What if we built ships controlled by artificial intelligences rather than human crews, assuming we could build sufficiently advanced AIs? They could do the fighting, and then phone home when the enemy has been turned to dust.

Computers can be hardened to stand significant amounts of radiation, excessive G-forces; they don't require oxygen or water; they're easy to decentralize or make redundant for robustness; they don't all die as soon as a stray railgun pellet puts a hole in a bulkhead, and it's a lot easier for an automated fleet with support vessels to replace damaged computers than to replace damaged humans.

They also have no creativity. Sure, you could program strategies and tactics into them, but there's always the chance that your enemies will come up with something new and obliterate your fancy computer system. Sending humans on board would give them a chance to do that back to the enemy.

You are right on the creativity point. Even so-called "random number generators" have a pattern to them eventually, and if you learn the pattern and the algorithms you can literally predict its every move. Of course they could also change things up from time to time to keep that from happening.

What about remote piloting? You'd basically get the best of both worlds from that; no life support, no g-force considerations, and expendable, but still controlled by an adaptable, unpredictable human mind tucked safely away on a carrier.

@guest_swithich_-, on Jul 11 2008, 01:16 AM, said in On Relative Sizes of Spacecraft:

I think that is why a moon base (the moon has large quantities of Plutonium and Helium-3, a very unstable atom that radiates energy...in process needed...or that is the theory) is so important. Also it would take less energy to launch a vehicle from the moon...because less gravity, plus the moon gives an extra speed boost and is close to anywhere humans might want to go.

Actually, this isn't true in the grand scheme of things. The moon is only 238,854 miles away from the Earth, which is piffle when you consider that Mars at its closest is 36 million miles away from the Earth, not to mention the fact that a space craft would actually have to go farther than that because the Earth and Mars are in constant orbit around the Sun. You'd also only get the benefit of a gravitational slingshot (I assume this is what you meant by "extra speed boost") if you launched your spacecraft off the Earth, as you can't expend very much Delta-V turning your ship around 180 degrees so it can perform a full orbit of the moon before heading to it's final destination. This wouldn't even be worth it in the long run due to the moon's small gravity.

I think we should first build a space elevator so we can lift all the supplies and materials we need to build an interplanetary spacecraft into space, so we don't have to expend fuel getting it off the Earth/Luna, or expend fuel lifting parts into space (e.g. nuclear reactor) using rockets. You could also use the space elevator as a slingshot, since it performs a full orbit of the Earth every 24 hours. (It's in geo-stationary orbit. If it wasn't, it'd wrap itself around the Earth and come down again in a flaming ball of carbon fiber.) I say we should draw up the plans for a space elevator first, and the plans for a battlecruiser next.

@archon, on Jul 11 2008, 08:19 AM, said in On Relative Sizes of Spacecraft:

You are right on the creativity point. Even so-called "random number generators" have a pattern to them eventually, and if you learn the pattern and the algorithms you can literally predict its every move. Of course they could also change things up from time to time to keep that from happening.

What about remote piloting? You'd basically get the best of both worlds from that; no life support, no g-force considerations, and expendable, but still controlled by an adaptable, unpredictable human mind tucked safely away on a carrier.

The problem with this is that you have to take the vast distances of space into consideration. At 1 lightsecond (the distance between the Earth and the Moon) a remote pilot would be getting a video feed/sensor input that was 1 second old, and all the commands he'd give to his spacecraft would take one second to get to the craft and two seconds for the feedback from his command to get back to him. If he's piloting a fighter craft in the thick of a space fighter dogfight or through AF (anti-fighter) fire from capital ships, he's going to be space dust before he can curse.

Quote

They also have no creativity. Sure, you could program strategies and tactics into them, but there's always the chance that your enemies will come up with something new and obliterate your fancy computer system. Sending humans on board would give them a chance to do that back to the enemy.

And this assumption that computers will never be able to do all the things a human brain can do comes from where? Unless, of course, there really is a 'ghost in the machine', so to speak, and that creativity is intrinsically tied to a human soul. Which I'm not going to get into, because whether or not a computer can have a soul is invariably going to be the personal choice of the author or writer or whatever.

