Yeah, ok, you're right. It was more the frame of mind than the actual capability that I was paraphrasing (I did say that the NT core was capable of this). Windows didn't ship with a telnet server until after the demise of telnet, and generally the system is set up in such a way that it's a far cry from a multi-user environment. But indeed, technically you are correct, even though you need a very expensive edition to do something that practically any other OS does out of the box (multi-user/machine gui/app access).
I think you misunderstand what most people mean with multi-user. In computing land this means that the operating system supports multiple users doing stuff on the machine at the same time, not that you have different logins/passwords for an essentially single-user environment. Although the NT kernel indeed has true multi-user support at its core(*), you need to get the 'Terminal Server' edition of the OS, not the 'Home', 'Professional', or even the 'Server' editions. These are crippled to single user systems. IIRC, the TS was introduced with w2k, not before.
(*) Citrix made use of this by offering a true multi-user windows before Microsoft did.
On point 4. It's spot on, not bullshit. I gather you're a window user, but in Unix land you never ever run the GUI as root. Never. What you do is log in as a normal user, browse the internet as a normal user and when you located whatever it is you need to do as root, you go to a console, su and do the root thing there. Why? This makes sure that if you as user catch something on the big bad internet, it doesn't hose your entire system right away. If you run this piece of shit IE as Administrator, any flaw in IE can take over your system, when run as user, it can only take over with user priviliges and might give you time to take countermeasures.
The naming convention used is quite logical if you bother to look further.
Ok, looking a bit further, it looks like it should have been called a ArrayListSerializableClonableRandomAccess, but that might be too much even for Java. Given the naming of HashSet and HashMap, I think we could settle for RandomAccessList;-)
And let's not mention the fact that 'Vector' is a very inaccurate name for an 'Array'. Vector comes from C++ (STL), and was named thus and not 'array' simply because it was found that many legacy C and C++ programs already used the name 'array' all over the place to describe *gasp* arrays. It would have been a pain to remove all these. So now with C++ we're in the situation where std::vector denotes an array, while std::valarray denotes a vector (you know, that mathematical object that contains a few primitive types and needs to be manipulated as a whole.).
The original Java designers apparently didn't stop to think why the this was called 'vector' in the first place and just put it in the spec, this while they didn't have legacy code to contend with. Once they found their 'Vector' was too slow for its most common use they invented something weirdly called an 'ArrayList' (which looks more like the std::vector). Now wow, that's a weird name: is it an array or a list??? They probably refrained from using Array because at that point they needed to conted with legacy code:-) Ah well, I think that nobody would argue that Java at its core is very consistent anyway (String.size, Collection.size(), T[].length?).
... is an argument against the existence of higher being, and as such is not falsifiable.
Not true, theories denying the existence of a higher being are perfectly falsifiable. "There is no God" can be falsified by showing God (maybe in a picture of her or something irrefutable like that). "There is a God", however, cannot be falsified and is therefore not a valid scientific theory.
I don't know of specific advances in mathematics, but if you shave off a tiny bit of the power of the general O(N^3) inversion for a special case of practical importance, it's a simple calculation to determine the size of the matrix to make the calculation a thousand times faster, a hundred thousand times faster, etc. Algorithms work that way, unlike computers. Furthermore, Moore's law only talks about the number of transistors which is a special case in building a general-purpose machine. CPU speed, memory/cache/harddisk access also count. Each of these have increased but not all in line with Moore's law.
So the clock of your microwave oven can only count for 16 seconds? I sure hope you've got some decent power in the thing because a decent meal would take quite a bit of energy to heat up in that time.
Re:If You Want a Serious Answer... Don't Get Cute
on
Rob Pike Responds
·
· Score: 1
MAD being pure evil is a judgement call as well. As far as I can see it kept the world 'dangerously safe' for a couple of decades by having a cold war instead of an all-out third world war; a thing that could easily have happened without MAD.
Usually the speed of progress is measured by the amount of papers that are published in journals. A few guys at the Physical review letters at one point extrapolated the trend from the last 30 years and obtained the prediction that with current progress in science, in 2030 the speed with which shelf-space would be filled with the journal pages would exceed the speed of light. However, they could safely concluded that this wasn't a violation of general relativity as no actual information is transmitted in these pages.
Afganistan today: Kabul in the hands of the elected government, outside of Kabul is in the hands of a few warlords and the Taliban again. The Taliban haven't gone, they've hidden themselves in the border area with Pakistan (which they 0wn) and are now moving back in to the place they occupied when he Soviet Union was in control (remember that one?). Western troops are probably stuck there for a few decades because of the explosiveness of the situation. Organizations for help have left the country because it's too dangerous.
