What Hard SF Universe Could Beat the Federation?

[kinnison]

OK Junghali, reverse-engineering that sort of tech difference might be very difficult. Leave aside the enormous disparity in the number of people available to work on the problem (I doubt that there were a thousand physicists anywhere in the eighteenth century, and the HSF civilisation would probably have a hundred million) and consider speed.

Imagine an AI "brain" with roughly equal complexity to a human one, but working at computer speeds. A reasonable estimate for the speed difference is a factor of a million - in fact it's probably more; a human brain works at about 20Hz whereas a computer works at maybe 3GHz. But let's be conservative and say a million. Let's also say that most of the development work in the HSF civilisation is done by AI - a bit like the Culture in this respect.

Now; let's say that the HSF civilisation has extremely good computers, practical nanotech and space travel, which puts it at maybe 2050. Star Trek (Next Generation) is set in c. 2375, IIRC. This is a timespan of about 325 years, and technology and science have gone from somewhat worse than the HSF civ to Trek tech in that time.

So what is a useful ballpark figure for how long the reverse-engineering would take, to get the science behind it? (remember, they already know Trek tech is possible, and they may have some observational data.) A crude estimate might be 325 years divided by a million. How long is that?

About three hours. And once the science is done actually making the stuff is trivial.

A nanotech, transapient AI, type II civilisation is powerful. More powerful than anyone in our society can imagine.

Given all that, Ghost Rider, if one stipulates that the HSF civilisation is prevented by deific fiat from gaining the Trek technology - then of course it will lose. Trek has a hard fight of it, though - much harder than some people here think, IMHO.

[Darth Holbytlan]

kinnison
You do realize that a good chunk of the time required to develop scientific knowledge and technologies is in running experiments, right? Being a fast thinker isn't going to make creating or running the scientific experiments or prototype testing go any faster.

[18-Till-I-Die]

I understand what he's saying kinda. But the point is moot as the whole thread concept tacitly implies that they cant "steal" so-called "soft" scifi technology.

That being said, even if they DID it would STILL be of no use.

Because even IF they develop warp, somehow, they would not be able to retrofit their shits with it in any reasonable timeframe, because it would require almost complete redesigns. At least if i understand the way warp works, it would. The hypothetical ships of an advanced "lighthugger" civilization would require a great deal of stuff to be completely rebuilt from the ground up. And THEN they still have to communicate this to the rest of their civilization, without FTL communication. Which would take centuries.

Unless MULTIPLE planets of the HSF civ get Warp tech at the same time, which is more than a little rediculous.

[Teleros]

I understand what he's saying kinda. But the point is moot as the whole thread concept tacitly implies that they cant "steal" so-called "soft" scifi technology.

They can steal it, use it, and even maintain it if that means not violating physics as we know it, but they can't reverse-engineer it or maintain things that require technobabble to maintain.

And THEN they still have to communicate this to the rest of their civilization, without FTL communication.

Or they fly out using said warp drive. However as has been said, there is still the problem that they can't develop Fed warp drives.

Just to remind some people of the OP:

Should the hard 'verse capture Federation equipment, they can use it for as long as it works, or can be recharged/refueled/maintained using technological knowlege that follows the laws of physics. However, they cannot reverse engineer Federation tech, produce it, etc.

[Ariphaos]

-While most HSF civ's are doomed noone has really said how Trek could overcome its vulnerability to information warfare (in the case of extremely advanced HSF civs that span a large number of systems) or possibility of the HSF civ making alliances and/or trade agreements with neutral or friendly Trek civs.

A Type III civilization existing as a singular entity somewhat stretches the concept of HSF on several levels - if you have a Type III civilization, it would be very difficult to justify its existence as a single entity. But if that were the case, and you pulled the entirety of the Federation into the Type III's galaxy with immediate total war being declared first by the Type III civ (a set of protocols for what to do when a new star pops out of nowhere or something - sterilize and settle)...

[Ariphaos]

The above really extends to any significant multi-stellar empire, I should add. Civilizations need rapid communication lines to remain homogeneous.

