What a complicated way to say it:)
for that matter , homeomorphism isn't required if I remember correctly as it applies to topological spaces and requires continuity.
Bijectivity is enough.
Also, for higher orders, take f(x)=cte and everything equals everything:)
But anyway, it's more of a logic problem isn't it ?
donkeys are less than 50m tall, I'm less than 50m tall, therefore I must be a donkey.. no wait.. I really am... err... oh well..
I don't think it's just a kinetic energy problem.
In the end, the bullet starting at velocity zero will gain as much kinetic energy as you give it, so heavier bullets will be slower and have the same kinetic energy than a regular one.
Let's say you give the bullet energy E, and it becomes pure kientic energy (no loss). than for a bullet b with mass Mb=k.Ma (for bullet a), you should obtain since 1/2 Ma.Va^2=1/2 Mb.Vb^2
Vb=Va/sqrt(k) so that it is slower for a heavier bullet
However, something conserved during an impact is the momentum, and Pb=MbVb=kMa.Va/sqrt(k)=Pa.sqrt(k) which is the one increased factor.
I really don't know anything about weapons, but i (therefore) think it must have to do with momentum rather than with kinetic energy. Also, its greater density makes it less likely to deform under impact, so I suppose that 'more energy is transfered to the target' as there is less loss due to deformation (of the bullet, of course, target deformation must probably be wellcome in some way:))
They mean for free fall
The batteries are at the bottom of the device putting its mass center at its feet
Also, it has a much better aerodynamic design taking advantage of it, so that air resistance is futile:)
They should have learnt now, and pay more attention to what I am doing
To me, Pulsar's ages have been a mistery since the first time I ever heard of them....
No. It's a threefold increase in crackable key-size.
If you could crack a 10 digits key, it will now cost you the same to carck a 30 digits one.
So, as the cost does grow exponentially with the key size, it is correct to talk of "order of magnitude".
Read well: *cost*.
If I understood well the article (ok, I probabily didn't), then it's not such a big deal (I think).
You are not cracking any faster than before. it is not an improvement so that now everybody is going to crack easily huge keys.
It's about making special purpose hardware in order to crack something more eficiently.
For instance, in the article it presents a wonderful sort algorithm that sorts N values in N "steps". Cool:) but you have to see what are the steps. on a regular computer, each of those steps costs you N caculations. so this is an O(N^2) algorhitm for you and me. By redefining the "unit of caculation" or "step" (which isn't just an ugly trick, it is meaningful with special purpose hardware) you can transform this very bad algorhithm into a very good O(N) one.
But the article is even less interesting than that (on my point of view) because it doesn't come with such a trick for factorizing integers. instead, it comes with an observation on overall cost. (which you and me generaly translate as time, but not everybody else)
The thing is, not only do you need O(f(N)) time to compute, with f *nasty*:) but you also need O(g(N)) memory, with g similarily nasty. the cost involves the computing time but also the hardware itself. If you find a way to make the algorithm parrallel, you divide it in 10 computers, so it will be 10 times faster, but 10 times as expensive and the cost is the same.
In this article, he proposes a different special purpose circuit that will allow you to have a somewhat less nasty g, so the O(g(N)) cost will be much more affordable.
In the end, it will cost you the same for keys now with 3 times more digits.
That's nice and all, but I'm sure it would be much more significant on general purpose hardware, even if not so wonderful at first sight, because something that is not considered in the article is 'real life cost of hardware'.
I mean, spcecific hardware, because of its specificity if of course much more expensive than general purpose hardware. There is also the availability problem, so in the end I don't find it a so interesting achievment.
It would be much more effective if for instance, you could easily maximize parrallelism with aceptable latency, so that you could build an Internet wide cracker, that would be much cheaper (I think).
How is this "unsafety" managed ?
I really wonder how the VM knows the code is safe or not.
If it is just marked at compile time, then it won't be long till a modified compiler will mark just about anything as safe even if not. We all know why and from who.
So how does the VM executes new downloaded code ?
Does it verify it completely before just to know if indeed it is safe or not ? and then isn't this a rather long process ?
How does this work ? in the end where do you want your trust to go today ?
Something cool also would be to translate mp3 links into a "lalala pattern database":)
This way, you could make a search for a song you don't know the title, or not even who sings it (or plays it if it's just music.)
You just sing how it's like to the microphone, and it looks for similar "lalala patterns":)
I already made somthing like that : you sing what you want and you get this answer:
6 matches:
- R.E.M: losing my religion
- Rammstein: Wollt Ihr das bett in flammen sehen
- Tatiana: yukaidi yukaida
- Alphaville: Forever Young
- Farinelli: Alto Giove
- Fight Club: theme
There is a link for useful feedback where users generaly complain they didn't find what they were looking for, but they get what they deserve:
"You are a loser and should learn to sing before complaining. what you sang 'ladadiladadada' (emailed to all in the lab) *did* match those 6 titles"
I'd see another (more complicated) solution:
Having a central server resposible for synchronization, but leting clients do most of the job.
That could involve some cheating, but not that much if you don't let them do just anything.
The map is something they can have localy. it doesn't mean they can change it, if there is a change they have to be synced about it from the server. if they want to make a change (using a spell or something) than the server validates it and propagates the changes to all the clients.
But to have a very good performance, you really need to let some cheating capabilities to the client.
ie. letting them handle some of the gameplay. You could however make it very dificult using a kind of p2p system.
The idea is, the server knows about all the world changes etc. but only let clients know about it whenever necesary. each clients encounter happens "localy only" and is handled by those clients.
the clients handle it by putting one of them as the temporary server for the encounter.
The election of who is the local server (which is not a complete server as it is a reporting slave from the main one and cannot do aboslutely everything itself) would eb with an algo. that involves calculating best computing capacity+best average relative pings+some random maybe.
So there could be some cheating, but this would require a handful of co-cheaters working together on a previously agreed encounter.
