You can't do meaningful experiments without some idea of what the theory says will happen. Numerics of this sort provide that for complex physical cases which are essentially impossible to work out with pen and paper. So yes, this is a step towards getting knowledge of the universe and how it might work.
Also, understanding does NOT require the tie to experiment since you can have mathematical understanding of a particular theory independant of whether that theory properly models reality. For instance, I can go and work out what orbits look like in a five dimensional space. If I go and check my results versus reality 'hey, it doesn't match up!'. So from that point of view, all I've learned is that space on that scale isn't five dimensional. But lets say I don't even bother to check versus reality. I've still learned something about the mathematical properties of the theory and I've gained intuition about how things behave - namely, I've learned that closed (classical) orbits seem to only be able to exist in 3d. How strange! And I've learned that if I had a situation where the system was, say, restricted to a lower dimension (examples in electromagnetism) then I can expect large changes to the dynamics. Or perhaps a better example is, we can learn a lot about phase transitions in three dimensions by doing problems in four dimensions where they can be solved exactly and then doing an expansion around the four dimensional solution to approximate the solution in 3d. That approach doesn't depend on the underlying Hamiltonian you're solving being the correct one for some physical system - it is a purely mathematical understanding which can generically be applied to many different theories. So the benefit is, in the future when I find a better Hamiltonian for my phase transition, a better dynamical theory for gravitation, etc, I can apply the techniques I've learned from before to those as well.
Why, that would just create an arms race as then ad checkers would adapt to the adaptations made by the ad makers. The end result of such would be that DVRs would evolve sentience and ads would evolve bacteria-like mutability and randomness. Sort of like spam blockers and spam. I for one... aw, skip it.
Sure but the tradeoffs still exist. Hardware acceleration means money towards that particular card means money that isn't spent on other aspects of the system. Also, restricting physics to the hardware puts a limit on the sort of physics that can be done if it HAS to go through the hardware. Since most physics can be expressed in terms of or at least take advantage of large matrix operations, that might be the best and most general way to go. Fourier transforms on the hardware are useful for certain kinds of physics, but I doubt it'd be too useful for the kind of thing that goes into games unless people start solving the PDEs for water ripples or whatever. But I don't actually know how its planned to go. Some of this could probably be taken care of decently by making graphics card interfaces a bit more general - since you can already do a lot of linear algebra stuff on 3d cards as it is. Afaik, the limit on graphics cards has more to do with sending information to the card and receiving it back than the actual computational difficulty of what it does, which would mean that on average there are some spare cycles that could be utilized there.
Specificity of the hardware was a big problem with the first 3d hardware. Basically the game's appearance was limited by the card and not the programmers. Its a bit better nowadays with shaders, and you can actually use some tricks to do general linear algebra on 3d cards but its not as fast as it could be since the card wasn't made for that purpose.
The problem is tradeoff. Lets ignore development time, since putting in an actual physics engine can actually speed that up if it means you don't have to explicitly program stuff, the same way that putting in random generation stops you from having to tweak each individual NPC's name and appearance.
No, the tradeoff I'm talking about here is system specs. Oblivion is a good game, immersive, fun. It was a good game back when it was called Morrowind and ran on computers that would be laughable by today's standards.
Now, I'm not saying they should have kept the graphics the same in Oblivion. But I'm saying they could have and if the rest of the game is solid, that wouldn't make it a not-fun game. Since systems have gotten better, then they can choose to improve the graphics and thats great. But the sort of casual physics they're talking about in that article isn't something that could realistically be done on modern systems. I was even surprised that the physics in Oblivion could be done until I realized that they had an on-off switch for it, so stationary objects weren't simulated. Meaning they had to do at most a couple dozen nodes at once - not a big deal.
Or, to put it another way. I can make a game that solves compressible Navier-Stokes to derive the weather patterns so that the player can influence the weather via the butterfly effect. Or I can stick in a random distribution. If its free, I might as well do the former. But it isn't, so if I want anyone to be able to play my game, I choose to do the latter which is almost as good. Putting in by hand swirling smoke gives you something which takes you as a developer more time to do, but the benefit is that the computational difficulty drops and you have spare cycles to do even more interesting stuff. I'd rather have my cycles used for a really clever AI, or even an evolving world, than simulating the grass. And since I have a finite computational power, thats a choice that must eventually be made.
It's only a double standard if you take all people to be a single entity. Realize that in any sizeable group of people, you should hear conflicting opinions expressed. If you don't, thats stranger than if you do.
Another explanation is that when large companies do business, their output is often somewhat homogenized since they have to target a broader audience to sustain themselves. Whereas smaller companies/developers can be niche. So a statement like that can be in support of niche development which is optimized for a smaller community at the cost of a smaller audience and a lower chance of success in business. And from that point of view, it makes a bit of sense - the large companies are seen as blocking the market, so one wants them to fail, whereas the niche developers are doing new stuff and would take a bigger hit from each illegitimate copy, and so those you're accusing of holding a double standard wish to support them since they so obviously need it.