I also argue that logistics and raw power have as much to do with winning a war as 'creativity'. Maybe you can come up with that one out of a billion brilliant strategy that'll turn the tides of a war, but my ships still accelerate faster, are more resistant to damage, can operate for longer periods of time, and have more space for weapons and military equipment in lieu of life support and radiation shielding. In the end, you aren't going to be able to win a fistfight against a tank, no matter how 'creative' you are.

I think there have to be better reasons to justify putting people on ships. Something unique beyond the rather cliche 'people have imaginations' could make for very interesting storytelling.

Quote

Even so-called "random number generators" have a pattern to them eventually

This is only true insofar as you're talking about pseudorandom number generators implemented using software. You can get truly random numbers through certain natural phenomenon, such as radioactive decay.

This post has been edited by UE_Research & Development: 11 July 2008 - 11:04 AM

@jacabyte, on Jul 11 2008, 03:09 PM, said in On Relative Sizes of Spacecraft:

The problem with this is that you have to take the vast distances of space into consideration. At 1 lightsecond (the distance between the Earth and the Moon) a remote pilot would be getting a video feed/sensor input that was 1 second old, and all the commands he'd give to his spacecraft would take one second to get to the craft and two seconds for the feedback from his command to get back to him. If he's piloting a fighter craft in the thick of a space fighter dogfight or through AF (anti-fighter) fire from capital ships, he's going to be space dust before he can curse.

True; I guess I'm operating under the highly shaky premise of combat taking place very near the mothership. Or hedging on inventing an ansible.

I'd never thought of using half-life decay for random numbers; that's quite crafty.

The problem with A.I.s are hacking. Theoretically, if you had really good computer systems on your ships, you could send hackers up into space, hack into the enemy ship's systems, and disable them. In case of a A.I. ship, if the computer goes off, your ship is going to be disabled until someone goes on board and flicks the metaphorical on switch again. Sure, you could have auto-restart mechanisms, but a good hacker would take those down. Heck, he'd probably go and delete the whole operating system, which would really screw over an enemy system. To counter this, you'd want other people up in space to stop them, because an A.I. isn't going to stop a clever hacker.

Time is an issue here, of course, so it'd be likely that the hackers would be operating out of your most defended ships, such as your capitals, so they'd have time to get the job done.

@archon, on Jul 11 2008, 07:19 AM, said in On Relative Sizes of Spacecraft:

Even so-called "random number generators" have a pattern to them eventually, and if you learn the pattern and the algorithms you can literally predict its every move.

Correct. Modern video games don't actually generate random numbers. The have a huge spreadsheet with thousands of different numbers, and go down it. This enables synchronization in multiplayer.

@jacabyte, on Jul 11 2008, 08:09 AM, said in On Relative Sizes of Spacecraft:

Actually, this isn't true in the grand scheme of things. The moon is only 238,854 miles away from the Earth, which is piffle when you consider that Mars at its closest is 36 million miles away from the Earth, not to mention the fact that a space craft would actually have to go farther than that because the Earth and Mars are in constant orbit around the Sun. You'd also only get the benefit of a gravitational slingshot (I assume this is what you meant by "extra speed boost") if you launched your spacecraft off the Earth, as you can't expend very much Delta-V turning your ship around 180 degrees so it can perform a full orbit of the moon before heading to it's final destination. This wouldn't even be worth it in the long run due to the moon's small gravity.

I think we should first build a space elevator so we can lift all the supplies and materials we need to build an interplanetary spacecraft into space, so we don't have to expend fuel getting it off the Earth/Luna, or expend fuel lifting parts into space (e.g. nuclear reactor) using rockets. You could also use the space elevator as a slingshot, since it performs a full orbit of the Earth every 24 hours. (It's in geo-stationary orbit. If it wasn't, it'd wrap itself around the Earth and come down again in a flaming ball of carbon fiber.) I say we should draw up the plans for a space elevator first, and the plans for a battlecruiser next.

