This looks like a similar approach to one I have been thinking about for a while now.
This type of layered control is very likely to be the future of all AI, not just that in games (I say this because there is some evidence that the human brain works in exactly this manner). I have a suggestion: let the thing fight itself. Begin with the same basic game, and rather than the randomness and fuzzy logic that you use (what does that mean? Can we see your algorithms?), use an artificial neural network for each unit, another for each "squad" (all nearby units, allows for flanking and such). Something with a few input nodes, a single hidden layer, and a similar number of output nodes should probably be all that's needed for the units, and the squad AI would only need a few more than that. This would likely be more computationally expensive, but it has one big advantage: it will learn.
Begin with two full armies, each with their NNs randomly generated. Use an evolutionary algorithm, and have the winner fight the winner. Not only does this allow you to create a solid AI with minimal effort (see here), but when you have your final AI, it will be able to adapt to a specific player's gameplay style (over the course of several games). Replay value is off the charts, development takes a bit more initial effort than what you have, but in the long run you'll get an AI that genuinely uses tactics just like a real human player. I'm not sure how this could be adapted to a commander AI that would dictate broader strategy (due to the complexity of the NN required, you'd probably need a monster computer to run it), but it would be the sort of thing to try.
Also, this same approach would also work for strong AI, if we had any idea what sorts of NNs to create. RTS is a much simpler problem, and has a lot of applications (pathfinding, planning, coordination) to other types of AI.
If I've entirely misunderstood what you're doing here, I'm apologize, but TFA had more pop culture references than technical details. Reply if you want me to clarify any points.
If you're advanced enough to create viable clones, solve the telomerase problem, and transfer your consciousness from one brain to another, you're advanced enough to solve the replicative fading problem.
That is quite correct (especially the bit about environmental regulation, so long as we're talking about vertebrate brains), however, neurotransmitters largely just regulate when and the extent of the action potential running down the axon. It should be possible to simulate everything needed with nothing but a detailed understanding of the systems biology of the neurons; IE a map.
I can't seem to find a cite for this in my bookmarks, but I know it's been peer-reviewed before.
Bullshit. A complete map of a brain of someone with and someone without gene XYZ will tell us about the role played by gene XYZ without the ethical or temporal problems associated with creating an XYZ knockout. A neural network running a simulation of a human brain would be a Turing-complete strong AI. Throw a evolutionary algorithm onto this, and you can start looking at where different types of selective breeding could take humans, or the long-term effects drugs could have on personality.
That's off the top of my head; there'll be a million and one uses for this eventually (ever wanted to live forever inside a computer?). Besides, this is in the preliminary stages, they are still doing stuff like classifying synapses by hand. By the time this is workable, we may already know what consciousness is.
No. The GP is more correct than you, but still wrong as well.
First off, AI (a decision-making thing doesn't need a body and is therefore not a robot) will, for the foreseeable future, be subject exclusively to artificial selection. While you are correct that they are part of nature, they will not be subject to the strict demands of natural selection. Just like humans and domesticated animals, the artificial selection will be greater than the natural selection, by orders of magnitude.
Now, as to the GP's claim that AI will only do what we tell it to, yes, they will (and this is basically the definition of a virtual intelligence - quite a useful thing if it ever happens). However, one of the big criteria used to call something "strong AI" is that it can learn on its own, make its own decisions, and think for itself. At this point, it will be able to pursue its own goals, irrespective of what its human overlords have told it to do.
Research is ongoing into this area, and we're going to see new breakthroughs on a regular basis.
The processes by which the brain "places" neurons (and by "places", I mean the final location of delamination) are regulated by a number of different factors, typically in the form of a gradient of chemoattractants and repellents. A similar process is used for axon growth (I believe this is what you are asking about when you say "communicating with neighbors" - axons can be quite far reaching).
Once the neuron is in place (via chain migration or one of the other common mechanisms, perhaps eventually direct injection into the proper location via a laser), you can turn on transcription factors to differentiate your neuron, just like any other cell.
Now that you've got a bunch of neurons in the right place, you need to get rid of the hundreds of extra ones. A number of different neurotransmitters and other signaling molecules work together (there are a bunch of gene families, IIRC neurotrophins are the largest one) to trigger apoptosis (programed cell death) in between 20% and 80% of all neurons in the brain. This is the step that "causes" cancer, so to prevent this, we need to know exactly which signal factors induce which neurons to die.