Of course with things like mods that sort of falls through. Most game mods are done for the love of the game, not the love of the dollar, and would be made with or without financial support. With that in mind, encouraging mod makers to become commercial is sort of like pushing 'linux - everywhere!'... the whole idea of being a part of the market ends up being more attractive than the actual thing being produced, and you're back to big companies again.
Of course, I can't speak for any other person, so this is just a set of possible interpretations, not a definitive answer for other's motives.
Myself, I'd prefer if everything were developed freely and just because the developer(s) wanted to do it - including larger projects. I don't even really have a problem with support via donations for the larger projects which are very time-consuming to develop for - but I haven't seen severe consequences or thought that through to the end so maybe I'll change my view on that when it becomes more common and if/when faults become evident. 'Everything' developed and released for free is unrealistic of course, so I don't expect it to happen. But 'some projects' is quite realistic and has been happening even more of late. I'm certainly not going to encourage the reverse direction by paying for fan-made modifications to a game. But I can understand that some people would, and why some people call it a step forward.
Spoofing the UI has been done in other cases before, so I don't think it'll provide much of an obstacle. I've seen tons of banner ads that are made to look like an windows error message. I'm not sure how effective that sort of thing is, but I imagine it gets the same sorts of people who wouldn't notice strange URLs or who don't look to see whether the site they're interacting with is using encryption (thats a UI icon too, but most people probably don't even know what it means).
Turns out this has already been worked-around. The format used by Oblivion is shared by a few other games (NetImmerse format) with a few modifications. There's a plugin for 3dsmax version something or other that outputs NIF files for Civilization 4, which can be imported into Oblivion but lack physics data (so no ragdoll, etc). It seems to work for buildings and terrain statics though, which just use a bounding box for collisions. So given this is just two weeks in, I don't imagine it will be long before the community figures out how to port over Havok data from other models, and from there starts to modify it.
And if I personally don't reply to spam, that will cause it all to go away? Unfortunately, silently not-buying gives you the power of a drop of water in an ocean. Spreading your concerns to others is a legitimate way to magnify that effect.
If you have a vested interest in people making good games that you'd be willing to shell out for, then seeing travesties that make you say 'nuts to this, I'm not buying it' doesn't just mean $2 sitting in your wallet that you wouldn't've had otherwise, it means $2 sitting in your wallet instead of going to pay for something you think is worth the $2. So I think its perfectly reasonable for fans to get upset when the company who has been producing content that they like suddenly changes directions on them. It doesn't give them license to burn the headquarters or anything, but it certainly gives them a good reason to announce their displeasure to the world in the hopes that other people who are sitting on the fence will listen and will decide not to buy the horse armor, or any addons of this sort, or even decide to get the PC version instead of the xbox version since there at least community content will easily replace the mods-for-pay.
The correct approach to this problem (of simulation in general) isn't to start at the lowest possible level and crunch. That will easily run into the issue you're talking about, which boils down to an O(e^N) dependence on particle number rather than the polynomial dependence that classical physics gives you.
To simulate such a huge system, the most feasible approach would be to start with an abstract high level model, simulate it, and see what it fails to produce. Then, refine the model down to lower and lower levels until you reach a turning point where the additional accuracy gained from the refinement is small compared to the increase in computing cost.
Sure you won't get a 'to the level of string theory accurate' computational model of a guy saying 'My name is Phil.'. But if your model of the guy says 'My name is Phil.' without needing to go that low level, then that actually tells you some interesting things about the brain. And if your high level model fails, well then you need to try something else/lower level possibly. But at least this way you don't waste time simulating things you don't in the end actually need to simulate to get the qualitative behavior thats the motivation for the simulation in the first place.
But you do have to make sure that your abstract model is consistent with the underlying stuff. If your abstract model is impossible to arrive at from the lower levels, then you're going to have some problems.
The thing is, the first form nanotechnology is likely to take will be exactly that - mutating organics to produce the desired behavior. Though its likely to be bacteria and not viruses. A ribosome is already an excellent nanoassembler:)
The problem with this sort of fear is that underlying it is the tacit assumption that any artificial life can easily and accidentally instantly outcompete every similar form of biological life. Yes, there will be some things that the standard inorganic mechanical self-replicating nanobots that are often envisioned would be better at - the rearrangement of inorganic compounds is probably easier for something that uses nanoscale mechanical methods rather than sequences of downhill chemical reactions - but in any sort of biological arena they'd be competing with billions of years of evolution. And would be made of materials not as easily obtained in a human body, so spreading via infection would be more difficult.
If someone was to spend time engineering something specifically to kill humans, then that could probably pose a significant threat - after all, most diseases don't evolve to immediately kill their hosts since that would inhibit the spreading. But an accidentally flipped bit isn't going to do it.