The global effects of their cheating can be greatly disminished if the local server has no spawning ability (things like that, which means, when the control is transfered form the central to the local server, some variables, like XP etc.. have a precalculated max. so they cannot tell the server "hey now I have x ammo or x xpoints etc...)
so if it involves unimportant NPCs, they can be handled by the local server, if they are important NPC, the central server is involved in the equation and is then the chosen local server.
The only cheating that can be done there is (I think) that the chosen local server has power to destroy its oponents. Still, at reporting time,from the clients and the local server, anomalies can be detected.
But the reports can be transfered to a sub-server that analyzes the outcomes with the known variables, to determine if there is sure or probable cheating. (it can takes its time, after all the actions will be taken in due time). if there is only probablñe cheating instead of sure cheating (ie. the server was just an unlikely good player or extremely lucky on random dices) than it can be "marked" as probable cheater. and be tracked. f a patron of him being extremely good when server and just average or bellow when client, than we have a smart cheater.
It depends on what kind of calculations you are doing,
But if they are time-sensitive, then besides all the almost purely hadware solutions so far proposed,
You'd also need to have os system calls that give you fake date/time related answers. (to be different of course of the 'real ones' also available)
If you want to have per process "cryogenics" you'll also have to keep track of the different date/time status for each one of them, whoich could cause trouble if they comunicate now and then...
I don't think this is that much similar, and I fail to apreciate the advantages:
One advantage of this approach is that if you're installing on multiple identical machines, you would only go through the process once. Once it's done, you'd have a set of "instant" install CDs. No menus, no further tweaking, just a direct blast onto the hard drive(s).
But you can do just the same localy with a Mandrake-like duplication system. I'm not saying this isn't an advantage, but it's not intrinsicaly due to the fact you built it on the server site.
A second advantage is that a server site can have a compiler farm, making the build process MUCH quicker than would be possible for an individual.
This depends a lot on how may users are going to use the server farm at the same time, and in the end, your farm will have finite resources while you can have 'infinite' distributed computing resources. But ok I don't think everybody would ask for compilation at the same time:)
A third advantage is that if someone sends in a configuration which matches one that's already been done, the compiler farm only needs to rebuild updated packages. The rest has already been done. The CDs can then be built out of freshly-compiled binaries and pre-compiled ones.
That's my main objection, I don't think you are going to find many matches here, first there are many many different hardware architecture.
Just think about the cpus, ss2 ? mmx ? floating point error ? smp ? you would end with a lot of different packages just to match the hardware, but my main concern is that from here, you explode to match the many many different configurations options that every package has to offer. I think that those who want to fine tune their system to the point they recompile everything will want to have their say about having php with or without support for say, mysql, they will want to have their say about how appache is built with or without mod_perl, and they'll definitely want to decide if they want a static build or dynamic libraries for this or this package. they'll want to decide what libc they want, and imagine the kernel options ! they'll want ext3 here, reiserfs there, not as a module, they'll want xfs here but as a module, with frame buffers, without, etc...That's a whole lot of different packages...
A fourth advantage is start-up time. Because you're downloading a very basic bootstrap, rather than a mini-distro, the time to download, install and run is going to be much much less.
Well yes, but then you'll have to download your customized iso. you could as well download a very basic mini-distro asking you everything you want about everything that would then only download the sources you will need and then creating the iso on your machine and you would also decrease total downloading time, but it would be extremely inconviegnent because you would never have a 'usable cd' as in 'usable anywhere, independent'
Also, your server has to build a new customized iso each time someone wants one, because even if for some reason you do have all the possible binaries anybody would want, you will still have to put them together on the unique way everyone wants it because one wants kde, another gnome another xfce etc. and of course you wouldn't waste bandwith sending him those binaries if he doesn't need them. So you will have to send him a customized iso with a specific newly computed name to avoid confusion, that is, it cannot be cached anywhere by any squid-like caching proxies (ok not so much an issue maybe) but it also cannot be mirrored think about how your server farm will have to deal with just the bandwidth that becasue of you very definition of the site cannot be shared. How many distributions aren't backed by a handful of mirrors ? and they don't have so high bandwidth requirements because the very same iso is usable by many people on the same area, if I download it, all my friends can use it (and they do). in this case, everybody has to download his own iso. the only one guy who will see an advantage here is the one who has 30 identical machine with the same requirements. cool, but I bet he will have at least one server with, if the same hardware (unlikely) will surely have different configurations (kernel with packet filtereing etc... different software, squid, apache, kups or lpd...) so this will be a second download. when he could have dealt with it just with one download in a sourcer's case. so you'd have to compare 2 download times plus 2 remote compiling time against 1 download time versus 2 local (but parrallel) compilation times. I don't know many cases where this will represent a quicker install in your proposed configuration.
The last advantage is when it comes to updating your system. Again, with all the compiling being done on a remote compiler farm, the time it would take to do a basic update would be minimal, compared to Sorcerer, and far more optimal, compared to Up2Date or Red-Carpet.
Hmm I don't know this sourcerer but it seems you're implying that the only way to update your system is to burn a new cd and then install all over again (??) you can do that to have always the latest install cd, but I don't think you have to do it just to update your system. I suppose you just have to recompile the few instaled packages that have update which must be done by dowbloading just their patches (I certainly hope so !)
Now that was something !
you know what ? I don't even know where to start..
So, let's see.. My Boss, as many others *knows* Microsoft invented the Internet, computers etc..
If I start telling him ms word's.doc format is a "proprietary secret format" There is absolutely no way he will understand what the f*ck I am talking about. And in fact nobody sending me.docs attachments is going to.
And if I even care to tell him I am *puzzled* because he sent me a 237,545 bytes text in 867,345 bytes he will realize I'm just some kind of morron from another planet and fire me before I cause any harm to his business.
While I do agree.doc attchments are a pain in the ass and it would be nice to do something about it, RMS is not giving us any help to achieve it.
