Your view is a bit simplistic, especially given the global nature of the world's economies and geo-politics. At this point, it would be virtually impossible for the US to cease all foreign operations
I was careful to say "preferring non-involvement", not "avoiding action at all cost".
I'm sure I don't need to remind you, but the US nearly doomed England and the rest of Europe by a lack of military support in WW2 until Japan brought the war home.
Let's stop for a moment and think about what would have happened if the US hadn't gotten involved in WWII. So, Germany takes over Britain and a few more European countries. Do you really think that that ridiculous government with its ridiculous ideology and murderous tendencies would have lasted very long? Do you think the British or French would have put up with German occupation for very long? Do you think a country that classifies science it doesn't like as "degenerate" and had just killed or exiled most of its elite would manage to keep up technologically? The "Reich" would have fallen apart after a few years, probably with fewer deaths than WWII actually caused.
Of course, in hindsight, US involvement in WWII probably was overall positive. It came too late to save the Jews, but at least, it propelled Europe rapidly into its current liberal and democratic existence, something that would probably have taken decades longer if the Germans had (temporarily) prevailed. But it is dangerous to paint pictures of "doom" because then you suddenly see just about any action, however disproportionate, as justifiable. Which is just what we are seeing today in US politics.
For instance, if Isreal hadn't destroyed the nuclear power plant Iraq was building, it's certain that Iraq would now be a nuclear power
Yeah, big deal. What would Iraq be doing with it? Any place they could use it that we care about would risk swift and terrible retaliation, and Iraq could be doing a lot more damage with biological weapons.
Israel probably did the right thing anyway, but that's because the destruction of the reactor was a small, tactical effort (and even that was at least officially condemned by the US and the UN). Starting a war over an issue like that would have been something entirely different.
What made the British Empire unusual is that it was one of the few empires that realized it had a responsibility for the people it placed in the empire (cases can be made for the Romans and Chinese there as well).
Most colonial powers felt that they were bringing the light of civilization to "the savages". That's why they brought religion, education, and administration with them. But what you consider "taking responsibility" may just as well be viewed as the extermination of entire cultures.
Consider, just as one terrible example, Africa before, during, and after colonization. Both before an after, terrible tribal wars, massacres, poverty, and raw, naked power terrifying the weak. During colonization? In some places the same or worse (Congo, the Boer States). In other times and places much, much better (much of British ruled Africa).
I think your analysis already presupposes a Western point of view. For example, people aren't poor until they come to accept Western ideas of wealth.
My point is not that arrogance or bullying is somehow good. It is that I don't think the Europeans have so much improved morally as become throughly cynical and insular.
My point was that I don't think the British needed the Nazis to teach them about totalitarianism or mass murder--the British had figured out and practiced those things for themselves on and off, as had most Western nations.
As for what Europeans are doing today, I think their attitude of preferring non-involvement is the better one. It is the Americans that now erroneously believe that they can carry the torch of freedom and democracy around the world and do good everywhere. It won't work. America's foreign policy has caused, and will continue to cause, enormous suffering. Sooner or later, the Europeans and the UN will likely put pressure on the US to stop its foreign adventures and clean up its domestic problems.
But the option not to conquer the Empire did not exist. If Britain had not done so, one of the other European powers (France, Spain, Portugal, Holland) would have done and used the resources to conquer Britain.
That's all well and good, but why do you think Germany started WWI and WWII? Did Britain have any more justification for wanting to be powerful, independent, and autonomous in the 19th century than Germany did in the 20th century? Did British aspirations justify the death of millions of natives abroad and the destruction of entire cultures any more than Germany's aspirations (failed to) justify an aggressive war in Europe and the destruction of millions of Jews and other "undesirables"?
All the major European nations were evil empires until the 19th and 20th century. Britain didn't need Nazi Germany to teach it about expansionism, racism, genocide, or totalitarianism, the British knew about those things well enough all by themselves and had practiced them, on and off, for centuries. It was only WWII that brought that to an end and finally caused European nations to come to terms with their histories.
Realizing that evil empire building was an all-European activity is important, not in order to attempt to exculpate Germany, but to understand that Britain isn't immune from these kinds of tendencies. British democracy might go bad in just the same ways that the Weimarer republic self-destructed. Fortunately, European integration and diversity has made this fairly unlikely these days.