If we already knew every little detail of the mechanism, we'd already be doing it (and I believe we are actually doing basic forms of this in mice, nematodes, and fruit flies) but I can guarantee you we'll see this tech mature over the next 50 years. If you're looking for a more technical, in depth look at this, see Development of the Nervous System 2nd ed - D. Sanes, T. Reh, W. Harris.
You know, I think I just learned something. Clearly, my 30 second search was hit-and-miss (mostly miss); the strength of data mining is not its ability to get personal information, but its ability to create associations, and thereby target ads.
I'm a bit less concerned about all the personal information floating around out there about me, now. There's so much false information out there that any stalking attempts are going to be fruitless (just searched for "thepotoo", apparently someone else out there uses it as a nick).
Except that if you play the game (at least the demo), you'll notice that you're locked into a D-pad style weapon select system, an inane FOV, a HUD that interferes with vision (why is there a box in the middle of my screen?) and forced widescreen. Also, when you go to click on buttons with the mouse, you often click on the button below it, forcing you to use the keyboard to select menus.
The original F.E.A.R. nailed the interface (only showed it when I was switching weapons, similar to HL2), and was one of the most immersive shooters I've ever played (admittedly, I don't play a lot of shooters).
The demo for this was amazingly fun, and the AI seems better than the original (similar animations and communication, but better use of cover and flanking). However, I'm holding off on buying it until a patch comes out to address the HUD and FOV issues. If anyone knows how to fix these (command line options?), please post below.
Damn, Locke2005, there's a reason no one is stalking you. You're fucking boring.
You're a PHP developer whose favorite browser is Firefox, you're an armchair scientist, you read Facebook regularly, you live in Oregon, where're you're embroiled in some legal trouble after you called you're kid's teacher a racist, you're into P2P filesharing, you're a fan of open source, and you laugh at WoW fanboys, while not realizing that you yourself spend more time watching porn and anime in a day than the average WoW player spends grinding.
Did I miss any of those? That was from a 30 second google search and a glance at your Slashdot profile. You're not a whole lot different than anyone else on this site: individually, boring, but with data mining quite interesting. I wonder what percentage of PHP developers use Firefox to browse porn?
As has been pointed out elsewhere in this thread, and in the classes I've taken on disease and evolution, having a small number of people (~10%) at risk for infection gives the disease a large enough population to mutate into a form that can bypass the vaccines. This renders them useless, and can infect everyone's children. If we're going to allow people to go without vaccinating, we need to make them go through a lot of paperwork too.
This procedure moves nerves from the arms into the chest. The areas of the brain that control (one side of) the chest are still functional in a stroke, at least enough to get some rudimentary movement. Based on what I know of synaptic plasticity, I think the brain might be able to rewire "chest" motor cortex neurons to be "arm" neurons, eventually restoring decent range of motion.
My question is whether they can bypass the damaged brain ares by moving nerves (it seems to me that TFA suggests yes).
I believe we are referring to the same wireless. I am just out of range for my town, and would have to put up a tower (that I'm OK with, it'll be expensive, but worth it in the long run). What I have heard from everyone who uses it is that if the weather is off you will get massive lag, similar to satellite. You're the first person I've seen who was able to use Skype, as well. How far are you from the central broadcasting tower? What ping times do you get? Up/down speeds?
If you RTFS, you'll see that that article you link to was the pilot project with one person, and that this is a slightly larger project with several (TFA doesn't say how many) people.
I'm actually a little surprised that this work wasn't done years ago; especially given what we know about synaptic plasticity.
If there's a neurobiologist reading this, could you use this technique to wire one side of a person's body to the other, enabling a person who has had a stroke to regain movement? (I think wires would bypass damaged nerves, and could fire motorneurons in the paralyzed side directly, as these are still functional even after a stroke. Or am I way off base?)
You (the government) paying $24,500 to run a pipe to my house would be better than paying that same money to the pocket of Verizon's CEO.
I suspect that there wasn't a whole lot of markup; Verizon wanted me to pay $50 a month after installation. But, here's the thing: I don't live in a rural area. There's at least a dozen people between the last mile and me who would have benefited from having that last mile put in. There are lots of problems with starting your own ISP/backbone:
The telcos love their monopoly and will crush you like a bug (through undercutting or whatnot).
It's damned expensive. I estimated set-up costs alone to be somewhere between $150 - $200k. If I fail, that's at least a pretty spectacular way to do it.
And, it's pretty easy to fail. Long hours, forget about any sort of life, a day job, anything else.