Yes: Kabul is a much better place now than it was with the Taliban in charge, the rest of the country is still pretty much fucked up and no real change in sight (yet, I hope).
Are you suggesting that the government stop trying to aprehend and prosecute hard drug users/dealers?
Will not happen in this lifetime, but of course, yes! Illegal hard drugs are not worse than the allowed hard drug alcohol, and less addictive/dangerous than the enormous amount of legal drugs the pharmaceutical industry produces (have you any idea how many people are hooked on painkillers, and how devastating that is for one's health?). Prosecuting users is absolutely silly, next thing you know the gov will put the death penalty on suicide. What about my right to fuck up my life the way I please?
I've posted this in another thread as well, but I'm interested in comments on the following strategy:
Maybe we should approach the battle against software patents the other way around. Instead of looking for ways to protect against patents and/or get patents for OSS, why not play the game out to its ultimate consequences. Start up a website and start matching patents with software products: when a likely match is found, notify the patent holder, the software vendor and possibly the software users of this potential patent violation. For the large fish (IBM, Microsoft, etc.) focus on the patents of patent trolls: litigation companies existing merely of a patent portfolio. If lots of companies are targetted, this would create a lot of publicity and when there's a website to back up the claims, there's a lot of proof about the untenability of the situation. Furthermore: it only takes work, not deep pockets to do this.
If the situation is as bad as is claimed, this would mean that any software vendor and/or webshop is forced to concede the fact that they can be put out of business overnight due to patents. Once this awareness is there it would just be a matter of time before the case law is overturned. As an example see this site. Imagine it targetted at real software products and real webshops. Probably we can match a couple of hundred patents to any particular product/shop, showing the seriousness of the situation, while at the same time confronting the potential victims with it. I know, this is a real weird proposal, but the question is: would it do the trick?
Maybe we should approach this the other way around. Instead of looking for ways to protect against patents and/or get patents for OSS, why not play the game to its ultimate consequence. Start up a website and start matching patents with software: when a likely match is found, notify the patent holder, the software vendor and possibly the software users of this potential patent violation. For the large fish (IBM, Microsoft, etc.) focus on patent troll's: litigation companies existing merely of a patent portfolio. If lots of companies are targetted, this would create a lot of publicity and when there's a website to back up the claims, there's a lot of proof about the untenability of the situation. Furthermore: it only takes work, not deep pockets to do this.
If the situation is as bad as is claimed, this would mean that any software vendor and/or webshop is forced to concede the fact that they can be put out of business overnight due to patents. Once this awareness is there it would just be a matter of time before the case law is overturned.
Sure we can build a virtual quantum computer, how do you think the quantum algorithms have been validated? The problem is that a virtual quantum computer loses its parallel properties, so that to run an 8-bit quantum algorithm on non-quantum hardware you need 2^8 calculations per processor step. This is *slightly* prohibitive for --say-- emulating a 32-bit quantum computer.
Such a more powerful search engine is already in place. (Note the exact title of today + restricting the search to slashdot brings back the original story).
One of the founders of Xs4All and the founder of HackTic is Rop Gonggrijp (now famous on slashdot for lending his car out in the terrible car accident a few week back). Xs4all is also the ISP that refused to take down Karin Spaink's website with Scientology papers on it, and went to court over it (which they won). They have a pretty extensive privacy statement for their users, and I do believe they abide that. All in all, this is one of the few ISP's left where the extra euros you spend actually amount to significant value.
Just as a bit of context, I'm a researcher in machine learning that's extremely jealous of the amounts of funding my collegues just received for their good-all-fashioned inference technology on the semantic web (They did a great job getting the money, but I think they're on the wrong path).
But okay. I don't think we're in disagreement here: I totally agree with you that the "low level tools" are the ones that are of interest here, I'm just sceptical about the scalability of the grand objective: crisp inference, based on this information. But, once the groundwork is done, machine learning will probably be the technology that will be used. Simple, fuzzy inference of the kind you describe and which made google big. Combine these 'simple' queries (find me cheap milk!) with likeliness of a match (is GallonOfMilk talking about dairy?), thrustworthiness of sites (Is the rest of the content about milk or milk-like products or products at all? Does the content match other milkselling sites?), and you've got something useful.
Thanks for the explanation, your story makes a bit more sense than the article(s) I read about ontologies, distributed ontologies, merging distributed ontologies and research into merging mutually contradicting distributed ontologies, which seems to be linked together with a brittle inference process that will fall flat on its face when inconsistent information is entered. You guessed, I remain sceptical.
About the web, it's perfectly capable of handling inconsistent information, as all the information it handles is valid HTML and links that resolve. The rest is simply data, strings of bits. It simply enforces consistency in its narrow domain of valid information. Not so with the Semantic Web.