Perhaps that realization is even more important for Americans, who have a powerful and increasingly centralized government. Americans seem to think that they don't have to worry about any of those pesky privacy and human rights issues because "The Constitution" will protect them like a magic shield, and that they are justified in all their actions because, almost by definition, anything America does is in the name of freedom and democracy.
I think we should really stop being so obsessed with attributing inventions to individual people. Morse's combination of single wire and serial code was clearly very practical and made the telegraph successful. But if it hadn't been Morse, someone else would have done the same thing within a few years: all the general ideas had been around. On the other hand, while the insight that electricity can be used for long distance signaling is great, it in itself does not lead to a viable and practical telegraph system.
The same is true for most of the "great" inventions or ideas we celebrate. It is very rare indeed that a ground breaking new idea appears out of the mainstream, and when it does, it usually doesn't catch on until the mainstream catches up with it and someone else gets the credit.
Yes, damn those evil Brits, rampaging all over the world building roads, schools, hospitals, dams, [...] creating legal systems
Exactly my point: going out all over the world, destroying one civilization after another, subjugating the native populations, and plundering its natural resources.
If only Stalin had conquered the world instead!
The world wasn't for anybody to "conquer": not for the British, not for the Spanish, not for the French. The fact that Stalin or the Nazis created evil regimes does not diminish the profound evil of those other empires. It's only in the second half of the 20th century that Europeans finally came to their senses.
you'd think we'd lost the war and the UK was a Nazi fucking state.
And what makes you think the UK wasn't? What does winning WWII have to do with the UK political system? Might makes right? The UK was committing atrocities and genocide for centuries before the Nazis even appeared on the scene--the UK just did it out of sight, overseas, to people who even today largely don't have much political say. The British Empire was one of the most evil institutions in human history.
Today, the UK has reformed, just like the rest of Europe, although it seems still quite a bit more right leaning.
How does removing a biometric identifier help you "get around a problem"? Right now, say, if your credit is bad and people refuse to give you a loan, pretending to be someone else is merely adding fraud to your list of problems. The fact that lack of sound biometric identifiers makes committing fraud harder is, in my opinion, a plus.
Yes, and if you have need of a monarch, you are in trouble. But because it's a fact of modern life that people have always need of their government, we have made government by the people and for the people. And you "always and in every possible condition of things have need" of your fellow citizens and society: without them, your food doesn't reach you, your water doesn't reach you, your electricity doesn't reach you, your roads don't get maintained, you don't get medical service, etc. Welcome to the real world.
And what freedom would that be in this case? The freedom to pretend to the government that you are someone else? The freedom to commit identity theft?
The government should be tightly regulated in its use of any personal information, something both the US and the UK are shamefully poor at. But fighting the introduction of well-designed national ID cards accomplishes nothing--it just means that the UK (and US) governments will keep mixing your files up with someone else's in addition to violating your privacy constantly.
The Home Secretary himeslf had his identity stolen [bbc.co.uk] by a journalist to highlight the dangers of identity theft, which will without a doubt rise if these new cards are introduced.
Current ID cards have almost no protection against identity theft. You see, even in the US we have national ID cards, they just don't work very well. New identity cards are an attempt to improve the situation.
If you think they aren't going to succeed, then you have to say why. But your blanket statement is simply logically flawed.
Really, if Europeans want to have ID cards, no one in the UK has a problem with that, and no one here is interested in arguing with Europeans who think that ID cards are "no problem at all".
The people of the UK are Europeans--it's a simple geographic and political fact.
Sure, some MS executives have said something like "we have lots of intellectual property in this great new technology", but every company that wants investors to drive up the stock price says that. Nobody has yet been able to produce any concrete, substantial intellectual property that Microsoft (or anybody else) holds in C# or.NET. The few patents that have surfaced have been peripheral and easy to work around. So, if you have concrete knowledge of this "legal minefield", please tell us all. Otherwise, please don't keep spreading FUD about Mono.
First of all, there is almost nothing technologically new in Windows NT: its "innovations" in areas like file systems, access control, security, etc., have pretty much all been tried before. People don't use them by choice. Furthermore, IBM has created a pretty complete Windows clone as part of OS/2, and they didn't get into legal trouble, so there is little reason to believe that ReactOS would have any more problems.