In fact, it's probably a way to get things even worse, not by having people send even more MS docs attachments, but by having them looking at any open source advocate as a fucking morron.
And that woul be more harmful to the movement than just keeping quiet about this annoying fact.
RMS is trying to convince (for some reason) a primitive canibal tribe that they should not sacrifice any more people to their God, the sun, because it's just a big ball of gaxs etc.. etc.. with an extended astrophysics course to support his claims. Cool, he's right, but they'll have him for dinner.
Nope this isn't for you.
they say in the article they are focusing on compatibility with key applications (office etc...)
They aren't trying to completely replace windows with a 100% compatible os, jut to port enough of the API so that office and some few other major applications work well enough.
prove me wrong, please:)
My pleasure:)
with the same seed:
dd if=/dev/urandom of=random bs=1M count=10
Here you have your original string amazingly compressed.
Of course, this isn't a general compression technique as it is only aimes towards compressing those pseudo-random strings.
But it's very simple, very fast, and have a much higher ratio than 100:1.The thing is, it is not based on repetitive patterns to compress, but rather on "encoding".
Now anyway, I don't see how you could deal with true random data and I don't realy think the article is much more than buzz, but I don't think it is proved that this result is unachievable.
At best, you could show it is impossible to achieve using pattern repetitions techniques.
I can think of several different ones, although they'd admitedly be hybrid ones.
For example, another idea I never gave a shot was this one:
Use some kind of sorting algorhithm to "shuffle" your original data. The algorithm is basicaly a random seed specified sequence of (inversible, have to see how to do that, would be two reciprocal algorithms that use the same seed to work forward and backward) sorting comands. (sorting or any other way to distort the data, I was thinking about a bubble sort because it's a nice vision to see all those 1s coming together at the surface letting the 0s under them (or the contrary)). Anyway, the trick would be to find a nice algorithm or a set of nice ones (you try several ones or the ones that statisticaly suit better your data type if youy have some knowledge about it) that reversibly distort your data with a simple seed into a hihgly compressible new data.
Depending on your original 1s and 0s distribution, some seeds will be more or less effective, I'd say a few tries should always come to something at least compressible beyond 1:1 which would be better than what we currently have. But 100:1:) As Neo would say.. "Wow !" dunno:)...
Oh well...I'll have to think about it better someday
Well it's not what it is, it's what it could be
There are some big privacy/security issues here and noone wants to let Microsoft handle that because Microsoft is not to betrusted.
So we are all against what Microsoft could do with such a "device" knowing Microsoft practices.
It's kind of the same problem we have with the DMCA saying that this device could be used to pirate music. Exept in that case it's bad because everybody else should be trusted in the first place , not Microsoft:)
Oh well I'm not only sarcastic, actualy I'm both against Microsft passport idea and against the DMCA, I'd however like to see better arguments against them (some that don't overlap at least).
Not that there aren't.. I just didn't see them thus far.
LOL
if you did read the article to know how the stuff is supposed to work, then if you want to see the inside of someone's body instead of inside the sphere, then you can imagine how it's going to be done.
At last, our Trolls favorite guy doing something useful for science.
(for the preliminary tests, until they can do with something smaller)
You seem to have a pretty warped sense of what entails a "natural" conclusion
Maybe:) If you mean that such a conclusion is a result of unforgivable ignorance
Actualy I don't understand what you mean, but I'm pretty sure it's not what I meant:) Otherwise, that's one of the most idiotic statements I've read...
*yawn*
Yeah well:) like you read a lot:) There is a huge difference between PROVING THINGS IN A FORMAL SYSTEM and KNOWING THINGS
Of course ! it just happens that in this case one follows from the other, it's nt like it's a general rule, I don't think you understood what I meant, but it must be because I wasn't clear about it. When you can grok that in fullness, you might redeem yourself for saying something so blatantly stupid as what I quote above.
*LOL*
I did 'grok' that in fullness and what you quoted alone lacks of context to be stupid by itself. Once in the context it is even less so I certainly won't 'redeem myself' but thank's for the good laugh anyway.
Look, here is the idea:
Before Godel played with the Diophantine equations version of "this statment is False", mathematicians didn't even think about such a thing as completeness/incompleteness of an axiom system.
Hilbert worked hard to formalize Mathematics so that people wouldn't argue anymore about proofs. Instead, they would just "compute" a proof with a mechanical mean of applying iference rules on the axiom set. Proofs would be computable statements leading to "true" or "false".
Godel's proof put an end to the beautifl dream stating that not all truths will be computable, unless some facts would be both true and false.
What does that mean ?
It means even if we carefuly chose our axiom set, that would form our "knowledge base", some truths couldn't be deduced from it, that is, they would be true just because they are. like Fermat Last theorem for example. Ok, it was proved, but it could have not been provable, so maybe we would have to add it to the axioms set.
This makes the axiom sets infinitelay growable, that I was translating as a truth that cannot be deduced from the one we already know is a knowledge we *yet* have to gain.
It follows 'naturally' that we can't have all the knowledge, there is always something else that is true, but that is not a natural consequences of the truths we already possess, making 'whole knowledge' impossible to have at any time.
Thus the 'natural' conclusion that Godel's proof indeed limits us not to know everything.
Have a nice day.
*LOL*
This isn't correct however,
one of my last physics teacher was also a guitar player thus would have been the music teacher:)
also one of my last mathematics teacher was a very impressive Harley guy. if you've seen "eaters of the dead", the viking king that dies as a hero in the end is exactly like him:)
Er. The proof is done using a discrete system. That seems to pretty concretely limit its result to discrete systems.
Sorry, I wasn't clear. I mean the result isn't limited to this particular system for those two reasons:
- It has direct consequences on logic reasoning, because demonstration are such a system. this is the most important thing there, and the one I was defending later.