Really, the whole 'artificial' thing is a red herring. It doesn't matter what the nanobot is made of. Organic chemicals, inorganic chemicals, etc. Frankly, if you want to engineer some sort of nasty self-replicating human-killing thing, you're probably better off starting with something that already self-replicates and already interacts with biological reaction pathways. Just because you make the jump to silicon doesn't mean you suddenly get to break all the limits that are in place on the performance of biological stuff. You're likely to get a whole new set of limits right on top of that.
Complete chaos is a lot more predictable than incomplete chaos. Incomplete chaos, you have to worry about when its ergodic and when it isn't. If there are aspects which are totally random, or at least sufficiently random that for all intents and purposes you can't predict the exact sequence of states then you can just use the distributions. The end result will be a theory that becomes more accurate the larger the system it's used to describe. It'll fail utterly on a group of three people but will work brilliantly on a group of three billion.
Theories developed via application of the scientific method have the ability to predict with far-above-chance accuracy future events and thus can be used as the centerpoints of design - we can build machines, etc that rely on these predictions and which as a result work and do the things we desire. The same cannot be said for 'the gods are angry'.
Only that few characters? I was thinking in terms of 32 character passwords (which amounts to a 128 bit key) and such. I don't mean a login password obviously, but for something like an encrypted harddrive, a 32 character password isn't too unreasonable (take the fifth letter of every other line on two specific nonconsecutive pages of a particular book or take a sentence that you have memorized and use the first character of each word or something like that). You could probably memorize the necessary information for either of those techniques without too much trouble. Though it does make one seem a bit paranoid.
Part of it is just to make it hard to get _everyone's_ data. While cracking one guy's password may be easy to do if you really want to know what he's been doing, if that costs a couple million in computing resources (not precisely the hardware, but rather the computer time, cryptanalysts, etc) you're not going to be able to simply do it to everyone. So I guess the question is, how long does a password-based encryption need to be these days to cost at least $1M in computing resources to crack?
I'm following you up to 'grows exponentially with computation time'. You mean that because of the exponential divergence of initial conditions? I suppose it depends on whether you want to extract an exact future system configuration or whether you want statistical information about turbulent flows. Throwing a smaller timestep at it won't make a difference anyhow if its the errorbars in your initial condition that are diverging with the chaotic dynamics, so really for long times the best you can hope for is a statistical description - that is, map out the entire region of phase space where the turbulent flow may lie.
I don't believe I ever said that any possible set of axioms would give you a self-consistent system. Now, I suppose it is a question to whether there are any self-consistent systems at all. If you could somehow prove that there are no self-consistent systems when they're sufficiently developed to include e.g. integer arithmetic then you might have a point. But I don't think Godel lets you do that any more than it lets you prove that a particular one is self-consistent.
No, I think I do understand; my point was that what I said in my original post is not ruled out by Godel. You can define your own axioms and play in that playground and prove things within that framework. The overall consistency of the system is not necessarily an issue so long as the regions in which you work do not actually violate that consistency. Proof of the overall consistency is not a requirement for proofs within the framework.
It seems to me though that you can generate statements that are trivially well-defined. An example would be some simple arithmetic upon a subset of numbers. If I define a deterministic procedure which lets me add two specific integers, then by applying that procedure I will always get the same result. How is that not a proof of something, even if it's something trivial? While its possible for me to generate certain statements which cannot be proven or even considered without the addition of axioms, that doesn't mean that there are no statements that I can prove.
Well any progress is progress. Reducing copyrights is a much more achievable short-term goal than eliminating them. The reason I don't like copyrights in general is that it makes certain forms of communication between two parties a crime against a third, even if that third is not explicitly involved in the interaction. Allowing people to control the distribution of what amounts to ideas means allowing people to use the law to ban ideas, restrict ideas, etc. I don't agree that control of one's creation is defensible from a moral perspective (at least from mine...) because it interferes with what to me is a deeper right of free communication.
In the long run I can't see that as being beneficial to the growth of science and the arts, even if it does encourage some people to produce things when they would otherwise not. My background is in academia, and you really do see a lot of research and ideas being produced and then made more or less freely available (journal publication models really need to be revamped, but look at arxiv and citeseer). So I'm not at all convinced that copyright is necessary to promote innovation.
Basically what I hope to see in the coming years is so much 'amateur' but freely available content that commercial production of content which centers around its restriction becomes so unprofitable that those companies go out of business or at least that the market becomes reduced. We've seen it happen partially in the opensource movement - its not really that non OSS companies have gone out of business or lost all their customers or anything that drastic, but on the other hand you can have a fully functioning operating system with most of the supporting software you'd want or need without shelling out a computer tax. You can see it sort of happening with some forms of music (the tracker community is quite active, for instance, but there's not as much quality amateur music with vocal elements from what I've seen). We've seen a few scant examples with video - Star Wreck is the main one that comes to mind.