A moderately silly question:

Would Trek weapons, warp drive, and such, actually make noise in space?

[NecronLord]

A moderately silly question:

Would Trek weapons, warp drive, and such, actually make noise in space?

Yes. But only if you're at TV-camera ranges from them.

[brianeyci]

About three hours. And once the science is done actually making the stuff is trivial.

You have a fundamental misunderstanding about how science and industry works. Firstly, quantity is not quality. It's entirely possible that a hundred million physicists are worse than a hundred. Secondly, making the "stuff" is not trivial. They would have to build an entirely new manufacturing infrastructure and a new energy source to a large scale (antimatter, and before you complain that fusion could provide the power necessary ships low on antimatter lose in Trek indicating their technology relies heavily on high energy density, a point NecronLord has made in other threads in the past). They would have to redesign everything, and if this took months or years or decades, any one of those, Star Trek wins.

Why assume that the AI wanking is hard physics at all? With all due respect to Starglider's profession, medical research still involves human beings. Computers are great at repetitive tasks, but leaps in physics usually involve insights, application and observation. You seem to think that science and mathematics is a straightforward endeavour whose only independent variable is time which you can manipulate with greater processing power. Computers absolutely suck at pattern recognition compared to human beings, and they can't even solve a fucking board game (Go) and get beaten by human children.

Also, Ender put it in good terms why reverse engineering doesn't work: it kills. Given the unstable nature of Trek technology, I can imagine massive industrial accidents attempting to duplicate it. Every industrial accident imaginable would happen because the AI being an AI would consider and attempt a large number of possibilities, even the ones easily eliminated by human experience, since it is a computer. I'm reminded of Terminator going through the phone book looking for every Sarah Connor, or again, Go. These industrial accidents would not be trivial and could result in large delays of months or realistically years or never, again giving victory to Trek.

I wonder how long it's been since you've taken science or math, but in general more time does not always equal more insight or knowledge and especially not more theory. Saying that hard SF universe could do well with enough time when it hasn't already is like saying with enough time bottom guy in the class could do as well as the genius.

[Nova Andromeda]

-While most HSF civ's are doomed noone has really said how Trek could overcome its vulnerability to information warfare (in the case of extremely advanced HSF civs that span a large number of systems) or possibility of the HSF civ making alliances and/or trade agreements with neutral or friendly Trek civs.

A Type III civilization existing as a singular entity somewhat stretches the concept of HSF on several levels - if you have a Type III civilization, it would be very difficult to justify its existence as a single entity.

-A HSF civ somewhere above Type II should do and is hard enough to predict the capabilities of. In terms of organization, you're restricting your thinking dumb ape (i.e., human) standards. A Type II civ should have more than enough accumulated science to have super intelligences and to upgrade their bodies and minds from whatever crap nature started them with. While light lag will make organization, research, etc. less efficient, it's nothing compared to the problems we dumb apes face when trying to organize things using our rather flawed brains.

To whoever it was saying they'll win via viruses; You want to explain how they'll magically compromise Star Trek ships?
...
they'd need to be able to take over the ship in the seconds it would take a compromised ship to jump a few light hours away for repairs.

-I addressed this previously. If there is any significant contact between the HSF civ and Trek before serious fighting begins then the HSF civ will have had plenty of opportunity to study Federation information technology. The HSF civ should also have a pretty good shot if it encounters Federation ships within light range and especially if it is able to capture a ship intact (studying a ship is far quicker than probing a functional one with various signals hoping to get feedback and then trying to exploit it). While the Federation could go from 0 to doomsday immediately (and avoid any contact) that's hardly their standard operating procedure, any HSF civ worth consideration isn't going to provoke them to do so, and it assumes the Federation just suddenly appears in the HSF civ's space and starts blasting away one day (not exactly a fair start).
-A ship, installation, etc. whose computer systems are compromised almost certainly isn't going to show any signs of a problem until it's way way too late (similar to modern spyware, viruses, worms, etc.).