- Also, it can be considered as a specific counter-example to a more general set of problems. This is a specific example to show us the thing doesn't hold, so yes it is limited to exactly where it shows how it doesn't hold, but it is not limited in the way it doesn't also tell you this is the only place where this happens. that's what I was meaning.
So you know some truthes cant be stated and here is an example. even if you don't know how bad the situation is, you can feel it's going to be even worse with more complex systems.
Take real numbers for example, there are infinitely many more real numbers that are not rational numbers, and then most of them are even transcendent, even though we only pointed a few of them.
So maybe "bad truthes" are like transcendent numbers. maybe most of truths aren't statable.Godel just proved mathematics to be incomplete "because of this" but its proof doesn't tell us "it is just incomplete here, and only here".
And it indeed is worse than that, Turing stopping problem widens the panorama of troubles, and Chaitnin "unknowable" widens it further more.
Yeah yeah, and here we call particle physicists "parti-culeros":)
But it's just a game. All my mathematics physics and chemistry teachers were always bitching at each other's classes or "way of seeing things".
"Ok, so here is a good equation that no mathematician can resolve, and we'll do that in just 1 second":)
"So the theorem is proved because 3 is not 0, unless you teach chemistry of course, then anything under ten is not enough not to be considered 0" blabla:)
And they of course were all very good friends:)
Hmm correct me if wrong but I somewhat disagree with some of your points: 1) Godel's proof only works in discrete systems that support (at least) a small number of operations.
Godel's proof is done using a discrete system, but there is nothing limiting its result there. it only lacks of substance to let us now how much we are in trouble with even more complicated problems, but it is enough to let us know that at least, beacause of those damn diophantine equations, our beloved set of basic axioms cannot exist. This pretty much kills the mathematicians "theory of everything"
Godel's proof does not say that it is impossible to know everything. It says that in these discrete systems, it is either a) impossible to make some valid statements (an incomplete system), or b) possible to make some invalid statements (an incorrect system).
Yes and no, it doesn't *say* you cannot know everything, but it is indeed a natural conclusion.
those valid statements that are impossble to make are in fact logic statements as proofs. So some truths cannot be proven whatever axiom set you chosed (unless as you say you axiom set in inconsistant, which is even worse than incomplete).So said truthes have to be added to the axiom list, growing it to infinity.
If we cannot have a finite set of axioms to describe everything, than we cannot *know* everything. we can know everything about a few things that are *covered* by a finite set of axioms, but not *everything* too bad:) or maybe it's better like that after all:)
3) Godel's proof only works if you are using boolean logic (and, in fact, works only because boolean logic is so bad at handling self-referential statements)
I still have a problem with this "only" it's correct of course, but it applies exactly on logic thus on our way to solve problems thus on our ability to cover *all problems* so it is indeed an important fact. This does not mean that the universe works the same way.
I fully agree, Godel's proof and its consequences don't limit our possibility to define how universe works with just a few (only one ?) set of laws. However, once we find the basic laws if we ever do so (which I doubdt) we might still observe some phenomenons that are not possible to explain with just the fundamental laws, creating "emerging laws" that would in fact be more than just a statistical consequence of more basic interactions.
You are however right in the general idea of your post that Godel definitely did not prove that was going necesarily to happen, he just proved it is a possibility.
At least that's how I understood it, but I can be wrong, that's for sure:)
The biggest mistake RMS did was probably to call his 'vision' "Free Software"
People get extreme around this word.
First, "freedom" as "absolute freedom" is a nonsense and cannot exist. it's unfortunately not so clear a problem as "This statement is False" so people don't realize it and still want "Freedom" to be absolute as in "everything I wever want". So they get in big ideological troubles, all of them rooted in the fact that they both understand Freedom as something different and generally too large, yet to strict to admit anyonelse's "freedom" (freedom definition, not freedom itself, although freedom itself sometimes).
Second, freedom has this great "added value" that everybody cares about it and would defend its own (definition) with so much conviction yet so little skill it's the mess we're witnessing every so often in slashdot.
Look at your own words, for instance:
"EX. I may not like Microsoft bashing Linux, but I will defend their right to do so."
So far so good. "Stallman needs to recognize this and embrace it."
Uh ?
Big dilema here, how would you defend someone's freedom to fight your freedom of defending his freedom ?
Not what's happening here, but it's the same problem:)
In the end, I think you should stand on this "I'll fight for your rights to..." as it's clearer and easier. I.E. you know what you are fighting for. (of course you'll have to fight for someone's rights to advocate censorship, but it's still simpler)
In the case of Stallman's "Free Software", you'll have to remember he's standing for something he started himself.
Also, it's not like the guy is an idiot, I'm pretty sure he actualy understands what he wants. Why does everybody undersntand what he wants better than he does himself ?
Maybe you want something else, maybe you should discuss with him how you'd like his vision to be called something else than "Free Software" so that you can use it for your own "vision" (I garantee you you'll end in the exact same situation, simply because of the "Free" word).
I doubdt he'd like it though, but it would be nice. Of course, Free Software as it is in RMS's words must just fit his own freedom understanding so that he probably wouldn't even see the need of another, more 'accurate' description. So how would something for "free" be more acurate than "free" uh ?:-)
Wether it is due to a strobing light source or a "frames capturing device" (television, cinema..) the idea is the same...
Look at your wheel , let's say it spins at (24*3)-2 revolutions per second.
If you capture 24 images per second, then you will see this sequence of images:
- wheel at position "0"
- wheel at position "-2"
- wheel at position "-4"
Of course the wheel actualy spinned nearly 3 times, but you just get the picture of it when it is slightly rotated in the other direction.
So you effectively watch it spinning backward.
This way, you can "freeze", play backward, play slowly etc.. any periodic/repetitive motion, just adjust your sampling frequency around the motion frequency. if it matches, you froze it. ...
Slashdot Mirror
What a complicated way to say it :) :)
for that matter , homeomorphism isn't required if I remember correctly as it applies to topological spaces and requires continuity.
Bijectivity is enough.