While there are quite a few IT managers who either don't know any better or have painted themselves into a corner when it comes to servers, and therefore need the biggest single processor bang for any buck, most people who really care about CPU performance care about bang-for-the-buck.
Unfortunately, none of the current crop of 64 bit processors deliver: the cost of true 64 bit systems (those capable of actually using more than 4 Gbytes of memory) generally starts somewhere upwards of $10000, and for that you do not get anywhere near 10 times the performance of a $1000 PC.
The main reason right now to get a 64 bit system at current prices is because the applications just cannot be shoehorned into a mere 2-4 Gbytes. If AMD can change that equation and deliver comparable bang-for-the-buck to current PCs, with 64 bit addressing being icing on the cake, they have a winner. None of the other players seem to be capable of doing that--they have tried and failed miserably so far.
You obviously dont understand what KDE brings, therefore it is not surprising you deem it unworthy.
Oh, I understand exactly what it brings: tons of really useful functionality, but I can only get most of it if I subscribe to the whole KDE master plan.
Now, Quanta COULD write such a higher level API, but wouldnt that be just recreating kioslaves, only for each app? That makes no sense.
Indeed, it wouldn't. But it also doesn't make sense to invest lots of time in something like kioslaves if it is only going to be useful for building KDE apps. What would make sense would be to create a library like kioslaves that would be used with lots of toolkits: wxWindows, FLTK, Gtk+, etc.
The KDE (and, to much the same degree, Gnome) attitude is: f*ck the rest of the universe, we are just going to live in our own little world and tinker with out intricate web of libraries. And then we are going to take over the world because we are a little better than everybody else. That's Windows thinking. That's mainframe thinking. It sucks. It's exactly what UNIX was created in reaction against.
I am not aware of a set of libraries that would support nearly as many protocols as the kioslaves with the same API.
Come on, pay at least a little attention. I'm not disputing the utility of the individual components that KDE provides. Quite to the contrary: I think the functionality that KDE components provide is extremely useful. All the worse that KDE is taking the approach of creating an integrated, consistent, interdependent set of libraries and software components that reuse each other. While those adjectives doubtlessly sound good to you, what they mean is that nobody is using kioslaves with other environments.
Lots of systems have taken KDE's all-or-nothing design approach over the last 30 years. It's great in the short run: it's really easy to do, and everything works together oh-so-nicely. But most of the time, the "all-or-nothing" turns into "nothing", and then no software components survive.
Pray, Mr. Troll, where would I find those miraculous libraries?
Circle the wagons! KDE is under attack! Retaliate with insults! Just don't bother trying to understand what someone is saying. Great going. And so typical.
The shuttle fleet should have been grounded long ago: it's a costly and inefficient way of moving stuff into orbit. And manned space exploration should have been put on hold decades ago as well--it, too, is costly, inefficient, dangerous, and unnecessary.
Hopefully, as a consequence of this strategy, manned flights will be scrapped for the foreseeable future and we can concetrate our efforts on unmanned probes.
Of course, what you say would make a modicum of sense if KDE (or GNOME) had the goal of "put[ting] in place a gigantic, sluggish infrastructure and try to force everything to use the same libraries". Since they don't, your comment is basically handwaving.
But they do: they have their own standards and software infrastructure for inter-client communication, audio, and a lot of other features.
For example: Suppose Quanta was not a KDE application. Now imagine a webmaster wants to use Quanta to edit pages on a website.
Since that is a very necessary feature, Quanta would have to implement some sort of ftp client. And perhaps also a scp/sftp client, a webdav client, and so on for every mechanism it wanted to support.
But... luckily Quanta *is* a KDE app. So, it got all that for free. And if tomorrow someone writes a mechanism to access any other remote site, Quanta will get it too.
There is nothing wrong with reusing software. But we already have libraries for dealing with FTP, SSH, WebDav, and all those other protocols.
The errors of KDE (and Gnome, for that matter) are that they are reinventing the wheel and picking a single winner. Rather than for the best FTP library to win through the choice of application developers, KDE's philosophy is that you use the KDE library. And if my application doesn't use the KDE library, then it won't integrate well with all the other KDE apps that do. And if you use one KDE library, probably you have to use more and more, and KDE servers, and lots of other stuff, since it's all interlinked and interdependent.