[kinnison]

brianeyci:

Sure. Humans have insight, intuition, etc. This actually relates to the really hard question about consciousness and sentience; where does the self come from, or in other words how does consciousness arise?

One train of thought in this field is that sentience is an emergent phenomenon and that it's been evolved for - that an organism with sentience does better in the evolutionary contest than one with a brain and processing power that is just about the same, but with no self-knowledge. Or that self-knowledge is an accidental byproduct of being able to model the behaviour of others in your social group (if I do this he'll do that).

Either way, the solution to AI may well be to do some sort of evolutionary software development, on A-life constructs. This is already being done, with results that the experimenters describe as "eerie". All that's lacking then is enough hardware. And then you have your artificial sentience.

Or if that's too difficult, do something easier (which we nevertheless don't yet know how to do, I admit); simply upload an existing meat brain's contents into the computing matrix.

We don't actually know that a sufficiently powerful and complex computer won't develop sentience on its own. Sheer power won't do it, most likely, but a sufficiently complex network might. For a fictional treatment of this idea, see Heinlein starting in the 50s.

As for the grunt work of redesign, this really is a trivial problem. Say that a large industrial complex (say present day America) would take a hundred years to do all the redesign work, CAD, drawing etc. together with all the false starts. This collapses, at computer speeds, to a hundred hours - less than a week - and the manufacturing steps are similarly speeded up if you have good enough nanotech. Say two or three hundred hours, then - a fortnight. Can the Feds kill a quadrillion people, and destroy all their industry, in a fortnight? I don't think so. Will they, without warning? I think so even less.

One last thing: Say the HSF civilisation captures one Trek ship. Trek technology includes replicators, does it not?

[brianeyci]

brianeyci:

Sure. Humans have insight, intuition, etc. This actually relates to the really hard question about consciousness and sentience; where does the self come from, or in other words how does consciousness arise?

One train of thought in this field is that sentience is an emergent phenomenon and that it's been evolved for - that an organism with sentience does better in the evolutionary contest than one with a brain and processing power that is just about the same, but with no self-knowledge. Or that self-knowledge is an accidental byproduct of being able to model the behaviour of others in your social group (if I do this he'll do that).

Either way, the solution to AI may well be to do some sort of evolutionary software development, on A-life constructs. This is already being done, with results that the experimenters describe as "eerie". All that's lacking then is enough hardware. And then you have your artificial sentience.

Explain Go. If you can't then you already admit humans have an advantage over computers, a rather significant one in pattern recognition which is essential in developing new theoretical science. What completely new theoretical fields has a computer opened up, ever?

Or if that's too difficult, do something easier (which we nevertheless don't yet know how to do, I admit); simply upload an existing meat brain's contents into the computing matrix.

Mind uploading is the ultimate wanking.

We don't actually know that a sufficiently powerful and complex computer won't develop sentience on its own. Sheer power won't do it, most likely, but a sufficiently complex network might. For a fictional treatment of this idea, see Heinlein starting in the 50s.

Since we're going by the definition of hard science fiction being fiction that does not violate physics, appealing to all the so-called hard science fiction doesn't apply in this thread since they routinely violate physics with faster than light travel.

As for the grunt work of redesign, this really is a trivial problem. Say that a large industrial complex (say present day America) would take a hundred years to do all the redesign work, CAD, drawing etc. together with all the false starts. This collapses, at computer speeds, to a hundred hours - less than a week - and the manufacturing steps are similarly speeded up if you have good enough nanotech. Say two or three hundred hours, then - a fortnight. Can the Feds kill a quadrillion people, and destroy all their industry, in a fortnight? I don't think so. Will they, without warning? I think so even less.

What part of time and processing power is not the only independent variable do you not understand? This is getting ridiculous.

They don't have to kill a quadrillion people. They only have to do enough damage to force a surrender. I don't know where you're pulling this quadrillion number out of your ass from, and all these numbers out of your ass. Why assume that nanotechnology works at all? Do you imagine huge piles of nanorobots able to fashion nearly anything and everything? I do not. I imagine that even if there's a 1% failure rate in nanorobots it will result in catastrophic failure. Macroscopic construction will always be superior to microscopic construction.