Also, for higher orders, take f(x)=cte and everything equals everything
But anyway, it's more of a logic problem isn't it ?
donkeys are less than 50m tall, I'm less than 50m tall, therefore I must be a donkey.. no wait.. I really am... err... oh well..
I don't think it's just a kinetic energy problem. :))
In the end, the bullet starting at velocity zero will gain as much kinetic energy as you give it, so heavier bullets will be slower and have the same kinetic energy than a regular one.
Let's say you give the bullet energy E, and it becomes pure kientic energy (no loss). than for a bullet b with mass Mb=k.Ma (for bullet a), you should obtain since 1/2 Ma.Va^2=1/2 Mb.Vb^2 Vb=Va/sqrt(k) so that it is slower for a heavier bullet
However, something conserved during an impact is the momentum, and Pb=MbVb=kMa.Va/sqrt(k)=Pa.sqrt(k) which is the one increased factor.
I really don't know anything about weapons, but i (therefore) think it must have to do with momentum rather than with kinetic energy. Also, its greater density makes it less likely to deform under impact, so I suppose that 'more energy is transfered to the target' as there is less loss due to deformation (of the bullet, of course, target deformation must probably be wellcome in some way
The headline says 170mm squared :)
that's not 170mm^2, but (170 mm) squared
They mean for free fall :)
The batteries are at the bottom of the device putting its mass center at its feet
Also, it has a much better aerodynamic design taking advantage of it, so that air resistance is futile
They should have learnt now, and pay more attention to what I am doing
To me, Pulsar's ages have been a mistery since the first time I ever heard of them....
No. It's a threefold increase in crackable key-size. :) but you have to see what are the steps. on a regular computer, each of those steps costs you N caculations. so this is an O(N^2) algorhitm for you and me. By redefining the "unit of caculation" or "step" (which isn't just an ugly trick, it is meaningful with special purpose hardware) you can transform this very bad algorhithm into a very good O(N) one. :) but you also need O(g(N)) memory, with g similarily nasty. the cost involves the computing time but also the hardware itself. If you find a way to make the algorithm parrallel, you divide it in 10 computers, so it will be 10 times faster, but 10 times as expensive and the cost is the same.
If you could crack a 10 digits key, it will now cost you the same to carck a 30 digits one.
So, as the cost does grow exponentially with the key size, it is correct to talk of "order of magnitude".
Read well: *cost*.
If I understood well the article (ok, I probabily didn't), then it's not such a big deal (I think).
You are not cracking any faster than before. it is not an improvement so that now everybody is going to crack easily huge keys.
It's about making special purpose hardware in order to crack something more eficiently.
For instance, in the article it presents a wonderful sort algorithm that sorts N values in N "steps". Cool
But the article is even less interesting than that (on my point of view) because it doesn't come with such a trick for factorizing integers. instead, it comes with an observation on overall cost. (which you and me generaly translate as time, but not everybody else)
The thing is, not only do you need O(f(N)) time to compute, with f *nasty*
In this article, he proposes a different special purpose circuit that will allow you to have a somewhat less nasty g, so the O(g(N)) cost will be much more affordable. In the end, it will cost you the same for keys now with 3 times more digits.
That's nice and all, but I'm sure it would be much more significant on general purpose hardware, even if not so wonderful at first sight, because something that is not considered in the article is 'real life cost of hardware'.
I mean, spcecific hardware, because of its specificity if of course much more expensive than general purpose hardware. There is also the availability problem, so in the end I don't find it a so interesting achievment.
It would be much more effective if for instance, you could easily maximize parrallelism with aceptable latency, so that you could build an Internet wide cracker, that would be much cheaper (I think).
Je suis la force qui fait rire
Je m'appelle casimir
*soupir*
How is this "unsafety" managed ?
I really wonder how the VM knows the code is safe or not.
If it is just marked at compile time, then it won't be long till a modified compiler will mark just about anything as safe even if not. We all know why and from who.
So how does the VM executes new downloaded code ?
Does it verify it completely before just to know if indeed it is safe or not ? and then isn't this a rather long process ?
How does this work ? in the end where do you want your trust to go today ?
Maybe an add-free cache could be interesting..
:) :)
Something cool also would be to translate mp3 links into a "lalala pattern database"
This way, you could make a search for a song you don't know the title, or not even who sings it (or plays it if it's just music.)
You just sing how it's like to the microphone, and it looks for similar "lalala patterns"
I already made somthing like that : you sing what you want and you get this answer:
6 matches:
- R.E.M: losing my religion
- Rammstein: Wollt Ihr das bett in flammen sehen
- Tatiana: yukaidi yukaida
- Alphaville: Forever Young
- Farinelli: Alto Giove
- Fight Club: theme
There is a link for useful feedback where users generaly complain they didn't find what they were looking for, but they get what they deserve:
"You are a loser and should learn to sing before complaining. what you sang 'ladadiladadada' (emailed to all in the lab) *did* match those 6 titles"
I'd see another (more complicated) solution:
Having a central server resposible for synchronization, but leting clients do most of the job.
That could involve some cheating, but not that much if you don't let them do just anything.
The map is something they can have localy. it doesn't mean they can change it, if there is a change they have to be synced about it from the server. if they want to make a change (using a spell or something) than the server validates it and propagates the changes to all the clients.
But to have a very good performance, you really need to let some cheating capabilities to the client.
ie. letting them handle some of the gameplay. You could however make it very dificult using a kind of p2p system.
The idea is, the server knows about all the world changes etc. but only let clients know about it whenever necesary. each clients encounter happens "localy only" and is handled by those clients. the clients handle it by putting one of them as the temporary server for the encounter.
The election of who is the local server (which is not a complete server as it is a reporting slave from the main one and cannot do aboslutely everything itself) would eb with an algo. that involves calculating best computing capacity+best average relative pings+some random maybe.
So there could be some cheating, but this would require a handful of co-cheaters working together on a previously agreed encounter.