KDE commits the same error as the Soviets: the problem of designing useful parts for a free market of ideas and components is so daunting that they instead fall back on central planning and central specifications. And the result is just like the Soviet Union as well: initially, a slick and consistent machinery, but it is already getting stale and it will sooner or later collapse under its own weight.
I'm tired of this "all consistent, all integrated" desktop madness. We have Mozilla for web browsing, e-mail, and calendaring. We have OpenOffice for MS Office-like uses. Maybe we need something that works a little more like Outlook than Mozilla does right now. Pick a good window manager, some small utilities, and you have a decent desktop.
I think efforts like Gnome and KDE that try to put in place a gigantic, sluggish infrastructure and try to force everything to use the same libraries are largely a waste of time and effort. Not even Microsoft or Apple are that consistent.
The promise of really low-cost 64bit systems is what makes AMD interesting. Server chips traditionally have a poor cost/performance ratios; they are for premium priced systems that only cover a small fraction of the true server market. If Opteron systems end up being not much cheaper than Itanium systems, they aren't interesting at all as far as I'm concerned.
I bought a couple of Macs about a year ago when OS X seemed ready for prime time. They look great, and they work reasonably well. It's nice to be able to port some UNIX/Linux software to it fairly easily, and it's nice that some system administration skills carry over.
But Apple has made the system much more proprietary and non-standard than it needs to be. The system administration database is different from mainstream UNIX systems made integrating the Macs into my home and work networks much more work than a Linux machine. The window system is completely different from UNIX, hard to port to, and rather sluggish. Apple's software package management is worse than even that of Windows. And the commercial software situation on Macintosh is not all that great either. A big disappointment, too, was that Apple had promised "free.Mac service with every iMac" and then started charging less than a year later.
Altogether, I think Apple has benefitted quite a bit from UNIX/Linux compatibility, by promising a no-hassles Linux-like environment and attracting some UNIX and Linux users. I don't think they really have delivered, and I will probably not be upgrading my Macs--I can get better functionality and more software for less money with Linux. On the other hand, Linux has not benefitted directly from OS X: there is little or no useful software that Apple has donated to the Linux community (Darwin is more of a distraction), and I don't think Apple's "switch" campaign has been all that effective.
I think in the long run, Apple will be forced to become more and more Linux compatible, and then maybe there will be more benefit to the Linux community. Until then, every Windows user that moves to Macintosh is still of some benefit to the Linux community.
The main problem with NT isn't that it's commercial or that it sucks technically, the main problem is that it's through-and-through proprietary--it's a single-vendor solution.
The military could and should go with software that is based on open standards: UNIX/POSIX, X11, etc. And in their implementations and deployments, they should then stick as much as possible to those open standards. They can then buy software and hardware from many different vendors and have a choice among multiple implementations, including some open source ones.
CART (Classification and Regression Trees) by Breiman et al. is a good reference for tree-based classification and makes the connection between exploration of hypotheses and validation very directly. (CART would probably be a good method to apply to this data.)
I'd also recommend Hastie et al. "The elements of statistical learning". For general machine learning, Mitchell's book "Machine Learning" is also frequently recommended.
Books on support vector machines generally also talk about the tradeoffs more theoretically (VC dimension, etc.). The following book may be reasonably readable: "An Introduction to Support Vector Machines and Other Kernel-based Learning Methods" -- by Nello Cristianini (Author), John Shawe-Taylor (Author);
In 1996, there were two huge sites designed in a modern way--the Olympics and Wimbledon. Of course, they had navigation bars and dynamically generated content.
Pearson's tables are for any correlation between random variables however constructed so long as they don't depart too far from normal distribution.
Yes, and that's in effect what I was saying, but that's not relevant. You get to apply Pearson's tables only once to a single linear relationship computed on one dataset. If you keep constructing different combinations of variables and keep testing for normality and correlation, the significance of the test sooner or later erodes. Sometimes you get lucky and you get to try a lot of hypotheses before that happens (which is why people often get away with using it), and sometimes you don't.
Look at it this way: pick any distribution of 47 (u,v) values that satisfies your normality and correlation test. Now, I can construct a pair of polynomials that maps any 47 datapoints (x,y) you actually measured onto those 47 (u,v) values. If you had just kept trying different polynomials and fitting their coefficients, you'd eventually have hit upon this. So, you can't just keep trying different combinations of variables: sooner or later, you are going to find one that satisfies whatever statistical tests you apply.