Space elevators...oh god, someone save me. Really, I don't even want to both arguing this point.

The topic of the moon, the primary use of fuel for any vehicle is leaving earth atmosphere and gravitational well. Thus if you have a moon base, the gravity is far less. The sling shot effect could still be gained by a sling shot around the Earth and then back around the moon, but the key gain comes from that fact that anything you build and launch from the moon only has 1/6th of the gravity to deal with. Thus you can build huge ships or vehicles and easily launch them into space.

On the slingshot effect, those orbital dynamics guys (the ones with Ph.D.s in physics) are amazing, there are literally "best paths" in the solar system and universe, and they literally will analyses hundreds or thousands of objects gravitational effects to get a spacecraft to a place the fastest. Amazing stuff really.

@joshtigerheart, on Jul 11 2008, 11:16 AM, said in On Relative Sizes of Spacecraft:

The problem with A.I.s are hacking. Theoretically, if you had really good computer systems on your ships, you could send hackers up into space, hack into the enemy ship's systems, and disable them. In case of a A.I. ship, if the computer goes off, your ship is going to be disabled until someone goes on board and flicks the metaphorical on switch again. Sure, you could have auto-restart mechanisms, but a good hacker would take those down. Heck, he'd probably go and delete the whole operating system, which would really screw over an enemy system. To counter this, you'd want other people up in space to stop them, because an A.I. isn't going to stop a clever hacker.

Time is an issue here, of course, so it'd be likely that the hackers would be operating out of your most defended ships, such as your capitals, so they'd have time to get the job done.

I'm a disagree on this point. You can only hack something if you have access to some sort of source code. Thus a password (basic and inventive) should keep most people out (as long as they can't pull the bios battery ). Seriously, you can build a system using encription that is always more powerful than the computer you are facing off against. Basically use RSA encryption and pick a large exponent (you know 10000 or 100000 something that makes a really big number) bingo you should be good.

From what I know of hacking, it mostly occurs on system that are meant to be entered by other people. So if you have a network and you have employees going in and out of the network you are open to hacking (as are unprotected and even protected wireless networks). But on a ship that has an internal computer system, I don't think so. Now on the AI fighter idea, I could see attempts at hacking, but you could always give the ship ability to override the control and have a default of return to the mothership (I.E. like the vehicles that drive themselves in the darpa challenge, have an overriding logical AI). Aka, someone hacks it, makes it do something illogical, it just shuts down and defaults to return to mothership.

Maybe that last part doesn't make any sense, but that is what I would try.

Perhaps I'm missing something, but I don't see how having a person on a fighter would make it that much less vulnerable to hacking. Even if you have a person pushing the buttons, at the end of the day, he/she's still relying on computers to carry out the orders.

On a totally different subject, what about The Moon is a Harsh Mistress -esque catapults as an alternative to an elevator?

The real problem with computerized A.Is with weapons is one hit from an EMP would render them as useless as a one-legged man in an ass kicking contest.

The other issue with computer-controlled fighters that hasn't been fully articulated as that with enough observation, they are predictable enough to have an advantage in aiming your fire. Unless we reach the level of sophistication of the Sci-Fi AI or somesuch (which I doubt will happen anytime soon, if at all), computer-controlled fighters are toast against an observant enemy.

@joshtigerheart, on Jul 11 2008, 02:16 PM, said in On Relative Sizes of Spacecraft:

The problem with A.I.s are hacking. Theoretically, if you had really good computer systems on your ships, you could send hackers up into space, hack into the enemy ship's systems, and disable them. In case of a A.I. ship, if the computer goes off, your ship is going to be disabled until someone goes on board and flicks the metaphorical on switch again. Sure, you could have auto-restart mechanisms, but a good hacker would take those down. Heck, he'd probably go and delete the whole operating system, which would really screw over an enemy system. To counter this, you'd want other people up in space to stop them, because an A.I. isn't going to stop a clever hacker.