One last thing: Say the HSF civilisation captures one Trek ship. Trek technology includes replicators, does it not?

Why assume they can capture a Trek ship at all when the Trek ship will see them coming light years away? Replicators are just another manufacturing tool, and your "nanorobots" could do just as good if they even fucking work. So replicators provide no real advantage.

What the fuck.

[Darth Wong]

Imagine an AI "brain" with roughly equal complexity to a human one, but working at computer speeds. A reasonable estimate for the speed difference is a factor of a million - in fact it's probably more; a human brain works at about 20Hz whereas a computer works at maybe 3GHz. But let's be conservative and say a million. Let's also say that most of the development work in the HSF civilisation is done by AI - a bit like the Culture in this respect.

This would be a great argument if the AI brain actually knew how everything worked beforehand, so all it had to do was crunch physics and engineering principles that were already well-developed and well-understood. Since this is not the case, you're full of shit.

Now; let's say that the HSF civilisation has extremely good computers, practical nanotech and space travel, which puts it at maybe 2050. Star Trek (Next Generation) is set in c. 2375, IIRC. This is a timespan of about 325 years, and technology and science have gone from somewhat worse than the HSF civ to Trek tech in that time.

So what is a useful ballpark figure for how long the reverse-engineering would take, to get the science behind it? (remember, they already know Trek tech is possible, and they may have some observational data.) A crude estimate might be 325 years divided by a million. How long is that?

About three hours. And once the science is done actually making the stuff is trivial.

Reverse-engineering is not like taking a known design and copying it, moron. And what about all of the engineering processes required? What about the materials? Suppose you know exactly how a modern nuclear bomb works, but you lack the technology to refine uranium? You know the science, so it should take about three hours with a powerful computer, right?

[Beowulf]

We used to think chess was too difficult for computers to be able to beat a grandmaster. And the programs we use can't even be described as AI.

Prove the four color theorem without the use of a computer. Can't feasibly be done. All proofs so far involve brute forcing the possible maps, and determining that all of them can be colored correctly.

[Darth Wong]

We used to think chess was too difficult for computers to be able to beat a grandmaster. And the programs we use can't even be described as AI.

Prove the four color theorem without the use of a computer. Can't feasibly be done. All proofs so far involve brute forcing the possible maps, and determining that all of them can be colored correctly.

And what does this prove, exactly? Are you saying that a sufficiently fast computer can just magically insta-learn science and engineering without having to do all of the time-consuming work? What's it supposed to do, derive everything from first principles?

[18-Till-I-Die]

Again i would point out that the whole point is rather moot. Even IF they get a few FTL capable ships it wont ammount to crap against the hundreds (thousands?) the Feds already have, and the UFP's capacity to build more at a considerably faster pace...seeing as they already have an extensive infrastructure built to do just that.

And they STILL have no FTL sensors and thus no way to even FIND the Federation's 150 worlds out of the numberless planets in the Milky Way.

And they STILL have no FTL communication to even tell their sister colonies about the discoveries.

And they STILL cant produce these new ships at any reasonable pace because it would require a massive redesign (antimatter would be the easiest problem to solve in that case) and that, even with all the AIs in the world, would take years, maybe months at the low end.


Saying "They can figure it out" is fine, but figuring out how it works and building thousands of duplicates are two different things.


And all while this is happening what is the Federation doing? Sitting around? Why allow some of their stuff to even fall into the HSF hands? Why not drop some antimatter on whatever site is being used to reverse-engineer the stuff?

The Feds aren't the most proactive folks but they aren't retards either.

[Teleros]

And they STILL have no FTL sensors and thus no way to even FIND the Federation's 150 worlds out of the numberless planets in the Milky Way.

And they STILL have no FTL communication to even tell their sister colonies about the discoveries.

If they capture a Fed ship they can use the ship's sensors for as long as they work, as well as use them as messenger boats - assuming they're not destroyed en route of course.