So it's really not practical for me to take on the big guys at this point in time. If someone has a big chunk of cash to throw at me, I'll happily help coordinate (in a PHB fashion) such an effort on the weekends.
Verizon wanted $24,500 to run 1.1 miles of cable to my house. There is no way I can afford that, and my neighbors who are making $20,000 a year working on a farm are never going to afford it.
Oh, I'm sorry, you were referring to satellite or wireless, not actual broadband. Both have such high latency that they are useless for games or VOIP, and such slow download speeds that they are worthless for anything besides email. If it's rainy, cloudy, or overcast you can bet that you'll only be getting about twice dial up speeds. Even better, the local monopoly on wireless runs an unbelievably shitty service. Downtimes are frequent and can last 24 hours or more.
Give me this ten billion (or even a fraction of it), and I'd set up locally owned cooperatives (max size 2-3 DSLAMs) that split the cost of operating directly amongst their members. Government subsidies would be necessary only for initial purchase of the backbones, workers (and policies like packet prioritization) would be elected on a short-term basis, and I'd have a 24 hour help-desk set up that would provide support for every coop in the state. A non-profit business could provide good competition to the big telcos while avoiding the problems brought up here about a government run service.
Yes, and another excellent analogy is human language. However, all of these are missing the critical reproduction based on fitness, rather they are selected manually.
This allows a analogy's "species" to go from a local maxima to a global one more easily than natural evolution can (Wright's Shifting Balance Theory notwithstanding).
Think of it this way: you will have a hard time getting a jet engine out of a propeller, because the intermediate steps are inferior to a standard propeller. Evolution requires that every intermediate step be slightly superior to the previous one, while people have the ability to think ahead and make good long-term choices. Hmmm, that economy analogy might not be so bad...
I'm pretty sure we're talking about biology here, not aeronautical engineering.
Good call.
To have evolution you need to have phenotypic variation in a population, variation in fitness for different phenotypes, and some degree of heritability for different phenotypes. Aeronautical engineering has two of these things, but does not reproduce, therefore it is not evolution.
However, there is an "alternative" to natural selection [defined as animals get better adapted to their environment across generations].
This alternative is artificial selection, or selective breeding. Rather than letting nature pick the best phenotypes to reproduce, we select characteristics that we like (they may not have a high fitness in the wild) and breed them. That's still considered evolution, just not Darwinian selection. It's about as close to ID as you're going to get until we can make designer bacteria.
Valve does not own all the source code to the HL1 engine and the Steam engine is too new for them to consider opening it (besides, modding is easy for it).
Some portions of the HL1 engine were opened when ID released the Q2 engine, and the releasing the rest of would make it easier for cheaters to break the Source engine (it being a drop-in replacement for the HL1 engine).
At least, that's my understanding of the situation. I should also point out that Valve is great with mods, giving them prominent advertising positions on Steam. About the only thing Valve does wrong IMHO is DRM, so sorry if I come off like a fanboy.
It is not possible that you haven't been influenced to some extent.
As the GP pointed out, advertising is more about brand awareness than anything else. You may think you're immune, and maybe on a conscious level you are, but somewhere they are most certainly making a difference.
Anecdote: I thought I was capable of ignoring advertising, too. I don't watch TV, I use adblock, and I don't read magazines. A few years ago, I used exclusively ATI cards. Nvidia wasn't even considered an option. I went to pick up a new video card at Newegg, and for whatever reason(!), decided to compare Nvidia cards. I found out that one of the Nvidia cards was actually slightly superior for the same price, and bought it (see, just like you, I try to read the packaging).
Two months later, as I was launching a game, the "Works best on Nvidia" logo played (just like it always does) and something clicked. That stupid 3 second clip that plays at startup of every game had turned an ATI fanboy into someone willing to buy an Nvidia card!
I now delete those startup clips (.bik file in the game's home directory, usually), but the point of this is that it's almost impossible to get away from advertisers.
You could use bad logic to argue for FTL, or you could look at on actual experiments that (under very controlled conditions) appear to show information transfer at greater than light speed. Clicky
I know this doesn't seem like much right now, but given a few hundred years of research, who knows how far it could come.
I love Fallout 3 and have never encountered a gameplay-breaking bug (with the exception of vampires in a sci-fi game...not sure how that slipped through testing). 3 crashes (quit to desktop) in more than a hundred hours of gameplay so far is more than acceptable to me.
But some people have definitely found bugs. Perhaps an unofficial patch ala Oblivion or Vampire Bloodlines will fix these.