The global effects of their cheating can be greatly disminished if the local server has no spawning ability (things like that, which means, when the control is transfered form the central to the local server, some variables, like XP etc.. have a precalculated max. so they cannot tell the server "hey now I have x ammo or x xpoints etc...)
so if it involves unimportant NPCs, they can be handled by the local server, if they are important NPC, the central server is involved in the equation and is then the chosen local server.
The only cheating that can be done there is (I think) that the chosen local server has power to destroy its oponents. Still, at reporting time,from the clients and the local server, anomalies can be detected.
But the reports can be transfered to a sub-server that analyzes the outcomes with the known variables, to determine if there is sure or probable cheating. (it can takes its time, after all the actions will be taken in due time). if there is only probablñe cheating instead of sure cheating (ie. the server was just an unlikely good player or extremely lucky on random dices) than it can be "marked" as probable cheater. and be tracked. f a patron of him being extremely good when server and just average or bellow when client, than we have a smart cheater.
It depends on what kind of calculations you are doing,
But if they are time-sensitive, then besides all the almost purely hadware solutions so far proposed,
You'd also need to have os system calls that give you fake date/time related answers. (to be different of course of the 'real ones' also available)
If you want to have per process "cryogenics" you'll also have to keep track of the different date/time status for each one of them, whoich could cause trouble if they comunicate now and then...
I don't think this is that much similar, and I fail to apreciate the advantages:
:)
... :)
One advantage of this approach is that if you're installing on multiple identical machines, you would only go through the process once. Once it's done, you'd have a set of "instant" install CDs. No menus, no further tweaking, just a direct blast onto the hard drive(s).
But you can do just the same localy with a Mandrake-like duplication system. I'm not saying this isn't an advantage, but it's not intrinsicaly due to the fact you built it on the server site.
A second advantage is that a server site can have a compiler farm, making the build process MUCH quicker than would be possible for an individual.
This depends a lot on how may users are going to use the server farm at the same time, and in the end, your farm will have finite resources while you can have 'infinite' distributed computing resources. But ok I don't think everybody would ask for compilation at the same time
A third advantage is that if someone sends in a configuration which matches one that's already been done, the compiler farm only needs to rebuild updated packages. The rest has already been done. The CDs can then be built out of freshly-compiled binaries and pre-compiled ones.
That's my main objection, I don't think you are going to find many matches here, first there are many many different hardware architecture. Just think about the cpus, ss2 ? mmx ? floating point error ? smp ? you would end with a lot of different packages just to match the hardware, but my main concern is that from here, you explode to match the many many different configurations options that every package has to offer. I think that those who want to fine tune their system to the point they recompile everything will want to have their say about having php with or without support for say, mysql, they will want to have their say about how appache is built with or without mod_perl, and they'll definitely want to decide if they want a static build or dynamic libraries for this or this package. they'll want to decide what libc they want, and imagine the kernel options ! they'll want ext3 here, reiserfs there, not as a module, they'll want xfs here but as a module, with frame buffers, without, etc...That's a whole lot of different packages...
A fourth advantage is start-up time. Because you're downloading a very basic bootstrap, rather than a mini-distro, the time to download, install and run is going to be much much less.
Well yes, but then you'll have to download your customized iso. you could as well download a very basic mini-distro asking you everything you want about everything that would then only download the sources you will need and then creating the iso on your machine and you would also decrease total downloading time, but it would be extremely inconviegnent because you would never have a 'usable cd' as in 'usable anywhere, independent' Also, your server has to build a new customized iso each time someone wants one, because even if for some reason you do have all the possible binaries anybody would want, you will still have to put them together on the unique way everyone wants it because one wants kde, another gnome another xfce etc. and of course you wouldn't waste bandwith sending him those binaries if he doesn't need them. So you will have to send him a customized iso with a specific newly computed name to avoid confusion, that is, it cannot be cached anywhere by any squid-like caching proxies (ok not so much an issue maybe) but it also cannot be mirrored think about how your server farm will have to deal with just the bandwidth that becasue of you very definition of the site cannot be shared. How many distributions aren't backed by a handful of mirrors ? and they don't have so high bandwidth requirements because the very same iso is usable by many people on the same area, if I download it, all my friends can use it (and they do). in this case, everybody has to download his own iso. the only one guy who will see an advantage here is the one who has 30 identical machine with the same requirements. cool, but I bet he will have at least one server with, if the same hardware (unlikely) will surely have different configurations (kernel with packet filtereing etc... different software, squid, apache, kups or lpd...) so this will be a second download. when he could have dealt with it just with one download in a sourcer's case. so you'd have to compare 2 download times plus 2 remote compiling time against 1 download time versus 2 local (but parrallel) compilation times. I don't know many cases where this will represent a quicker install in your proposed configuration.
The last advantage is when it comes to updating your system. Again, with all the compiling being done on a remote compiler farm, the time it would take to do a basic update would be minimal, compared to Sorcerer, and far more optimal, compared to Up2Date or Red-Carpet.
Hmm I don't know this sourcerer but it seems you're implying that the only way to update your system is to burn a new cd and then install all over again (??) you can do that to have always the latest install cd, but I don't think you have to do it just to update your system. I suppose you just have to recompile the few instaled packages that have update which must be done by dowbloading just their patches (I certainly hope so !)
Oh well.. maybe I just missed your point
It might be funny to use them against their own source and watch them go crazy *cheat* *cheat*...
Now that was something ! .doc format is a "proprietary secret format" There is absolutely no way he will understand what the f*ck I am talking about. And in fact nobody sending me .docs attachments is going to. .doc attchments are a pain in the ass and it would be nice to do something about it, RMS is not giving us any help to achieve it.
you know what ? I don't even know where to start..
So, let's see.. My Boss, as many others *knows* Microsoft invented the Internet, computers etc.. If I start telling him ms word's
And if I even care to tell him I am *puzzled* because he sent me a 237,545 bytes text in 867,345 bytes he will realize I'm just some kind of morron from another planet and fire me before I cause any harm to his business.