What you have to avoid, but not all all costs, is mindless "data trolling" -- just fishing for any correlations between random variables regardless of your prior hypotheses. The costs of not doing such data trolling are that frequently one has hypotheses that can't be directly measured or that are still in formation and one is searching for means of refining the hypotheses or ways of testing. This is tricky and can't be reduced to simple bromides like "don't data mine" or "correlation doesn't imply causation", etc.,
Maybe it's tricky for people who only have "Pearson's r" in their bag of tricks. But there is nothing "tricky" about it statistically. The question of what hypotheses one can explore and what the significance of the results is is reasonably well understood. Machine learning techniques routinely explore billions of possible formulas and hypotheses describing the relationship between variables of some dataset. But they use accurate, theoretically justified measures of the significance of such hypotheses. Pearson's r doesn't work for that; that's not a theoretical debate or a matter of preference, it just fails absolutely miserably when you actually use it.
valuable as those exoteric rules may be for some students just starting out, without damaging science and technology irrepairably.
The damage is that many universities teach outdated junk statistics to their students. Fortunately, it's slowly dying out, but I suspect it will still be another couple of generations of academics before it's completely gone.
What does that have to do with this data? Pearson's r is for linear relationships between two variables, not rational relationships between many variables. (And that's not even getting into the intrinsic problems with using Pearson's r as a statistical test for anything, even if its assumptions are satisfied.)
Slashdot Mirror
I was careful to say "preferring non-involvement", not "avoiding action at all cost".
I'm sure I don't need to remind you, but the US nearly doomed England and the rest of Europe by a lack of military support in WW2 until Japan brought the war home.
Let's stop for a moment and think about what would have happened if the US hadn't gotten involved in WWII. So, Germany takes over Britain and a few more European countries. Do you really think that that ridiculous government with its ridiculous ideology and murderous tendencies would have lasted very long? Do you think the British or French would have put up with German occupation for very long? Do you think a country that classifies science it doesn't like as "degenerate" and had just killed or exiled most of its elite would manage to keep up technologically? The "Reich" would have fallen apart after a few years, probably with fewer deaths than WWII actually caused.
Of course, in hindsight, US involvement in WWII probably was overall positive. It came too late to save the Jews, but at least, it propelled Europe rapidly into its current liberal and democratic existence, something that would probably have taken decades longer if the Germans had (temporarily) prevailed. But it is dangerous to paint pictures of "doom" because then you suddenly see just about any action, however disproportionate, as justifiable. Which is just what we are seeing today in US politics.
For instance, if Isreal hadn't destroyed the nuclear power plant Iraq was building, it's certain that Iraq would now be a nuclear power
Yeah, big deal. What would Iraq be doing with it? Any place they could use it that we care about would risk swift and terrible retaliation, and Iraq could be doing a lot more damage with biological weapons.
Israel probably did the right thing anyway, but that's because the destruction of the reactor was a small, tactical effort (and even that was at least officially condemned by the US and the UN). Starting a war over an issue like that would have been something entirely different.
Most colonial powers felt that they were bringing the light of civilization to "the savages". That's why they brought religion, education, and administration with them. But what you consider "taking responsibility" may just as well be viewed as the extermination of entire cultures.
Consider, just as one terrible example, Africa before, during, and after colonization. Both before an after, terrible tribal wars, massacres, poverty, and raw, naked power terrifying the weak. During colonization? In some places the same or worse (Congo, the Boer States). In other times and places much, much better (much of British ruled Africa).
I think your analysis already presupposes a Western point of view. For example, people aren't poor until they come to accept Western ideas of wealth.
My point is not that arrogance or bullying is somehow good. It is that I don't think the Europeans have so much improved morally as become throughly cynical and insular.
My point was that I don't think the British needed the Nazis to teach them about totalitarianism or mass murder--the British had figured out and practiced those things for themselves on and off, as had most Western nations.
As for what Europeans are doing today, I think their attitude of preferring non-involvement is the better one. It is the Americans that now erroneously believe that they can carry the torch of freedom and democracy around the world and do good everywhere. It won't work. America's foreign policy has caused, and will continue to cause, enormous suffering. Sooner or later, the Europeans and the UN will likely put pressure on the US to stop its foreign adventures and clean up its domestic problems.