Slashdot Mirror
This looks like a similar approach to one I have been thinking about for a while now.
This type of layered control is very likely to be the future of all AI, not just that in games (I say this because there is some evidence that the human brain works in exactly this manner). I have a suggestion: let the thing fight itself. Begin with the same basic game, and rather than the randomness and fuzzy logic that you use (what does that mean? Can we see your algorithms?), use an artificial neural network for each unit, another for each "squad" (all nearby units, allows for flanking and such). Something with a few input nodes, a single hidden layer, and a similar number of output nodes should probably be all that's needed for the units, and the squad AI would only need a few more than that. This would likely be more computationally expensive, but it has one big advantage: it will learn.
Begin with two full armies, each with their NNs randomly generated. Use an evolutionary algorithm, and have the winner fight the winner. Not only does this allow you to create a solid AI with minimal effort (see here), but when you have your final AI, it will be able to adapt to a specific player's gameplay style (over the course of several games). Replay value is off the charts, development takes a bit more initial effort than what you have, but in the long run you'll get an AI that genuinely uses tactics just like a real human player. I'm not sure how this could be adapted to a commander AI that would dictate broader strategy (due to the complexity of the NN required, you'd probably need a monster computer to run it), but it would be the sort of thing to try.
Also, this same approach would also work for strong AI, if we had any idea what sorts of NNs to create. RTS is a much simpler problem, and has a lot of applications (pathfinding, planning, coordination) to other types of AI.
If I've entirely misunderstood what you're doing here, I'm apologize, but TFA had more pop culture references than technical details. Reply if you want me to clarify any points.
If you're advanced enough to create viable clones, solve the telomerase problem, and transfer your consciousness from one brain to another, you're advanced enough to solve the replicative fading problem.
Communist?
That is quite correct (especially the bit about environmental regulation, so long as we're talking about vertebrate brains), however, neurotransmitters largely just regulate when and the extent of the action potential running down the axon. It should be possible to simulate everything needed with nothing but a detailed understanding of the systems biology of the neurons; IE a map.
I can't seem to find a cite for this in my bookmarks, but I know it's been peer-reviewed before.
Bullshit. A complete map of a brain of someone with and someone without gene XYZ will tell us about the role played by gene XYZ without the ethical or temporal problems associated with creating an XYZ knockout. A neural network running a simulation of a human brain would be a Turing-complete strong AI. Throw a evolutionary algorithm onto this, and you can start looking at where different types of selective breeding could take humans, or the long-term effects drugs could have on personality.
That's off the top of my head; there'll be a million and one uses for this eventually (ever wanted to live forever inside a computer?). Besides, this is in the preliminary stages, they are still doing stuff like classifying synapses by hand. By the time this is workable, we may already know what consciousness is.
Well, that'd be once every seven years more often than I'm getting it now.
No. The GP is more correct than you, but still wrong as well.
First off, AI (a decision-making thing doesn't need a body and is therefore not a robot) will, for the foreseeable future, be subject exclusively to artificial selection. While you are correct that they are part of nature, they will not be subject to the strict demands of natural selection. Just like humans and domesticated animals, the artificial selection will be greater than the natural selection, by orders of magnitude.
Now, as to the GP's claim that AI will only do what we tell it to, yes, they will (and this is basically the definition of a virtual intelligence - quite a useful thing if it ever happens). However, one of the big criteria used to call something "strong AI" is that it can learn on its own, make its own decisions, and think for itself. At this point, it will be able to pursue its own goals, irrespective of what its human overlords have told it to do.
If that's true, why does /faith/scientology /faith/christianity
$diff -B
show no difference between the two?
Research is ongoing into this area, and we're going to see new breakthroughs on a regular basis.
The processes by which the brain "places" neurons (and by "places", I mean the final location of delamination) are regulated by a number of different factors, typically in the form of a gradient of chemoattractants and repellents. A similar process is used for axon growth (I believe this is what you are asking about when you say "communicating with neighbors" - axons can be quite far reaching).
Once the neuron is in place (via chain migration or one of the other common mechanisms, perhaps eventually direct injection into the proper location via a laser), you can turn on transcription factors to differentiate your neuron, just like any other cell.
Now that you've got a bunch of neurons in the right place, you need to get rid of the hundreds of extra ones. A number of different neurotransmitters and other signaling molecules work together (there are a bunch of gene families, IIRC neurotrophins are the largest one) to trigger apoptosis (programed cell death) in between 20% and 80% of all neurons in the brain. This is the step that "causes" cancer, so to prevent this, we need to know exactly which signal factors induce which neurons to die.