My main fear is that some of these licenses may go from things which encourage the growth of that community to things as bad as copyright in stifling that growth once the community becomes sufficiently large. Look at some of the license compatibility quarrels in OSS. It prevents good programs from being made because of the legal hurdles, not even necessarily because the authors wanted to prevent that specific usage. When you have projects with a thousand authors, it becomes impossible to go and ask every author 'hey, want to let this other project use our code even though they have an extra attribution requirement and don't require no-encryption?'.
And of course anything like a license will have loopholes and tricks which people can exploit to apply force to others even if its against the original purpose of the license. This isn't so much an issue with standalone programs - the author can put whatever license they want on it. But it is an issue when a choice of license is forced because the author incorporates some external code.
Anyhow, just my thoughts on the whole copyright issue.
Math theorems don't necessarily have to apply to the physical universe. The axioms on which the theorems are built are explicitly part of the theorems, leading to a logically self-consistent system. That is, you define the particular 'universe' you want to study by setting down axioms, then you prove things which you know are true about that universe because you've derived them in a logical fashion from those axioms you've set down.
Since the discussion here is about the GPL then binaries in terms of compiled code is the most in-context case. But I wouldn't mind extending that to any sort of data in any shape or form.
Your second paragraph states my position accurately.
I realize that and I was attempting to point it out with my second paragraph. My point is that I'm willing to take the risk that I cannot defeat a technological measure over the risk of jailtime or large fines should I succeed or even attempt to do so. From my point of view it's better that information is not restricted by law than that I'm guaranteed the ability to force people to not use my information in a way that restricts my later access to it via technological means.
Personally, the main thing I ask out of the GPL and other OSS licenses is simply to prevent others from taking that code and making it illegal for me to distribute it and binaries generated from it (or modifications of it). As long as distribution, reverse engineering, etc aren't illegal, I don't see the GPL as being nearly as necessary as it is today. Actually forcing the person to distribute the source and binary together is an added effect, but it's largely secondary in my eyes to simply keeping actions involving the code and program from being contraband.
DRM essentially threatens the most important part of the GPL (in my eyes that is) without harming the weaker part. Now the binaries can be distributed but you can't run them if you don't have the user-specific key to unlock the DRM around the binaries. On the other hand, if it were legal to circumvent DRM, I hardly think that any such methods would prove practically effective. So I'm game for taking my chances in a world where someone can't use a license to levy legal threat against me using the information I have access to any way I want, even if it means I can't use a license to prevent people from obfuscating that information and making it harder for me to actually use it as I'd like.
I did in fact intend phase diagram, not phase plane. The idea is, there are different distinct states of the system, such that in different states different symmetries or qualitative properties exist. As some parameter is adjusted, there is a change from one state to the other. When that change is discontinuous, you're seeing a phase boundary. Systems very close to the phase boundary may exhibit certain well-studied behaviors. In particular, if a system is near a critical point there are many things that may be said about its statistical properties.
Linearity has nothing at all to do with the success of statistical mechanics. In fact, to a good extent the more nonlinear the system, the better stat mech applies. The reason for this is ergodicity. A purely linear system will have certain invariances that prevent a random distribution from accurately modelling the population. In essence, the linearity makes it so that the initial state of the system controls its state from that time on - there is no interaction between the different bases. A nonlinearity added to the system permits mixing between states which, averaged over a large time, will tend to produce distributions that are well-behaved and predictable. Hidden symmetries in the rules may of course cause this to fail - there is some constraint which your statistical sampling fails to acknowledge, and so you miscount.
You mention turbulence. The statistical mechanics of turbulence is actually studied and forms the basis of much turbulence simulation these days. Since it's impractical to model the flow down to the smallest scale, the small-scale properties are replaced with models of the turbulent energy and turbulent energy dissipation rate. The big result in that field is 'K41 theory', K for Kolmogorov and 41 for 1941, when it was published. Among other things, this predicts the form of the energy spectrum of homogeneous isotropic turbulence, so you can determine how much of the turbulent energy is stored in each size of vortex. That result has been experimentally confirmed in atmospheric turbulence, superfluid helium, and a bunch of other systems.
As far as controlling the form of a snowflake, I've seen some stuff along those lines. The main problem with predicting nonlinear systems like turbulence and snowflakes and population dynamics is that if you need to calculate a microscopic state, it is naturally impossible because of chaos - you have a sensitive dependance on initial conditions. On the other hand, if you wish to calculate the bulk properties, it works quite well - it's a _statistical_ approach. The point here is, the political system of a country is a bulk property, not a specific microscopic state. You might not be able to predict what person will get the presidential nomination for a particular party in a century, but you may very well be able to predict certain things about what their party's platform will be.