That's all well and good, but why do you think Germany started WWI and WWII? Did Britain have any more justification for wanting to be powerful, independent, and autonomous in the 19th century than Germany did in the 20th century? Did British aspirations justify the death of millions of natives abroad and the destruction of entire cultures any more than Germany's aspirations (failed to) justify an aggressive war in Europe and the destruction of millions of Jews and other "undesirables"?
All the major European nations were evil empires until the 19th and 20th century. Britain didn't need Nazi Germany to teach it about expansionism, racism, genocide, or totalitarianism, the British knew about those things well enough all by themselves and had practiced them, on and off, for centuries. It was only WWII that brought that to an end and finally caused European nations to come to terms with their histories.
Realizing that evil empire building was an all-European activity is important, not in order to attempt to exculpate Germany, but to understand that Britain isn't immune from these kinds of tendencies. British democracy might go bad in just the same ways that the Weimarer republic self-destructed. Fortunately, European integration and diversity has made this fairly unlikely these days.
Perhaps that realization is even more important for Americans, who have a powerful and increasingly centralized government. Americans seem to think that they don't have to worry about any of those pesky privacy and human rights issues because "The Constitution" will protect them like a magic shield, and that they are justified in all their actions because, almost by definition, anything America does is in the name of freedom and democracy.
The same is true for most of the "great" inventions or ideas we celebrate. It is very rare indeed that a ground breaking new idea appears out of the mainstream, and when it does, it usually doesn't catch on until the mainstream catches up with it and someone else gets the credit.
Exactly my point: going out all over the world, destroying one civilization after another, subjugating the native populations, and plundering its natural resources.
If only Stalin had conquered the world instead!
The world wasn't for anybody to "conquer": not for the British, not for the Spanish, not for the French. The fact that Stalin or the Nazis created evil regimes does not diminish the profound evil of those other empires. It's only in the second half of the 20th century that Europeans finally came to their senses.
And what makes you think the UK wasn't? What does winning WWII have to do with the UK political system? Might makes right? The UK was committing atrocities and genocide for centuries before the Nazis even appeared on the scene--the UK just did it out of sight, overseas, to people who even today largely don't have much political say. The British Empire was one of the most evil institutions in human history.
Today, the UK has reformed, just like the rest of Europe, although it seems still quite a bit more right leaning.
How does removing a biometric identifier help you "get around a problem"? Right now, say, if your credit is bad and people refuse to give you a loan, pretending to be someone else is merely adding fraud to your list of problems. The fact that lack of sound biometric identifiers makes committing fraud harder is, in my opinion, a plus.
Yes, and if you have need of a monarch, you are in trouble. But because it's a fact of modern life that people have always need of their government, we have made government by the people and for the people. And you "always and in every possible condition of things have need" of your fellow citizens and society: without them, your food doesn't reach you, your water doesn't reach you, your electricity doesn't reach you, your roads don't get maintained, you don't get medical service, etc. Welcome to the real world.
The government should be tightly regulated in its use of any personal information, something both the US and the UK are shamefully poor at. But fighting the introduction of well-designed national ID cards accomplishes nothing--it just means that the UK (and US) governments will keep mixing your files up with someone else's in addition to violating your privacy constantly.
Current ID cards have almost no protection against identity theft. You see, even in the US we have national ID cards, they just don't work very well. New identity cards are an attempt to improve the situation.
If you think they aren't going to succeed, then you have to say why. But your blanket statement is simply logically flawed.
Really, if Europeans want to have ID cards, no one in the UK has a problem with that, and no one here is interested in arguing with Europeans who think that ID cards are "no problem at all".
The people of the UK are Europeans--it's a simple geographic and political fact.
Sure, some MS executives have said something like "we have lots of intellectual property in this great new technology", but every company that wants investors to drive up the stock price says that. Nobody has yet been able to produce any concrete, substantial intellectual property that Microsoft (or anybody else) holds in C# or .NET. The few patents that have surfaced have been peripheral and easy to work around. So, if you have concrete knowledge of this "legal minefield", please tell us all. Otherwise, please don't keep spreading FUD about Mono.