I don't have a problem with rdf, as it's just a way to state stuff, and it seems genuinely useful. I do however think that the use of inference engines and formal proofs is a conceptual dead-end in a dynamic and inconsistent world. In contrast with your statement that all there is to the Semantic Web is rdf, the Semantic Web research is in the inference based upon these facts, not in the statements themselves (other then their (in)consistency). This is flawed for the same reasons expert systems never worked. An inference engine that cannot handle inconsistencies, cannot learn, and cannot revise its results in the presence of new information is only very limitedly useful.
I am however sure that some good use of the rdf information can be made using statistical (Bayesian) inference, which can handle inconsistency, revision and extra evidence gracefully, so maybe regardless of the doomed research that is going on, we will end up with a semantic web after all.
Any system that is used in the real world and cannot at its core handle inconsistent information (and no deductive system can) is fundamentally flawed for this use. Expect from the semantic web the same as for automatic translation (60s) and expert systems (80s).
This gets one page. If someone could reverse-engineer the "sig" argument I'm sure you could specify a page number.
No need to reverse engineer, just enter the search word 'the' in the search box on the page and google will gladly give you a list of all pages in the book (with the page number the header of the search result). From there you can get to every page. That is, until google gives a friendly warning that it will not serve more pages (because the limit has been reached), after which it ever so gently will serve more pages;-).
Slashdot Mirror
Yeah, ok, you're right. It was more the frame of mind than the actual capability that I was paraphrasing (I did say that the NT core was capable of this). Windows didn't ship with a telnet server until after the demise of telnet, and generally the system is set up in such a way that it's a far cry from a multi-user environment. But indeed, technically you are correct, even though you need a very expensive edition to do something that practically any other OS does out of the box (multi-user/machine gui/app access).
(*) Citrix made use of this by offering a true multi-user windows before Microsoft did.
On point 4. It's spot on, not bullshit. I gather you're a window user, but in Unix land you never ever run the GUI as root. Never. What you do is log in as a normal user, browse the internet as a normal user and when you located whatever it is you need to do as root, you go to a console, su and do the root thing there. Why? This makes sure that if you as user catch something on the big bad internet, it doesn't hose your entire system right away. If you run this piece of shit IE as Administrator, any flaw in IE can take over your system, when run as user, it can only take over with user priviliges and might give you time to take countermeasures.
Ok, looking a bit further, it looks like it should have been called a ArrayListSerializableClonableRandomAccess, but that might be too much even for Java. Given the naming of HashSet and HashMap, I think we could settle for RandomAccessList ;-)
The original Java designers apparently didn't stop to think why the this was called 'vector' in the first place and just put it in the spec, this while they didn't have legacy code to contend with. Once they found their 'Vector' was too slow for its most common use they invented something weirdly called an 'ArrayList' (which looks more like the std::vector). Now wow, that's a weird name: is it an array or a list??? They probably refrained from using Array because at that point they needed to conted with legacy code :-) Ah well, I think that nobody would argue that Java at its core is very consistent anyway (String.size, Collection.size(), T[].length?).
Not true, theories denying the existence of a higher being are perfectly falsifiable. "There is no God" can be falsified by showing God (maybe in a picture of her or something irrefutable like that). "There is a God", however, cannot be falsified and is therefore not a valid scientific theory.
I don't know of specific advances in mathematics, but if you shave off a tiny bit of the power of the general O(N^3) inversion for a special case of practical importance, it's a simple calculation to determine the size of the matrix to make the calculation a thousand times faster, a hundred thousand times faster, etc. Algorithms work that way, unlike computers. Furthermore, Moore's law only talks about the number of transistors which is a special case in building a general-purpose machine. CPU speed, memory/cache/harddisk access also count. Each of these have increased but not all in line with Moore's law.
I think tampering with the election should be a felony anyway, so let's indeed restrict Florida from voting for life.
Vietnam?
So the clock of your microwave oven can only count for 16 seconds? I sure hope you've got some decent power in the thing because a decent meal would take quite a bit of energy to heat up in that time.
MAD being pure evil is a judgement call as well. As far as I can see it kept the world 'dangerously safe' for a couple of decades by having a cold war instead of an all-out third world war; a thing that could easily have happened without MAD.
And then again, so what? Let them collect the data about aisle traversal time and optimize the hell out of their display. What's the beef with that?
Usually the speed of progress is measured by the amount of papers that are published in journals. A few guys at the Physical review letters at one point extrapolated the trend from the last 30 years and obtained the prediction that with current progress in science, in 2030 the speed with which shelf-space would be filled with the journal pages would exceed the speed of light. However, they could safely concluded that this wasn't a violation of general relativity as no actual information is transmitted in these pages.