Slashdot Mirror
You can't do meaningful experiments without some idea of what the theory says will happen. Numerics of this sort provide that for complex physical cases which are essentially impossible to work out with pen and paper. So yes, this is a step towards getting knowledge of the universe and how it might work.
Also, understanding does NOT require the tie to experiment since you can have mathematical understanding of a particular theory independant of whether that theory properly models reality. For instance, I can go and work out what orbits look like in a five dimensional space. If I go and check my results versus reality 'hey, it doesn't match up!'. So from that point of view, all I've learned is that space on that scale isn't five dimensional. But lets say I don't even bother to check versus reality. I've still learned something about the mathematical properties of the theory and I've gained intuition about how things behave - namely, I've learned that closed (classical) orbits seem to only be able to exist in 3d. How strange! And I've learned that if I had a situation where the system was, say, restricted to a lower dimension (examples in electromagnetism) then I can expect large changes to the dynamics. Or perhaps a better example is, we can learn a lot about phase transitions in three dimensions by doing problems in four dimensions where they can be solved exactly and then doing an expansion around the four dimensional solution to approximate the solution in 3d. That approach doesn't depend on the underlying Hamiltonian you're solving being the correct one for some physical system - it is a purely mathematical understanding which can generically be applied to many different theories. So the benefit is, in the future when I find a better Hamiltonian for my phase transition, a better dynamical theory for gravitation, etc, I can apply the techniques I've learned from before to those as well.
Why, that would just create an arms race as then ad checkers would adapt to the adaptations made by the ad makers. The end result of such would be that DVRs would evolve sentience and ads would evolve bacteria-like mutability and randomness. Sort of like spam blockers and spam. I for one... aw, skip it.
Sure but the tradeoffs still exist. Hardware acceleration means money towards that particular card means money that isn't spent on other aspects of the system. Also, restricting physics to the hardware puts a limit on the sort of physics that can be done if it HAS to go through the hardware. Since most physics can be expressed in terms of or at least take advantage of large matrix operations, that might be the best and most general way to go. Fourier transforms on the hardware are useful for certain kinds of physics, but I doubt it'd be too useful for the kind of thing that goes into games unless people start solving the PDEs for water ripples or whatever. But I don't actually know how its planned to go. Some of this could probably be taken care of decently by making graphics card interfaces a bit more general - since you can already do a lot of linear algebra stuff on 3d cards as it is. Afaik, the limit on graphics cards has more to do with sending information to the card and receiving it back than the actual computational difficulty of what it does, which would mean that on average there are some spare cycles that could be utilized there.
Specificity of the hardware was a big problem with the first 3d hardware. Basically the game's appearance was limited by the card and not the programmers. Its a bit better nowadays with shaders, and you can actually use some tricks to do general linear algebra on 3d cards but its not as fast as it could be since the card wasn't made for that purpose.
The problem is tradeoff. Lets ignore development time, since putting in an actual physics engine can actually speed that up if it means you don't have to explicitly program stuff, the same way that putting in random generation stops you from having to tweak each individual NPC's name and appearance.
No, the tradeoff I'm talking about here is system specs. Oblivion is a good game, immersive, fun. It was a good game back when it was called Morrowind and ran on computers that would be laughable by today's standards.
Now, I'm not saying they should have kept the graphics the same in Oblivion. But I'm saying they could have and if the rest of the game is solid, that wouldn't make it a not-fun game. Since systems have gotten better, then they can choose to improve the graphics and thats great. But the sort of casual physics they're talking about in that article isn't something that could realistically be done on modern systems. I was even surprised that the physics in Oblivion could be done until I realized that they had an on-off switch for it, so stationary objects weren't simulated. Meaning they had to do at most a couple dozen nodes at once - not a big deal.
Or, to put it another way. I can make a game that solves compressible Navier-Stokes to derive the weather patterns so that the player can influence the weather via the butterfly effect. Or I can stick in a random distribution. If its free, I might as well do the former. But it isn't, so if I want anyone to be able to play my game, I choose to do the latter which is almost as good. Putting in by hand swirling smoke gives you something which takes you as a developer more time to do, but the benefit is that the computational difficulty drops and you have spare cycles to do even more interesting stuff. I'd rather have my cycles used for a really clever AI, or even an evolving world, than simulating the grass. And since I have a finite computational power, thats a choice that must eventually be made.
It's only a double standard if you take all people to be a single entity. Realize that in any sizeable group of people, you should hear conflicting opinions expressed. If you don't, thats stranger than if you do.
Another explanation is that when large companies do business, their output is often somewhat homogenized since they have to target a broader audience to sustain themselves. Whereas smaller companies/developers can be niche. So a statement like that can be in support of niche development which is optimized for a smaller community at the cost of a smaller audience and a lower chance of success in business. And from that point of view, it makes a bit of sense - the large companies are seen as blocking the market, so one wants them to fail, whereas the niche developers are doing new stuff and would take a bigger hit from each illegitimate copy, and so those you're accusing of holding a double standard wish to support them since they so obviously need it.