While I do agree
In fact, it's probably a way to get things even worse, not by having people send even more MS docs attachments, but by having them looking at any open source advocate as a fucking morron. And that woul be more harmful to the movement than just keeping quiet about this annoying fact.
RMS is trying to convince (for some reason) a primitive canibal tribe that they should not sacrifice any more people to their God, the sun, because it's just a big ball of gaxs etc.. etc.. with an extended astrophysics course to support his claims. Cool, he's right, but they'll have him for dinner.
Nope this isn't for you.
they say in the article they are focusing on compatibility with key applications (office etc...)
They aren't trying to completely replace windows with a 100% compatible os, jut to port enough of the API so that office and some few other major applications work well enough.
prove me wrong, please :) :) :) As Neo would say.. "Wow !" dunno :)...
My pleasure
with the same seed:
dd if=/dev/urandom of=random bs=1M count=10
Here you have your original string amazingly compressed.
Of course, this isn't a general compression technique as it is only aimes towards compressing those pseudo-random strings.
But it's very simple, very fast, and have a much higher ratio than 100:1.The thing is, it is not based on repetitive patterns to compress, but rather on "encoding".
Now anyway, I don't see how you could deal with true random data and I don't realy think the article is much more than buzz, but I don't think it is proved that this result is unachievable.
At best, you could show it is impossible to achieve using pattern repetitions techniques.
I can think of several different ones, although they'd admitedly be hybrid ones.
For example, another idea I never gave a shot was this one:
Use some kind of sorting algorhithm to "shuffle" your original data. The algorithm is basicaly a random seed specified sequence of (inversible, have to see how to do that, would be two reciprocal algorithms that use the same seed to work forward and backward) sorting comands. (sorting or any other way to distort the data, I was thinking about a bubble sort because it's a nice vision to see all those 1s coming together at the surface letting the 0s under them (or the contrary)). Anyway, the trick would be to find a nice algorithm or a set of nice ones (you try several ones or the ones that statisticaly suit better your data type if youy have some knowledge about it) that reversibly distort your data with a simple seed into a hihgly compressible new data.
Depending on your original 1s and 0s distribution, some seeds will be more or less effective, I'd say a few tries should always come to something at least compressible beyond 1:1 which would be better than what we currently have. But 100:1
Oh well...I'll have to think about it better someday
Well it's not what it is, it's what it could be :)
There are some big privacy/security issues here and noone wants to let Microsoft handle that because Microsoft is not to betrusted.
So we are all against what Microsoft could do with such a "device" knowing Microsoft practices.
It's kind of the same problem we have with the DMCA saying that this device could be used to pirate music. Exept in that case it's bad because everybody else should be trusted in the first place , not Microsoft
Oh well I'm not only sarcastic, actualy I'm both against Microsft passport idea and against the DMCA, I'd however like to see better arguments against them (some that don't overlap at least).
Not that there aren't.. I just didn't see them thus far.
LOL
if you did read the article to know how the stuff is supposed to work, then if you want to see the inside of someone's body instead of inside the sphere, then you can imagine how it's going to be done.
At last, our Trolls favorite guy doing something useful for science.
(for the preliminary tests, until they can do with something smaller)
You seem to have a pretty warped sense of what entails a "natural" conclusion :) :) :) like you read a lot :)
Maybe
If you mean that such a conclusion is a result of unforgivable ignorance
Actualy I don't understand what you mean, but I'm pretty sure it's not what I meant
Otherwise, that's one of the most idiotic statements I've read...
*yawn*
Yeah well
There is a huge difference between PROVING THINGS IN A FORMAL SYSTEM and KNOWING THINGS
Of course ! it just happens that in this case one follows from the other, it's nt like it's a general rule, I don't think you understood what I meant, but it must be because I wasn't clear about it.
When you can grok that in fullness, you might redeem yourself for saying something so blatantly stupid as what I quote above.
*LOL*
I did 'grok' that in fullness and what you quoted alone lacks of context to be stupid by itself. Once in the context it is even less so I certainly won't 'redeem myself' but thank's for the good laugh anyway.
Look, here is the idea:
Before Godel played with the Diophantine equations version of "this statment is False", mathematicians didn't even think about such a thing as completeness/incompleteness of an axiom system.
Hilbert worked hard to formalize Mathematics so that people wouldn't argue anymore about proofs. Instead, they would just "compute" a proof with a mechanical mean of applying iference rules on the axiom set. Proofs would be computable statements leading to "true" or "false".
Godel's proof put an end to the beautifl dream stating that not all truths will be computable, unless some facts would be both true and false.
What does that mean ?
It means even if we carefuly chose our axiom set, that would form our "knowledge base", some truths couldn't be deduced from it, that is, they would be true just because they are. like Fermat Last theorem for example. Ok, it was proved, but it could have not been provable, so maybe we would have to add it to the axioms set.
This makes the axiom sets infinitelay growable, that I was translating as a truth that cannot be deduced from the one we already know is a knowledge we *yet* have to gain.
It follows 'naturally' that we can't have all the knowledge, there is always something else that is true, but that is not a natural consequences of the truths we already possess, making 'whole knowledge' impossible to have at any time.
Thus the 'natural' conclusion that Godel's proof indeed limits us not to know everything.
Have a nice day.
*LOL* :) :)
This isn't correct however, one of my last physics teacher was also a guitar player thus would have been the music teacher
also one of my last mathematics teacher was a very impressive Harley guy. if you've seen "eaters of the dead", the viking king that dies as a hero in the end is exactly like him
Er. The proof is done using a discrete system. That seems to pretty concretely limit its result to discrete systems.
Sorry, I wasn't clear. I mean the result isn't limited to this particular system for those two reasons:
- It has direct consequences on logic reasoning, because demonstration are such a system. this is the most important thing there, and the one I was defending later.