Yes: Kabul is a much better place now than it was with the Taliban in charge, the rest of the country is still pretty much fucked up and no real change in sight (yet, I hope).
Will not happen in this lifetime, but of course, yes! Illegal hard drugs are not worse than the allowed hard drug alcohol, and less addictive/dangerous than the enormous amount of legal drugs the pharmaceutical industry produces (have you any idea how many people are hooked on painkillers, and how devastating that is for one's health?). Prosecuting users is absolutely silly, next thing you know the gov will put the death penalty on suicide. What about my right to fuck up my life the way I please?
Maybe we should approach the battle against software patents the other way around. Instead of looking for ways to protect against patents and/or get patents for OSS, why not play the game out to its ultimate consequences. Start up a website and start matching patents with software products: when a likely match is found, notify the patent holder, the software vendor and possibly the software users of this potential patent violation. For the large fish (IBM, Microsoft, etc.) focus on the patents of patent trolls: litigation companies existing merely of a patent portfolio. If lots of companies are targetted, this would create a lot of publicity and when there's a website to back up the claims, there's a lot of proof about the untenability of the situation. Furthermore: it only takes work, not deep pockets to do this.
If the situation is as bad as is claimed, this would mean that any software vendor and/or webshop is forced to concede the fact that they can be put out of business overnight due to patents. Once this awareness is there it would just be a matter of time before the case law is overturned. As an example see this site. Imagine it targetted at real software products and real webshops. Probably we can match a couple of hundred patents to any particular product/shop, showing the seriousness of the situation, while at the same time confronting the potential victims with it. I know, this is a real weird proposal, but the question is: would it do the trick?
If the situation is as bad as is claimed, this would mean that any software vendor and/or webshop is forced to concede the fact that they can be put out of business overnight due to patents. Once this awareness is there it would just be a matter of time before the case law is overturned.
Sure we can build a virtual quantum computer, how do you think the quantum algorithms have been validated? The problem is that a virtual quantum computer loses its parallel properties, so that to run an 8-bit quantum algorithm on non-quantum hardware you need 2^8 calculations per processor step. This is *slightly* prohibitive for --say-- emulating a 32-bit quantum computer.
Such a more powerful search engine is already in place. (Note the exact title of today + restricting the search to slashdot brings back the original story).
One of the founders of Xs4All and the founder of HackTic is Rop Gonggrijp (now famous on slashdot for lending his car out in the terrible car accident a few week back). Xs4all is also the ISP that refused to take down Karin Spaink's website with Scientology papers on it, and went to court over it (which they won). They have a pretty extensive privacy statement for their users, and I do believe they abide that. All in all, this is one of the few ISP's left where the extra euros you spend actually amount to significant value.
But okay. I don't think we're in disagreement here: I totally agree with you that the "low level tools" are the ones that are of interest here, I'm just sceptical about the scalability of the grand objective: crisp inference, based on this information. But, once the groundwork is done, machine learning will probably be the technology that will be used. Simple, fuzzy inference of the kind you describe and which made google big. Combine these 'simple' queries (find me cheap milk!) with likeliness of a match (is GallonOfMilk talking about dairy?), thrustworthiness of sites (Is the rest of the content about milk or milk-like products or products at all? Does the content match other milkselling sites?), and you've got something useful.
About the web, it's perfectly capable of handling inconsistent information, as all the information it handles is valid HTML and links that resolve. The rest is simply data, strings of bits. It simply enforces consistency in its narrow domain of valid information. Not so with the Semantic Web.
I don't have a problem with rdf, as it's just a way to state stuff, and it seems genuinely useful. I do however think that the use of inference engines and formal proofs is a conceptual dead-end in a dynamic and inconsistent world. In contrast with your statement that all there is to the Semantic Web is rdf, the Semantic Web research is in the inference based upon these facts, not in the statements themselves (other then their (in)consistency). This is flawed for the same reasons expert systems never worked. An inference engine that cannot handle inconsistencies, cannot learn, and cannot revise its results in the presence of new information is only very limitedly useful.
I am however sure that some good use of the rdf information can be made using statistical (Bayesian) inference, which can handle inconsistency, revision and extra evidence gracefully, so maybe regardless of the doomed research that is going on, we will end up with a semantic web after all.
Any system that is used in the real world and cannot at its core handle inconsistent information (and no deductive system can) is fundamentally flawed for this use. Expect from the semantic web the same as for automatic translation (60s) and expert systems (80s).
No need to reverse engineer, just enter the search word 'the' in the search box on the page and google will gladly give you a list of all pages in the book (with the page number the header of the search result). From there you can get to every page. That is, until google gives a friendly warning that it will not serve more pages (because the limit has been reached), after which it ever so gently will serve more pages ;-).
Why are you using a browser that can be broken that easily?