Of course with things like mods that sort of falls through. Most game mods are done for the love of the game, not the love of the dollar, and would be made with or without financial support. With that in mind, encouraging mod makers to become commercial is sort of like pushing 'linux - everywhere!'... the whole idea of being a part of the market ends up being more attractive than the actual thing being produced, and you're back to big companies again.
Of course, I can't speak for any other person, so this is just a set of possible interpretations, not a definitive answer for other's motives.
Myself, I'd prefer if everything were developed freely and just because the developer(s) wanted to do it - including larger projects. I don't even really have a problem with support via donations for the larger projects which are very time-consuming to develop for - but I haven't seen severe consequences or thought that through to the end so maybe I'll change my view on that when it becomes more common and if/when faults become evident. 'Everything' developed and released for free is unrealistic of course, so I don't expect it to happen. But 'some projects' is quite realistic and has been happening even more of late. I'm certainly not going to encourage the reverse direction by paying for fan-made modifications to a game. But I can understand that some people would, and why some people call it a step forward.
Spoofing the UI has been done in other cases before, so I don't think it'll provide much of an obstacle. I've seen tons of banner ads that are made to look like an windows error message. I'm not sure how effective that sort of thing is, but I imagine it gets the same sorts of people who wouldn't notice strange URLs or who don't look to see whether the site they're interacting with is using encryption (thats a UI icon too, but most people probably don't even know what it means).
Turns out this has already been worked-around. The format used by Oblivion is shared by a few other games (NetImmerse format) with a few modifications. There's a plugin for 3dsmax version something or other that outputs NIF files for Civilization 4, which can be imported into Oblivion but lack physics data (so no ragdoll, etc). It seems to work for buildings and terrain statics though, which just use a bounding box for collisions. So given this is just two weeks in, I don't imagine it will be long before the community figures out how to port over Havok data from other models, and from there starts to modify it.
And if I personally don't reply to spam, that will cause it all to go away? Unfortunately, silently not-buying gives you the power of a drop of water in an ocean. Spreading your concerns to others is a legitimate way to magnify that effect.
If you have a vested interest in people making good games that you'd be willing to shell out for, then seeing travesties that make you say 'nuts to this, I'm not buying it' doesn't just mean $2 sitting in your wallet that you wouldn't've had otherwise, it means $2 sitting in your wallet instead of going to pay for something you think is worth the $2. So I think its perfectly reasonable for fans to get upset when the company who has been producing content that they like suddenly changes directions on them. It doesn't give them license to burn the headquarters or anything, but it certainly gives them a good reason to announce their displeasure to the world in the hopes that other people who are sitting on the fence will listen and will decide not to buy the horse armor, or any addons of this sort, or even decide to get the PC version instead of the xbox version since there at least community content will easily replace the mods-for-pay.
The correct approach to this problem (of simulation in general) isn't to start at the lowest possible level and crunch. That will easily run into the issue you're talking about, which boils down to an O(e^N) dependence on particle number rather than the polynomial dependence that classical physics gives you.
To simulate such a huge system, the most feasible approach would be to start with an abstract high level model, simulate it, and see what it fails to produce. Then, refine the model down to lower and lower levels until you reach a turning point where the additional accuracy gained from the refinement is small compared to the increase in computing cost.
Sure you won't get a 'to the level of string theory accurate' computational model of a guy saying 'My name is Phil.'. But if your model of the guy says 'My name is Phil.' without needing to go that low level, then that actually tells you some interesting things about the brain. And if your high level model fails, well then you need to try something else/lower level possibly. But at least this way you don't waste time simulating things you don't in the end actually need to simulate to get the qualitative behavior thats the motivation for the simulation in the first place.
But you do have to make sure that your abstract model is consistent with the underlying stuff. If your abstract model is impossible to arrive at from the lower levels, then you're going to have some problems.
The thing is, the first form nanotechnology is likely to take will be exactly that - mutating organics to produce the desired behavior. Though its likely to be bacteria and not viruses. A ribosome is already an excellent nanoassembler :)
The problem with this sort of fear is that underlying it is the tacit assumption that any artificial life can easily and accidentally instantly outcompete every similar form of biological life. Yes, there will be some things that the standard inorganic mechanical self-replicating nanobots that are often envisioned would be better at - the rearrangement of inorganic compounds is probably easier for something that uses nanoscale mechanical methods rather than sequences of downhill chemical reactions - but in any sort of biological arena they'd be competing with billions of years of evolution. And would be made of materials not as easily obtained in a human body, so spreading via infection would be more difficult.
If someone was to spend time engineering something specifically to kill humans, then that could probably pose a significant threat - after all, most diseases don't evolve to immediately kill their hosts since that would inhibit the spreading. But an accidentally flipped bit isn't going to do it.