- Also, it can be considered as a specific counter-example to a more general set of problems. This is a specific example to show us the thing doesn't hold, so yes it is limited to exactly where it shows how it doesn't hold, but it is not limited in the way it doesn't also tell you this is the only place where this happens. that's what I was meaning.
So you know some truthes cant be stated and here is an example. even if you don't know how bad the situation is, you can feel it's going to be even worse with more complex systems.
Take real numbers for example, there are infinitely many more real numbers that are not rational numbers, and then most of them are even transcendent, even though we only pointed a few of them.
So maybe "bad truthes" are like transcendent numbers. maybe most of truths aren't statable.Godel just proved mathematics to be incomplete "because of this" but its proof doesn't tell us "it is just incomplete here, and only here".
And it indeed is worse than that, Turing stopping problem widens the panorama of troubles, and Chaitnin "unknowable" widens it further more.
Yeah yeah, and here we call particle physicists "parti-culeros" :) :) :) :)
But it's just a game. All my mathematics physics and chemistry teachers were always bitching at each other's classes or "way of seeing things".
"Ok, so here is a good equation that no mathematician can resolve, and we'll do that in just 1 second"
"So the theorem is proved because 3 is not 0, unless you teach chemistry of course, then anything under ten is not enough not to be considered 0" blabla
And they of course were all very good friends
Hmm correct me if wrong but I somewhat disagree with some of your points:
:) or maybe it's better like that after all :)
:)
1) Godel's proof only works in discrete systems that support (at least) a small number of operations.
Godel's proof is done using a discrete system, but there is nothing limiting its result there. it only lacks of substance to let us now how much we are in trouble with even more complicated problems, but it is enough to let us know that at least, beacause of those damn diophantine equations, our beloved set of basic axioms cannot exist. This pretty much kills the mathematicians "theory of everything"
Godel's proof does not say that it is impossible to know everything. It says that in these discrete systems, it is either a) impossible to make some valid statements (an incomplete system), or b) possible to make some invalid statements (an incorrect system).
Yes and no, it doesn't *say* you cannot know everything, but it is indeed a natural conclusion. those valid statements that are impossble to make are in fact logic statements as proofs. So some truths cannot be proven whatever axiom set you chosed (unless as you say you axiom set in inconsistant, which is even worse than incomplete).So said truthes have to be added to the axiom list, growing it to infinity.
If we cannot have a finite set of axioms to describe everything, than we cannot *know* everything. we can know everything about a few things that are *covered* by a finite set of axioms, but not *everything* too bad
3) Godel's proof only works if you are using boolean logic (and, in fact, works only because boolean logic is so bad at handling self-referential statements)
I still have a problem with this "only" it's correct of course, but it applies exactly on logic thus on our way to solve problems thus on our ability to cover *all problems* so it is indeed an important fact.
This does not mean that the universe works the same way.
I fully agree, Godel's proof and its consequences don't limit our possibility to define how universe works with just a few (only one ?) set of laws. However, once we find the basic laws if we ever do so (which I doubdt) we might still observe some phenomenons that are not possible to explain with just the fundamental laws, creating "emerging laws" that would in fact be more than just a statistical consequence of more basic interactions.
You are however right in the general idea of your post that Godel definitely did not prove that was going necesarily to happen, he just proved it is a possibility.
At least that's how I understood it, but I can be wrong, that's for sure
The biggest mistake RMS did was probably to call his 'vision' "Free Software" :) :-)
People get extreme around this word.
First, "freedom" as "absolute freedom" is a nonsense and cannot exist. it's unfortunately not so clear a problem as "This statement is False" so people don't realize it and still want "Freedom" to be absolute as in "everything I wever want". So they get in big ideological troubles, all of them rooted in the fact that they both understand Freedom as something different and generally too large, yet to strict to admit anyonelse's "freedom" (freedom definition, not freedom itself, although freedom itself sometimes).
Second, freedom has this great "added value" that everybody cares about it and would defend its own (definition) with so much conviction yet so little skill it's the mess we're witnessing every so often in slashdot.
Look at your own words, for instance: "EX. I may not like Microsoft bashing Linux, but I will defend their right to do so."
So far so good.
"Stallman needs to recognize this and embrace it."
Uh ?
Big dilema here, how would you defend someone's freedom to fight your freedom of defending his freedom ?
Not what's happening here, but it's the same problem
In the end, I think you should stand on this "I'll fight for your rights to..." as it's clearer and easier. I.E. you know what you are fighting for. (of course you'll have to fight for someone's rights to advocate censorship, but it's still simpler)
In the case of Stallman's "Free Software", you'll have to remember he's standing for something he started himself.
Also, it's not like the guy is an idiot, I'm pretty sure he actualy understands what he wants. Why does everybody undersntand what he wants better than he does himself ?
Maybe you want something else, maybe you should discuss with him how you'd like his vision to be called something else than "Free Software" so that you can use it for your own "vision" (I garantee you you'll end in the exact same situation, simply because of the "Free" word).
I doubdt he'd like it though, but it would be nice. Of course, Free Software as it is in RMS's words must just fit his own freedom understanding so that he probably wouldn't even see the need of another, more 'accurate' description.
So how would something for "free" be more acurate than "free" uh ?
Wether it is due to a strobing light source or a "frames capturing device" (television, cinema..) the idea is the same...
...
Look at your wheel , let's say it spins at (24*3)-2 revolutions per second.
If you capture 24 images per second, then you will see this sequence of images:
- wheel at position "0"
- wheel at position "-2"
- wheel at position "-4"
Of course the wheel actualy spinned nearly 3 times, but you just get the picture of it when it is slightly rotated in the other direction.
So you effectively watch it spinning backward.
This way, you can "freeze", play backward, play slowly etc.. any periodic/repetitive motion, just adjust your sampling frequency around the motion frequency. if it matches, you froze it.