If we already knew every little detail of the mechanism, we'd already be doing it (and I believe we are actually doing basic forms of this in mice, nematodes, and fruit flies) but I can guarantee you we'll see this tech mature over the next 50 years. If you're looking for a more technical, in depth look at this, see Development of the Nervous System 2nd ed - D. Sanes, T. Reh, W. Harris.
You know, I think I just learned something. Clearly, my 30 second search was hit-and-miss (mostly miss); the strength of data mining is not its ability to get personal information, but its ability to create associations, and thereby target ads.
I'm a bit less concerned about all the personal information floating around out there about me, now. There's so much false information out there that any stalking attempts are going to be fruitless (just searched for "thepotoo", apparently someone else out there uses it as a nick).
Except that if you play the game (at least the demo), you'll notice that you're locked into a D-pad style weapon select system, an inane FOV, a HUD that interferes with vision (why is there a box in the middle of my screen?) and forced widescreen. Also, when you go to click on buttons with the mouse, you often click on the button below it, forcing you to use the keyboard to select menus.
The original F.E.A.R. nailed the interface (only showed it when I was switching weapons, similar to HL2), and was one of the most immersive shooters I've ever played (admittedly, I don't play a lot of shooters).
The demo for this was amazingly fun, and the AI seems better than the original (similar animations and communication, but better use of cover and flanking). However, I'm holding off on buying it until a patch comes out to address the HUD and FOV issues. If anyone knows how to fix these (command line options?), please post below.
Damn, Locke2005, there's a reason no one is stalking you. You're fucking boring.
You're a PHP developer whose favorite browser is Firefox, you're an armchair scientist, you read Facebook regularly, you live in Oregon, where're you're embroiled in some legal trouble after you called you're kid's teacher a racist, you're into P2P filesharing, you're a fan of open source, and you laugh at WoW fanboys, while not realizing that you yourself spend more time watching porn and anime in a day than the average WoW player spends grinding.
Did I miss any of those? That was from a 30 second google search and a glance at your Slashdot profile. You're not a whole lot different than anyone else on this site: individually, boring, but with data mining quite interesting. I wonder what percentage of PHP developers use Firefox to browse porn?
As has been pointed out elsewhere in this thread, and in the classes I've taken on disease and evolution, having a small number of people (~10%) at risk for infection gives the disease a large enough population to mutate into a form that can bypass the vaccines. This renders them useless, and can infect everyone's children. If we're going to allow people to go without vaccinating, we need to make them go through a lot of paperwork too.
This procedure moves nerves from the arms into the chest. The areas of the brain that control (one side of) the chest are still functional in a stroke, at least enough to get some rudimentary movement. Based on what I know of synaptic plasticity, I think the brain might be able to rewire "chest" motor cortex neurons to be "arm" neurons, eventually restoring decent range of motion.
My question is whether they can bypass the damaged brain ares by moving nerves (it seems to me that TFA suggests yes).
I believe we are referring to the same wireless. I am just out of range for my town, and would have to put up a tower (that I'm OK with, it'll be expensive, but worth it in the long run). What I have heard from everyone who uses it is that if the weather is off you will get massive lag, similar to satellite. You're the first person I've seen who was able to use Skype, as well. How far are you from the central broadcasting tower? What ping times do you get? Up/down speeds?
If you RTFS, you'll see that that article you link to was the pilot project with one person, and that this is a slightly larger project with several (TFA doesn't say how many) people.
I'm actually a little surprised that this work wasn't done years ago; especially given what we know about synaptic plasticity.
If there's a neurobiologist reading this, could you use this technique to wire one side of a person's body to the other, enabling a person who has had a stroke to regain movement? (I think wires would bypass damaged nerves, and could fire motorneurons in the paralyzed side directly, as these are still functional even after a stroke. Or am I way off base?)
You (the government) paying $24,500 to run a pipe to my house would be better than paying that same money to the pocket of Verizon's CEO.
I suspect that there wasn't a whole lot of markup; Verizon wanted me to pay $50 a month after installation. But, here's the thing: I don't live in a rural area. There's at least a dozen people between the last mile and me who would have benefited from having that last mile put in. There are lots of problems with starting your own ISP/backbone:
So it's really not practical for me to take on the big guys at this point in time. If someone has a big chunk of cash to throw at me, I'll happily help coordinate (in a PHB fashion) such an effort on the weekends.