First of all, there is almost nothing technologically new in Windows NT: its "innovations" in areas like file systems, access control, security, etc., have pretty much all been tried before. People don't use them by choice. Furthermore, IBM has created a pretty complete Windows clone as part of OS/2, and they didn't get into legal trouble, so there is little reason to believe that ReactOS would have any more problems.
Unfortunately, none of the current crop of 64 bit processors deliver: the cost of true 64 bit systems (those capable of actually using more than 4 Gbytes of memory) generally starts somewhere upwards of $10000, and for that you do not get anywhere near 10 times the performance of a $1000 PC.
The main reason right now to get a 64 bit system at current prices is because the applications just cannot be shoehorned into a mere 2-4 Gbytes. If AMD can change that equation and deliver comparable bang-for-the-buck to current PCs, with 64 bit addressing being icing on the cake, they have a winner. None of the other players seem to be capable of doing that--they have tried and failed miserably so far.
Oh, I understand exactly what it brings: tons of really useful functionality, but I can only get most of it if I subscribe to the whole KDE master plan.
Now, Quanta COULD write such a higher level API, but wouldnt that be just recreating kioslaves, only for each app? That makes no sense.
Indeed, it wouldn't. But it also doesn't make sense to invest lots of time in something like kioslaves if it is only going to be useful for building KDE apps. What would make sense would be to create a library like kioslaves that would be used with lots of toolkits: wxWindows, FLTK, Gtk+, etc.
The KDE (and, to much the same degree, Gnome) attitude is: f*ck the rest of the universe, we are just going to live in our own little world and tinker with out intricate web of libraries. And then we are going to take over the world because we are a little better than everybody else. That's Windows thinking. That's mainframe thinking. It sucks. It's exactly what UNIX was created in reaction against.
Come on, pay at least a little attention. I'm not disputing the utility of the individual components that KDE provides. Quite to the contrary: I think the functionality that KDE components provide is extremely useful. All the worse that KDE is taking the approach of creating an integrated, consistent, interdependent set of libraries and software components that reuse each other. While those adjectives doubtlessly sound good to you, what they mean is that nobody is using kioslaves with other environments.
Lots of systems have taken KDE's all-or-nothing design approach over the last 30 years. It's great in the short run: it's really easy to do, and everything works together oh-so-nicely. But most of the time, the "all-or-nothing" turns into "nothing", and then no software components survive.
Pray, Mr. Troll, where would I find those miraculous libraries?
Circle the wagons! KDE is under attack! Retaliate with insults! Just don't bother trying to understand what someone is saying. Great going. And so typical.
Hopefully, as a consequence of this strategy, manned flights will be scrapped for the foreseeable future and we can concetrate our efforts on unmanned probes.
But they do: they have their own standards and software infrastructure for inter-client communication, audio, and a lot of other features.
For example: Suppose Quanta was not a KDE application. Now imagine a webmaster wants to use Quanta to edit pages on a website. Since that is a very necessary feature, Quanta would have to implement some sort of ftp client. And perhaps also a scp/sftp client, a webdav client, and so on for every mechanism it wanted to support. But... luckily Quanta *is* a KDE app. So, it got all that for free. And if tomorrow someone writes a mechanism to access any other remote site, Quanta will get it too.
There is nothing wrong with reusing software. But we already have libraries for dealing with FTP, SSH, WebDav, and all those other protocols.
The errors of KDE (and Gnome, for that matter) are that they are reinventing the wheel and picking a single winner. Rather than for the best FTP library to win through the choice of application developers, KDE's philosophy is that you use the KDE library. And if my application doesn't use the KDE library, then it won't integrate well with all the other KDE apps that do. And if you use one KDE library, probably you have to use more and more, and KDE servers, and lots of other stuff, since it's all interlinked and interdependent.
KDE commits the same error as the Soviets: the problem of designing useful parts for a free market of ideas and components is so daunting that they instead fall back on central planning and central specifications. And the result is just like the Soviet Union as well: initially, a slick and consistent machinery, but it is already getting stale and it will sooner or later collapse under its own weight.
I think efforts like Gnome and KDE that try to put in place a gigantic, sluggish infrastructure and try to force everything to use the same libraries are largely a waste of time and effort. Not even Microsoft or Apple are that consistent.