[Enigma]

And they STILL have no FTL sensors and thus no way to even FIND the Federation's 150 worlds out of the numberless planets in the Milky Way.

And they STILL have no FTL communication to even tell their sister colonies about the discoveries.

If they capture a Fed ship they can use the ship's sensors for as long as they work, as well as use them as messenger boats - assuming they're not destroyed en route of course.

If they capture one is one thing, but flying a Fed ship and using anything else onboard is another. How will they figure out the controls on the starship? For all we know they could accidentally vent all oxygen from the ship while trying to access the sensors.

[Teleros]

Well if they can capture it they're probably going to end up with a few of the crew, which'll help matters. But yes you're absolutely right that'll be yet another problem for them to overcome.

[Beowulf]

We used to think chess was too difficult for computers to be able to beat a grandmaster. And the programs we use can't even be described as AI.

Prove the four color theorem without the use of a computer. Can't feasibly be done. All proofs so far involve brute forcing the possible maps, and determining that all of them can be colored correctly.

And what does this prove, exactly? Are you saying that a sufficiently fast computer can just magically insta-learn science and engineering without having to do all of the time-consuming work? What's it supposed to do, derive everything from first principles?

Actually, it was more of a response to Brianeyci's idiotic "Explain Go." remark.

[Starglider]

A moderately silly question:

Would Trek weapons, warp drive, and such, actually make noise in space?

Yes, but you have to be using subspace microphones.

Computers are great at repetitive tasks, but leaps in physics usually involve insights, application and observation. You seem to think that science and mathematics is a straightforward endeavour whose only independent variable is time which you can manipulate with greater processing power. Computers absolutely suck at pattern recognition compared to human beings, and they can't even solve a fucking board game (Go) and get beaten by human children.

This is a software problem. Solving it is actually a lot more plausible than building all the incredible industrial and space infrastructure that people are invoking for this 'hard sci fi' civ.

The lower bound for what AIs could do is the same things humans can do, but at one hundred million times the speed, 24/7 with no breaks for sleep/food/etc, with total concentration, with instant recall of vast databases, with telepathic-equivalent communication with all other AGIs working on the problem etc etc. Transhuman AGI is geuniely very, very scary. However there are two problems with this;

1) Very few 'hard sci-fi civs' (in sci-fi novels) actually have it.
2) You're still bottlenecked on doing physical experiments. Sure you can analyse the data near-instantly, but building the equipment can't be significantly accelerated (though nanotech can help a lot here). Simulations cut down your prototyping cycle a lot, but they won't eliminate it, not given the sparse information available here.

Every industrial accident imaginable would happen because the AI being an AI would consider and attempt a large number of possibilities, even the ones easily eliminated by human experience, since it is a computer. I'm reminded of Terminator going through the phone book looking for every Sarah Connor, or again, Go. These industrial accidents would not be trivial and could result in large delays of months or realistically years or never, again giving victory to Trek.

A transhuman AI would not do this. This sounds like some bizarre and desperate attempt to connect a very basic general AI to massive nanomanufacturing assets in the hope that it will come up with something that works. There is no inherent reason why AIs are less inventive/sensible/whateve than humans - in fact they can in principle have a lot more of these things, given adequate software engineering.

Explain Go. If you can't then you already admit humans have an advantage over computers, a rather significant one in pattern recognition which is essential in developing new theoretical science.

Software problem.

Mind uploading is the ultimate wanking

Don't be ridiculous. Mind uploading is /far/ easier than building a solar-system-spanning industrial infrastructure. In fact it's almost certainly easier than building a self-supporting mars colony. Mind uploading research is already progressing at a fast pace; brain simulation is getting more and more accurate, and 3D scanning is rapidly advancing. It's completely physical plausible, several basic methods have been worked out, the engineering is being done. Non-destructive uploading is rather harder than destructive uploading, but that's a luxury feature.

They don't have to kill a quadrillion people. They only have to do enough damage to force a surrender.