Time is an issue here, of course, so it'd be likely that the hackers would be operating out of your most defended ships, such as your capitals, so they'd have time to get the job done.

Nonsense. As long as the computer is self-contained and can only take such simple commands as "Attack this ship", "Return to base", "Protect this ship" and so on and so forth, it can't be fully shut down/self-destructed/etc. As long as good encryption protocols are used and the IDs of friendly ships are hardcoded into the fighter, it won't attack friendlies. It is possible to encrypt data to the point where breaking the encryption requires a lucky guess, and bruteforcing it will require billions of years of computing time. The weakness is in telling the reciever what the encryption code will be, not the encryption itself.

@fireball, on Jul 11 2008, 07:45 PM, said in On Relative Sizes of Spacecraft:

The real problem with computerized A.Is with weapons is one hit from an EMP would render them as useless as a one-legged man in an ass kicking contest.

As will most human-piloted ships. Assuming that ships have computer-assisted controls as today's often do, an EMP will knock out the computers just as easily. The only difference is that the human is making the "turn left" decision, not the computer. The only way to avoid this is to make a fully mechanical ship (think world wars era).

Actually, we have ways of including EMP shielding in military hardware, now. Also, why does everyone that we'll still be using contemporary style electronics in future space-based assets, when optronics and photonic based computers and power systems (using light as a primary medium instead of electrons) could effectively render cheap shot EMP technology obsolete. This should, at least AFAIK, also help the issue of hull ionization.

edit: typing fixes

This post has been edited by Cydonius Libertas : 11 July 2008 - 07:55 PM

@lnsu, on Jul 12 2008, 12:31 AM, said in On Relative Sizes of Spacecraft:

The other issue with computer-controlled fighters that hasn't been fully articulated as that with enough observation, they are predictable enough to have an advantage in aiming your fire.

Actually, it has. That's what the talk about random-number generators was.

@lnsu, on Jul 11 2008, 06:31 PM, said in On Relative Sizes of Spacecraft:

As will most human-piloted ships. Assuming that ships have computer-assisted controls as today's often do, an EMP will knock out the computers just as easily. The only difference is that the human is making the "turn left" decision, not the computer. The only way to avoid this is to make a fully mechanical ship (think world wars era).

Actually, this isn't such a far fetched idea. (Scroll down about one third down the page)

Bear in mind that when I'm talking about hacking enemy spacecraft, I'm not talking on the level of a nerd hacking into your computer and moving your mouse around. I'm talking about professional hackers using highly sophisticated and specialized hardware and software, perhaps on a level we can't quite grasp yet. Any ship functioning on a basic level is going to need dozens of advanced functions for handling simple things like life support. You're also going to want to allow outside connections of some sort for your ships (and A.I.s if their advanced enough to do that sort of thing) to communicate and share data. For example, if you have two capital ships, you're going to want them to share data they're collecting from the battle so the captains on board can collaborate and effectively manage the other ships under their command.

A hacker would simply need to exploit these vulnerabilities, perhaps even breaking in through communication channels if the technology has that capability (unlike today I think ). Once inside, they could do all sorts of things. Or, more simply, find a spot on the ship's systems where data can be written, like the logs, and drop of some viruses and then watch the fireworks. And these viruses would likely be specifically designed for attacking a ship's computers, as opposed to those random things lurking on websites one typo away from the places you normally visit. Also, the hackers would likely be targeting larger ships. Hacking a fighter would take too much effort for the effects it'd have on the battle, since the enemy would have very large numbers. Hacking a capital ship, on the other hand, would be much more effective.

Bear in mind, processor power increases exponentially. While today it might takes years to run through encryptions, regular hardware and software of the future could probably crack our toughest defenses in a second. Specialized, sci-fi stuff, dedicated to hacking, would likely be able to process countless different combinations of passwords in seconds, for starters, and work on various security systems simultaneously without bogging down their processors. Granted, security will become tougher too, hence the need for a specialized computer using specialized software.