Really, the whole 'artificial' thing is a red herring. It doesn't matter what the nanobot is made of. Organic chemicals, inorganic chemicals, etc. Frankly, if you want to engineer some sort of nasty self-replicating human-killing thing, you're probably better off starting with something that already self-replicates and already interacts with biological reaction pathways. Just because you make the jump to silicon doesn't mean you suddenly get to break all the limits that are in place on the performance of biological stuff. You're likely to get a whole new set of limits right on top of that.
Complete chaos is a lot more predictable than incomplete chaos. Incomplete chaos, you have to worry about when its ergodic and when it isn't. If there are aspects which are totally random, or at least sufficiently random that for all intents and purposes you can't predict the exact sequence of states then you can just use the distributions. The end result will be a theory that becomes more accurate the larger the system it's used to describe. It'll fail utterly on a group of three people but will work brilliantly on a group of three billion.
Theories developed via application of the scientific method have the ability to predict with far-above-chance accuracy future events and thus can be used as the centerpoints of design - we can build machines, etc that rely on these predictions and which as a result work and do the things we desire. The same cannot be said for 'the gods are angry'.
Only that few characters? I was thinking in terms of 32 character passwords (which amounts to a 128 bit key) and such. I don't mean a login password obviously, but for something like an encrypted harddrive, a 32 character password isn't too unreasonable (take the fifth letter of every other line on two specific nonconsecutive pages of a particular book or take a sentence that you have memorized and use the first character of each word or something like that). You could probably memorize the necessary information for either of those techniques without too much trouble. Though it does make one seem a bit paranoid.
Part of it is just to make it hard to get _everyone's_ data. While cracking one guy's password may be easy to do if you really want to know what he's been doing, if that costs a couple million in computing resources (not precisely the hardware, but rather the computer time, cryptanalysts, etc) you're not going to be able to simply do it to everyone. So I guess the question is, how long does a password-based encryption need to be these days to cost at least $1M in computing resources to crack?
I'm following you up to 'grows exponentially with computation time'. You mean that because of the exponential divergence of initial conditions? I suppose it depends on whether you want to extract an exact future system configuration or whether you want statistical information about turbulent flows. Throwing a smaller timestep at it won't make a difference anyhow if its the errorbars in your initial condition that are diverging with the chaotic dynamics, so really for long times the best you can hope for is a statistical description - that is, map out the entire region of phase space where the turbulent flow may lie.
I don't believe I ever said that any possible set of axioms would give you a self-consistent system. Now, I suppose it is a question to whether there are any self-consistent systems at all. If you could somehow prove that there are no self-consistent systems when they're sufficiently developed to include e.g. integer arithmetic then you might have a point. But I don't think Godel lets you do that any more than it lets you prove that a particular one is self-consistent.
No, I think I do understand; my point was that what I said in my original post is not ruled out by Godel. You can define your own axioms and play in that playground and prove things within that framework. The overall consistency of the system is not necessarily an issue so long as the regions in which you work do not actually violate that consistency. Proof of the overall consistency is not a requirement for proofs within the framework.
It seems to me though that you can generate statements that are trivially well-defined. An example would be some simple arithmetic upon a subset of numbers. If I define a deterministic procedure which lets me add two specific integers, then by applying that procedure I will always get the same result. How is that not a proof of something, even if it's something trivial? While its possible for me to generate certain statements which cannot be proven or even considered without the addition of axioms, that doesn't mean that there are no statements that I can prove.
Well any progress is progress. Reducing copyrights is a much more achievable short-term goal than eliminating them. The reason I don't like copyrights in general is that it makes certain forms of communication between two parties a crime against a third, even if that third is not explicitly involved in the interaction. Allowing people to control the distribution of what amounts to ideas means allowing people to use the law to ban ideas, restrict ideas, etc. I don't agree that control of one's creation is defensible from a moral perspective (at least from mine...) because it interferes with what to me is a deeper right of free communication.
In the long run I can't see that as being beneficial to the growth of science and the arts, even if it does encourage some people to produce things when they would otherwise not. My background is in academia, and you really do see a lot of research and ideas being produced and then made more or less freely available (journal publication models really need to be revamped, but look at arxiv and citeseer). So I'm not at all convinced that copyright is necessary to promote innovation.
Basically what I hope to see in the coming years is so much 'amateur' but freely available content that commercial production of content which centers around its restriction becomes so unprofitable that those companies go out of business or at least that the market becomes reduced. We've seen it happen partially in the opensource movement - its not really that non OSS companies have gone out of business or lost all their customers or anything that drastic, but on the other hand you can have a fully functioning operating system with most of the supporting software you'd want or need without shelling out a computer tax. You can see it sort of happening with some forms of music (the tracker community is quite active, for instance, but there's not as much quality amateur music with vocal elements from what I've seen). We've seen a few scant examples with video - Star Wreck is the main one that comes to mind.