True, but the historical examples on that are mixed. Japan surrended after two nuclear bombs, a large amount of conventional bombing and a blockade of critical supplies. Nazi Germany didn't surrender until the country was completely occupied and ruined. Depending on the exact details of the civ it may plausibly fight to the last outpost or it may surrender as soon as it's clear that a war will be worse for it in the long run than a surrender.

One train of thought in this field is that sentience is an emergent phenomenon and that it's been evolved for - that an organism with sentience does better in the evolutionary contest than one with a brain and processing power that is just about the same, but with no self-knowledge

Well it's emergent in the sense that evolution produced it without an explicit plan to do so. The explicit human concept of self is closely tied to social modelling, but sentience in the sense of at least a minimal level of self-understanding is critical to general intelligence. However it doesn't have to be 'emergent'; you can explicitly design it in to an AI system if you know what you're doing (which regrettably we don't, for the most part, at present).

We don't actually know that a sufficiently powerful and complex computer won't develop sentience on its own. Sheer power won't do it, most likely, but a sufficiently complex network might. For a fictional treatment of this idea, see Heinlein starting in the 50s.

Actually yes we do. I've worked with a number of people who've analysed this in depth. The kind of dynamics you need to get open ended self-improving behaviour are actually fairly specific, most software systems do not qualify.

This collapses, at computer speeds, to a hundred hours - less than a week - and the manufacturing steps are similarly speeded up if you have good enough nanotech.

Nanotech has a lot of potential, but it can't necessarily build a capship in a day. The big problem with nanoassembly is that it's fundamentally limited to low-energy processes, because the working parts are so delicate. Trek technology relies critically on lots of bizarre materials (e.g. dilithium) that aren't necessarily compatible with nanoassembly. For example the TNG tech manual talks about the warp coils requiring many months to cast - a process which you probably can't finnesse with nanotech. Trek also has the advantage of being able to use force fields and artificial gravity in their construction process. DW's analogy of the difficulty of refining uranium is a good one and most likely applies to most Trek technology (anything involving subspace seems to rely on bizarre and hard-to-make materials).

One last thing: Say the HSF civilisation captures one Trek ship. Trek technology includes replicators, does it not?

Replicators seem to be strictly limited in what they can do (no replicating capships) and highly energy intensive.

Are you saying that a sufficiently fast computer can just magically insta-learn science and engineering without having to do all of the time-consuming work? What's it supposed to do, derive everything from first principles?

We can't actually be sure that this isn't possible. It may be that we already have enough experimental evidence to solve all of physics and we're just not intelligent enough to do so. Certainly it does seem likely to me that a superintellligence going over our particle physics experiment logs would generate additional findings that humans couldn't see.

However that isn't relevant in this case. The hard sci fi civ's past experiments are no longer valid. If they want to crack the Trek physics that suddenly apply, they're going to have to do a whole load of new experiments - for which the don't even have appropriate apparatus. For the Trek magic materials, there may literally be no substitute for trial and error - synthesising tens of thousands of compounds and alloys and exposing them to assorted wacky conditions to see if there's a new physical effect in operation. This can't even be done in parallel, because you probably have to build instruments that can detect the appropriate subspace phenomena before you can start testing materials for warp coils and shield generators.

And all while this is happening what is the Federation doing? Sitting around? Why allow some of their stuff to even fall into the HSF hands? Why not drop some antimatter on whatever site is being used to reverse-engineer the stuff?

The Feds aren't the most proactive folks but they aren't retards either.

Agree. If it's a genuine 'total war scenario' the Federation aren't going to be so stupid as to let their stuff get captured and reverse engineered at lesuire. Almost all of the usual pro-Wars arguments against ridiculous Trektard 'we'll just trivially reverse engineer Imperial technology' apply.

[Darth Wong]

One other problem with the HSF-wankers is that they keep assuming a highly advanced, highly developed HSF civilization would have an extremely robust industrial infrastructure, designed to withstand severe damage from alien attackers and continue pumping out military equipment and supplies.