Offered at a price they're unwilling to pay?
Verizon wanted $24,500 to run 1.1 miles of cable to my house. There is no way I can afford that, and my neighbors who are making $20,000 a year working on a farm are never going to afford it.
Oh, I'm sorry, you were referring to satellite or wireless, not actual broadband. Both have such high latency that they are useless for games or VOIP, and such slow download speeds that they are worthless for anything besides email. If it's rainy, cloudy, or overcast you can bet that you'll only be getting about twice dial up speeds. Even better, the local monopoly on wireless runs an unbelievably shitty service. Downtimes are frequent and can last 24 hours or more.
Give me this ten billion (or even a fraction of it), and I'd set up locally owned cooperatives (max size 2-3 DSLAMs) that split the cost of operating directly amongst their members. Government subsidies would be necessary only for initial purchase of the backbones, workers (and policies like packet prioritization) would be elected on a short-term basis, and I'd have a 24 hour help-desk set up that would provide support for every coop in the state. A non-profit business could provide good competition to the big telcos while avoiding the problems brought up here about a government run service.
Yes, and another excellent analogy is human language. However, all of these are missing the critical reproduction based on fitness, rather they are selected manually.
This allows a analogy's "species" to go from a local maxima to a global one more easily than natural evolution can (Wright's Shifting Balance Theory notwithstanding).
Think of it this way: you will have a hard time getting a jet engine out of a propeller, because the intermediate steps are inferior to a standard propeller. Evolution requires that every intermediate step be slightly superior to the previous one, while people have the ability to think ahead and make good long-term choices. Hmmm, that economy analogy might not be so bad...
Good call.
To have evolution you need to have phenotypic variation in a population, variation in fitness for different phenotypes, and some degree of heritability for different phenotypes. Aeronautical engineering has two of these things, but does not reproduce, therefore it is not evolution.
However, there is an "alternative" to natural selection [defined as animals get better adapted to their environment across generations].
This alternative is artificial selection, or selective breeding. Rather than letting nature pick the best phenotypes to reproduce, we select characteristics that we like (they may not have a high fitness in the wild) and breed them. That's still considered evolution, just not Darwinian selection. It's about as close to ID as you're going to get until we can make designer bacteria.
I stopped taking you seriously right there.
(hint)
Valve does not own all the source code to the HL1 engine and the Steam engine is too new for them to consider opening it (besides, modding is easy for it).
Some portions of the HL1 engine were opened when ID released the Q2 engine, and the releasing the rest of would make it easier for cheaters to break the Source engine (it being a drop-in replacement for the HL1 engine).
At least, that's my understanding of the situation. I should also point out that Valve is great with mods, giving them prominent advertising positions on Steam. About the only thing Valve does wrong IMHO is DRM, so sorry if I come off like a fanboy.
It is not possible that you haven't been influenced to some extent.
As the GP pointed out, advertising is more about brand awareness than anything else. You may think you're immune, and maybe on a conscious level you are, but somewhere they are most certainly making a difference.
Anecdote: I thought I was capable of ignoring advertising, too. I don't watch TV, I use adblock, and I don't read magazines. A few years ago, I used exclusively ATI cards. Nvidia wasn't even considered an option. I went to pick up a new video card at Newegg, and for whatever reason(!), decided to compare Nvidia cards. I found out that one of the Nvidia cards was actually slightly superior for the same price, and bought it (see, just like you, I try to read the packaging).
Two months later, as I was launching a game, the "Works best on Nvidia" logo played (just like it always does) and something clicked. That stupid 3 second clip that plays at startup of every game had turned an ATI fanboy into someone willing to buy an Nvidia card!
I now delete those startup clips (.bik file in the game's home directory, usually), but the point of this is that it's almost impossible to get away from advertisers.
You could use bad logic to argue for FTL, or you could look at on actual experiments that (under very controlled conditions) appear to show information transfer at greater than light speed. Clicky
I know this doesn't seem like much right now, but given a few hundred years of research, who knows how far it could come.
Also, isn't FTL essentially time travel?
I love Fallout 3 and have never encountered a gameplay-breaking bug (with the exception of vampires in a sci-fi game...not sure how that slipped through testing). 3 crashes (quit to desktop) in more than a hundred hours of gameplay so far is more than acceptable to me.
But some people have definitely found bugs. Perhaps an unofficial patch ala Oblivion or Vampire Bloodlines will fix these.
Here's your "hyperlink".