speed by simply adding more hardware and power.

Yeah. The brute force solution is not elegant but it is guaranteed to work.

[Formless]

How exactly does the bolded work again? You act as if there's some clear path to superhuman intelligence once you have human-level intelligence that can engage in self-modification of its programming.

The simplest one would be increasing clock speed by simply adding more hardware and power.

Yeah. The brute force solution is not elegant but it is guaranteed to work.

Everything going faster does make them smarter though, since they can run through more options in the same time. Even if the intelligence is nothing more than guess and check, running faster means it can do more guesses in the same time.

No its not. Can run multiple options =! WILL run multiple options. If you are simulating the brain of an uneducated moron, he won't think of doing that, he's just going to feel bored and alienated because his day subjectively feels like it takes a week. Because it DOES take a week because you've fucked up his sleep cycle. You are trying to apply computing principals to the brain when the brain does not work on the same principals as a computer. Intelligence is far more complicated than that.

[Sarevok]

No its not. Can run multiple options =! WILL run multiple options. If you are simulating the brain of an uneducated moron, he won't think of doing that, he's just going to feel bored and alienated because his day subjectively feels like it takes a week. Because it DOES take a week because you've fucked up his sleep cycle. You are trying to apply computing principals to the brain when the brain does not work on the same principals as a computer.

I would urge you to read Stargliders FAQ on artificial intelligence http://bbs.stardestroyer.net/viewtopic.php?f=5&t=136633. T o summarize there are many different methods that may lead to an AGI. Brain simulation is a unlikely path to yield success.

[Formless]

No its not. Can run multiple options =! WILL run multiple options. If you are simulating the brain of an uneducated moron, he won't think of doing that, he's just going to feel bored and alienated because his day subjectively feels like it takes a week. Because it DOES take a week because you've fucked up his sleep cycle. You are trying to apply computing principals to the brain when the brain does not work on the same principals as a computer.

I would urge you to read Stargliders FAQ on artificial intelligence http://bbs.stardestroyer.net/viewtopic.php?f=5&t=136633. T o summarize there are many different methods that may lead to an AGI. Brain simulation is a unlikely path to yield success.

Now we're not even disagreeing.

[Formless]

No its not. Can run multiple options =! WILL run multiple options. If you are simulating the brain of an uneducated moron, he won't think of doing that, he's just going to feel bored and alienated because his day subjectively feels like it takes a week.

That doesn't mean he isn't intelligent, it just means he is uneducated.

What, in practice, is the difference?

None, because learning capacity is an integral part of intelligence.

Because it DOES take a week because you've fucked up his sleep cycle. You are trying to apply computing principals to the brain when the brain does not work on the same principals as a computer. Intelligence is far more complicated than that.

If we actually apply computing principles, the smart thing is to have it automatically underclock itself when not in use, like computers do. When thinking (high CPU use), the clock rate automatically, and near instantly, is boosted. When not thinking (idle cpu), the clock rate instantly and automatically drops.

Except that unlike a computer, which only processes one thing at a time, the human brain is processing many things at once at all times. It has no central clock that you can speed up and slow down arbitrarily, and you would be hard pressed to do this anyway because its always thinking. It does not have times where it is idle in the same way a computer does.