My main fear is that some of these licenses may go from things which encourage the growth of that community to things as bad as copyright in stifling that growth once the community becomes sufficiently large. Look at some of the license compatibility quarrels in OSS. It prevents good programs from being made because of the legal hurdles, not even necessarily because the authors wanted to prevent that specific usage. When you have projects with a thousand authors, it becomes impossible to go and ask every author 'hey, want to let this other project use our code even though they have an extra attribution requirement and don't require no-encryption?'.
And of course anything like a license will have loopholes and tricks which people can exploit to apply force to others even if its against the original purpose of the license. This isn't so much an issue with standalone programs - the author can put whatever license they want on it. But it is an issue when a choice of license is forced because the author incorporates some external code.
Anyhow, just my thoughts on the whole copyright issue.
Math theorems don't necessarily have to apply to the physical universe. The axioms on which the theorems are built are explicitly part of the theorems, leading to a logically self-consistent system. That is, you define the particular 'universe' you want to study by setting down axioms, then you prove things which you know are true about that universe because you've derived them in a logical fashion from those axioms you've set down.
Since the discussion here is about the GPL then binaries in terms of compiled code is the most in-context case. But I wouldn't mind extending that to any sort of data in any shape or form.
Your second paragraph states my position accurately.
I realize that and I was attempting to point it out with my second paragraph. My point is that I'm willing to take the risk that I cannot defeat a technological measure over the risk of jailtime or large fines should I succeed or even attempt to do so. From my point of view it's better that information is not restricted by law than that I'm guaranteed the ability to force people to not use my information in a way that restricts my later access to it via technological means.
Personally, the main thing I ask out of the GPL and other OSS licenses is simply to prevent others from taking that code and making it illegal for me to distribute it and binaries generated from it (or modifications of it). As long as distribution, reverse engineering, etc aren't illegal, I don't see the GPL as being nearly as necessary as it is today. Actually forcing the person to distribute the source and binary together is an added effect, but it's largely secondary in my eyes to simply keeping actions involving the code and program from being contraband.
DRM essentially threatens the most important part of the GPL (in my eyes that is) without harming the weaker part. Now the binaries can be distributed but you can't run them if you don't have the user-specific key to unlock the DRM around the binaries. On the other hand, if it were legal to circumvent DRM, I hardly think that any such methods would prove practically effective. So I'm game for taking my chances in a world where someone can't use a license to levy legal threat against me using the information I have access to any way I want, even if it means I can't use a license to prevent people from obfuscating that information and making it harder for me to actually use it as I'd like.
I did in fact intend phase diagram, not phase plane. The idea is, there are different distinct states of the system, such that in different states different symmetries or qualitative properties exist. As some parameter is adjusted, there is a change from one state to the other. When that change is discontinuous, you're seeing a phase boundary. Systems very close to the phase boundary may exhibit certain well-studied behaviors. In particular, if a system is near a critical point there are many things that may be said about its statistical properties.
Linearity has nothing at all to do with the success of statistical mechanics. In fact, to a good extent the more nonlinear the system, the better stat mech applies. The reason for this is ergodicity. A purely linear system will have certain invariances that prevent a random distribution from accurately modelling the population. In essence, the linearity makes it so that the initial state of the system controls its state from that time on - there is no interaction between the different bases. A nonlinearity added to the system permits mixing between states which, averaged over a large time, will tend to produce distributions that are well-behaved and predictable. Hidden symmetries in the rules may of course cause this to fail - there is some constraint which your statistical sampling fails to acknowledge, and so you miscount.
You mention turbulence. The statistical mechanics of turbulence is actually studied and forms the basis of much turbulence simulation these days. Since it's impractical to model the flow down to the smallest scale, the small-scale properties are replaced with models of the turbulent energy and turbulent energy dissipation rate. The big result in that field is 'K41 theory', K for Kolmogorov and 41 for 1941, when it was published. Among other things, this predicts the form of the energy spectrum of homogeneous isotropic turbulence, so you can determine how much of the turbulent energy is stored in each size of vortex. That result has been experimentally confirmed in atmospheric turbulence, superfluid helium, and a bunch of other systems.
As far as controlling the form of a snowflake, I've seen some stuff along those lines. The main problem with predicting nonlinear systems like turbulence and snowflakes and population dynamics is that if you need to calculate a microscopic state, it is naturally impossible because of chaos - you have a sensitive dependance on initial conditions. On the other hand, if you wish to calculate the bulk properties, it works quite well - it's a _statistical_ approach. The point here is, the political system of a country is a bulk property, not a specific microscopic state. You might not be able to predict what person will get the presidential nomination for a particular party in a century, but you may very well be able to predict certain things about what their party's platform will be.