Why the fuck should this be the case? If this HSF civ has so expansively colonized its own system, it has almost certainly come to the conclusion that FTL travel is impossible, so it will not be attacked by fleets of alien warships. It would certainly not design defenses against something which appears to be impossible and which has never happened in its long history.

Moreover, when I see people assuming that its infrastructure would be highly robust, I have to ask what they're smoking. The natural trend in any kind of industry is toward greater efficiency, not redundant capabilities (which are necessary for the kind of robustness we're talking about). With no apparent external threat, there is no reason whatsoever to believe that they would have the kind of redundancies which are necessary in order to secure their industrial infrastructure from attack. Most likely, it would actually be extremely fragile: only minor pinpoint strikes would be necessary to very seriously cripple it.

[Ariphaos]

If they capture one is one thing, but flying a Fed ship and using anything else onboard is another. How will they figure out the controls on the starship? For all we know they could accidentally vent all oxygen from the ship while trying to access the sensors.

If they capture a ship with its crew (alive or dead but not overly damaged), there is little stopping such a civilization from taking their brains apart and acquiring the relevant information. Synapse and dendrite modeling, is well within the bounds of HSF.

[Enigma]

If they capture one is one thing, but flying a Fed ship and using anything else onboard is another. How will they figure out the controls on the starship? For all we know they could accidentally vent all oxygen from the ship while trying to access the sensors.

If they capture a ship with its crew (alive or dead but not overly damaged), there is little stopping such a civilization from taking their brains apart and acquiring the relevant information. Synapse and dendrite modeling, is well within the bounds of HSF.

What is to stop the senior crew from just self-destructing or make it very hard for the invaders to commandeer the ship? There is a lot of damage the Fed crew can do before the HSF invaders reach the bridge or the engineering section.

[Ariphaos]

One other problem with the HSF-wankers is that they keep assuming a highly advanced, highly developed HSF civilization would have an extremely robust industrial infrastructure, designed to withstand severe damage from alien attackers and continue pumping out military equipment and supplies.

Because 'military equipment' and 'power gathering / redirection equipment' are synonymous in such a scenario. I've mentioned this already, is there some reason this would not be the case?

Why the fuck should this be the case? If this HSF civ has so expansively colonized its own system, it has almost certainly come to the conclusion that FTL travel is impossible, so it will not be attacked by fleets of alien warships. It would certainly not design defenses against something which appears to be impossible and which has never happened in its long history.

Why would it need to? A near-type II civilization is handling hundreds of yottawatts of energy, using quadrillions to quintillions of components and millions to billions of entirely separate power distribution networks, which likely need to be able to switch targets relatively rapidly - if infrequently.

You need some seventy million rings of km-wide solar collectors / emitters in order to do this, each of which would contain at least as many individual devices. Is there something non-HSF about using advanced solar cells and FET lasers to this effect?

Moreover, when I see people assuming that its infrastructure would be highly robust, I have to ask what they're smoking. The natural trend in any kind of industry is toward greater efficiency, not redundant capabilities (which are necessary for the kind of robustness we're talking about). With no apparent external threat, there is no reason whatsoever to believe that they would have the kind of redundancies which are necessary in order to secure their industrial infrastructure from attack. Most likely, it would actually be extremely fragile: only minor pinpoint strikes would be necessary to very seriously cripple it.

How could such a network be constructed in the first place if it were not exceptionally robust? Robustness isn't a design -goal- here, it's a necessity of the swarm itself. Vulcan orbits are not perfectly stable, and the occasional extrasolar meteor needs to be accounted for.

Above and well beyond that, no single point in the power network can be capable of handling more than an order of magnitude or so more energy than an individual collector. IF it -weren't- robust, then it would require the same sorts of magical materials that are commonplace in soft sci-fi.

What sort of structure could possibly harness and distribute any significant fraction of a star's energy without being composed of millions to billions of largely independent subsystems, with each of those being composed of just as many subsystems of their own? What sort of focus point could possibly exist to handle more than a trillionth of the sun's total output?

Even if you could concoct some super-magical material to do so, why would you?