And, of course, breaking into your enemy's ship wouldn't be a guaranteed thing. On the other hand, keeping the enemy out of your systems wouldn't be guaranteed either, especially if they've updated their software more recently than you have.

Something I thought of earlier today would make the need for space-borne fighters.

Bounty Hunters.

There's always the guys who'll work alone, and working with a big-@$$ crew may not be exactly efficient for them. Having a fighter for them to buzz around in makes more sense than having a cruiser/carrier just to take down some smaller vessel.

@JTH: In a closed-loop network, such as that inside a capital ship, the only way to gain significant access is through a physical connection, which requires boarding the ship, in which case hacking is the least of your problems. At best with a properly designed non-wireless network, the only "hacking" that the enemy could do assuming every ship was human-controlled would be to feed in false recon/instruction data from other friendlies through either man-in-the-middle attacks, or completely jamming the communications by spamming the right aera of the radio spectrum with a huge mount of incomprehensible data. There is also an advantage to using analoge voice-based data transmission, (everything through radio). As a trained fighter pilot, you'd probably become pretty good at recognizing your wingmen's voices and the voice of the carrier radio transmission guy. Voices are hard to spoof.

Like I said though, a closed-network on your capital ship isn't likely, as you're going to want to be sending tactical data between your other ships, stations, command, and others. Radio would be too slow and very open for the enemy to hear. And with advanced sci-fi computer processors and software, they'd be able to crack your wind talking with little effort just by using their computer to figure out a few words and letters. Space is huge and, despite what things such as Star Wars depict, it isn't going to be divided into "sectors" that you could point to on a map. You couldn't say a ship is in sector 3 (or really something like sector 434,567,423). Well, you could, but then everyone would have to know where sector 3 is, which would require transmitting at least some sort of three-dimensional map. The command ship will probably be constantly receiving data from subordinate ships, such as shield and hull strength levels, damage reports, energy status, ammunition supplies, and much more. A radio would get cluttered like crazy with all of that.

Again, hackers wouldn't go for communications (unless they were shutting them down). They'd probably do what was quick and potent, like dumping as many viruses as they could onto the network and shutting down whatever they could get their hands on while they were doing that before the system pinned down their signal and locked down out (or locked down the whole system, but that'd be bad for your allies). Also, this wouldn't be like hacking a personal computer or the internet, it'd be an entirely different ball game that none of us could quite predict how it would work. I'm sure the levels of computational power that would be required to pull it off is over our heads. And I'm sure it'd be dang difficult to pull off. Unless your spies managed to get actual access data to you, such as the equivalent of passwords, you probably wouldn't use it as your primary strategy for victory, but rather a secondary thing that, if successful, could really screw up your enemy.

You can design a computer system that's so hacker-resistant that nobody can reasonably penetrate the system. Most modern operating systems are designed to accept configuration information from outside sources- hence all the wonderful add-ons that make your browser so potent. The analogy is that of a house with a single door and a lock. Windows computers are like houses designed so that people can go inside and do whatever they want. Their locks have keys so the right people can open the doors. Hackers have lockpicks to open doors as well. The key thing is that the gateway is built into the operating system intentionally-it's the security that's compromised. A lockpick won't allow a thief to enter a house with no doors.

Do you know why militaries today field hackers? Because the 'soft targets'- infrastructure, industry, and banking- that a military hacker would target are supported by computers that run general-purpose operating systems like Windows and various flavors of UNIX. No serious military hacker is talking about hacking into, say, an F-22 and shutting down its computer in mid-flight, because the F-22's computers are running code designed so that it can't be compromised by outside sources. And before you complain about command and control, realize that fighters are exchanging huge amounts of info with command centers, each others, and AWACS planes.

Oh, one more thing. Windows and UNIX are heavily documented. If you're writing software for a ship, that software isn't going to be using standards, it's going to be classified, and you probably aren't going to be able to look at the source code or APIs. Obviously people's computers aren't exploding every day, and this is with an operating system where new bugs are discovered and exploited every week.

This post has been edited by UE_Research & Development: 12 July 2008 - 08:57 PM