The promise of really low-cost 64bit systems is what makes AMD interesting. Server chips traditionally have a poor cost/performance ratios; they are for premium priced systems that only cover a small fraction of the true server market. If Opteron systems end up being not much cheaper than Itanium systems, they aren't interesting at all as far as I'm concerned.
But Apple has made the system much more proprietary and non-standard than it needs to be. The system administration database is different from mainstream UNIX systems made integrating the Macs into my home and work networks much more work than a Linux machine. The window system is completely different from UNIX, hard to port to, and rather sluggish. Apple's software package management is worse than even that of Windows. And the commercial software situation on Macintosh is not all that great either. A big disappointment, too, was that Apple had promised "free .Mac service with every iMac" and then started charging less than a year later.
Altogether, I think Apple has benefitted quite a bit from UNIX/Linux compatibility, by promising a no-hassles Linux-like environment and attracting some UNIX and Linux users. I don't think they really have delivered, and I will probably not be upgrading my Macs--I can get better functionality and more software for less money with Linux. On the other hand, Linux has not benefitted directly from OS X: there is little or no useful software that Apple has donated to the Linux community (Darwin is more of a distraction), and I don't think Apple's "switch" campaign has been all that effective.
I think in the long run, Apple will be forced to become more and more Linux compatible, and then maybe there will be more benefit to the Linux community. Until then, every Windows user that moves to Macintosh is still of some benefit to the Linux community.
The military could and should go with software that is based on open standards: UNIX/POSIX, X11, etc. And in their implementations and deployments, they should then stick as much as possible to those open standards. They can then buy software and hardware from many different vendors and have a choice among multiple implementations, including some open source ones.
I'd also recommend Hastie et al. "The elements of statistical learning". For general machine learning, Mitchell's book "Machine Learning" is also frequently recommended.
Books on support vector machines generally also talk about the tradeoffs more theoretically (VC dimension, etc.). The following book may be reasonably readable: "An Introduction to Support Vector Machines and Other Kernel-based Learning Methods" -- by Nello Cristianini (Author), John Shawe-Taylor (Author);
In 1996, there were two huge sites designed in a modern way--the Olympics and Wimbledon. Of course, they had navigation bars and dynamically generated content.
Yes, and that's in effect what I was saying, but that's not relevant. You get to apply Pearson's tables only once to a single linear relationship computed on one dataset. If you keep constructing different combinations of variables and keep testing for normality and correlation, the significance of the test sooner or later erodes. Sometimes you get lucky and you get to try a lot of hypotheses before that happens (which is why people often get away with using it), and sometimes you don't.
Look at it this way: pick any distribution of 47 (u,v) values that satisfies your normality and correlation test. Now, I can construct a pair of polynomials that maps any 47 datapoints (x,y) you actually measured onto those 47 (u,v) values. If you had just kept trying different polynomials and fitting their coefficients, you'd eventually have hit upon this. So, you can't just keep trying different combinations of variables: sooner or later, you are going to find one that satisfies whatever statistical tests you apply.
What you have to avoid, but not all all costs, is mindless "data trolling" -- just fishing for any correlations between random variables regardless of your prior hypotheses. The costs of not doing such data trolling are that frequently one has hypotheses that can't be directly measured or that are still in formation and one is searching for means of refining the hypotheses or ways of testing. This is tricky and can't be reduced to simple bromides like "don't data mine" or "correlation doesn't imply causation", etc.,
Maybe it's tricky for people who only have "Pearson's r" in their bag of tricks. But there is nothing "tricky" about it statistically. The question of what hypotheses one can explore and what the significance of the results is is reasonably well understood. Machine learning techniques routinely explore billions of possible formulas and hypotheses describing the relationship between variables of some dataset. But they use accurate, theoretically justified measures of the significance of such hypotheses. Pearson's r doesn't work for that; that's not a theoretical debate or a matter of preference, it just fails absolutely miserably when you actually use it.
valuable as those exoteric rules may be for some students just starting out, without damaging science and technology irrepairably.
The damage is that many universities teach outdated junk statistics to their students. Fortunately, it's slowly dying out, but I suspect it will still be another couple of generations of academics before it's completely gone.
What does that have to do with this data? Pearson's r is for linear relationships between two variables, not rational relationships between many variables. (And that's not even getting into the intrinsic problems with using Pearson's r as a statistical test for anything, even if its assumptions are satisfied.)