[brianeyci]

This is a software problem. Solving it is actually a lot more plausible than building all the incredible industrial and space infrastructure that people are invoking for this 'hard sci fi' civ.

The lower bound for what AIs could do is the same things humans can do, but at one hundred million times the speed, 24/7 with no breaks for sleep/food/etc, with total concentration, with instant recall of vast databases, with telepathic-equivalent communication with all other AGIs working on the problem etc etc. Transhuman AGI is geuniely very, very scary. However there are two problems with this;

1) Very few 'hard sci-fi civs' (in sci-fi novels) actually have it.
2) You're still bottlenecked on doing physical experiments. Sure you can analyse the data near-instantly, but building the equipment can't be significantly accelerated (though nanotech can help a lot here). Simulations cut down your prototyping cycle a lot, but they won't eliminate it, not given the sparse information available here.

We're already going by the idea that what is generally called hard science fiction isn't that hard at all. You say it's a software problem, but if it's an unsolvable software problem it's still a demonstration that in the real world, with many more choices and variables than a Go game, AI would fail miserably.

A transhuman AI would not do this. This sounds like some bizarre and desperate attempt to connect a very basic general AI to massive nanomanufacturing assets in the hope that it will come up with something that works. There is no inherent reason why AIs are less inventive/sensible/whateve than humans - in fact they can in principle have a lot more of these things, given adequate software engineering.

People suggested reverse engineering. Reverse engineering kills, AI or no AI. If an AI does not explore all possible routes, at the very least it will explore dangerous routes unknown to it since it lacks the background knowledge, just like his human counterparts would reverse engineering something they have no idea of its limits or dangers or safety margins. What do you imagine an AI doing and why is it different than what I imagine it?

Why believe that transhuman AI is hard science at all? AI wankers also forget the human factor. It is hard to imagine any civilization putting its fate in the hands of intelligent machines, and even the most stupid of stupid would want control and use machines as tools rather than overlords.

You also got it backwards. I don't have to show that AI's are less inventive, sensible or intuitive than humans. I'm waiting to see why I should believe they can be more. Beowulf brought up chess, but I don't see why that's any different than saying when science was immature it believed black people were stupid. AI people may not like to admit it, but their field has failed to produce its holy veil of machine sentience (as you guys admit what we have now is not close to what you would like to call AI) and the AI field is mature. This may be all there is to it. Why do you think universities scaled back on AI labs in the past two decades compared to its heyday? Sorry.

Don't be ridiculous. Mind uploading is /far/ easier than building a solar-system-spanning industrial infrastructure. In fact it's almost certainly easier than building a self-supporting mars colony. Mind uploading research is already progressing at a fast pace; brain simulation is getting more and more accurate, and 3D scanning is rapidly advancing. It's completely physical plausible, several basic methods have been worked out, the engineering is being done. Non-destructive uploading is rather harder than destructive uploading, but that's a luxury feature.

So explain to me why we can make a Mars colony now, with current day technology, if the political and economic factors were aligned, but we can't upload a mind now. Is it just a lag in certain fields of study? I feel perfectly justified calling it wanking too, because all mind uploading does is preserve dying geniuses. To hear mind uploaders they seem to think Einsteins will result as a result of uploading ordinary people. This just won't happen: best case scenario they'll continue to be selfish human or alien beings (assuming HSF isn't human) and demand a robot body and enjoy pleasures they could never dream of. Not work on problems to advance their civilization. Mind uploading solves nothing.

True, but the historical examples on that are mixed. Japan surrended after two nuclear bombs, a large amount of conventional bombing and a blockade of critical supplies. Nazi Germany didn't surrender until the country was completely occupied and ruined. Depending on the exact details of the civ it may plausibly fight to the last outpost or it may surrender as soon as it's clear that a war will be worse for it in the long run than a surrender.

Well then the Federation will just have to disable its war making ability. Fighting to the last outpost will never be necessary, because the Federation won't ever land troops on foreign soil unless it absolutely has to (they use Klingons for that.)