The advantage of a committee is not that the committee is better than one man with a vision (I can't think of any instance where that's true, even when the committee includes the original one man) but rather that the committee can keep going even when some of its members leave or are busy. The one-man-band approach can't do that.
Don't know, but does it matter? Would you want to work for or employ a paranoid maniac who hires private detectives to watch not only you, but your family and friends too? Would you do so even if there were no National Security justifications? Since that's the situation here.
Guess you are, with eyes that see and ears that hear in far greater fidelity than the rest of us, and at much higher frame/frequency rates too. Must be absolutely terrible watching movies and seeing all those film grains all over the place and those inaccurate least significant bits in the audio signal!
There's a physiological limit to our perceptive abilities, and going beyond it (or, frankly, even very close up to it) is a waste of money.
You forget that sometimes programs need access to shitloads of data. Big simulations, big images, big databases, etc. Operating systems have supported files larger than 4GB for a long time now for good reason, and the HPC world has been 64-bit for over a decade, with other sorts of big-iron server not far behind. (More than 4GB of program code is still obscene though. How can you possibly debug anything that large?)
If there was a critical threat to the UK, how come the PM flew out on holiday two days ago?
Come one now, that's an easy one! Until arrests are made, things have to look normal. Moreover, there's no real point in him being there anyway. This is all currently being run at an "Operations" level, whereas politicians work more at a "Strategy" level. What earthly good would having him about shoving his oar in do?
This is why pure machine language code will always be faster - even if more horribly difficult - than any high level code. Having a machine - programmed by fallible humans - checking fallible code - is just another path for potential failure.
Umm, that's really rather off what most people observe in reality. While in the limit the directly-written machine code (MC for short) might be better than anything out of a compiler, the level of guru-dom required to do this on modern architectures is really ridiculously high and the error rate (per kilobyte of object code) with direct writing of MC is through the roof as the real problem is that the error rate per line of input code (however expressed) is probably close to constant anyway. It's way cheaper to throw a few extra processor cycles at a problem than it is to train people to write everything in MC. Especially if you are targetting multiple architectures; MC is utterly non-portable you know. All that abstraction means you can spend most of your time dealing with the higher-level aspects of the problem ("is the size of the window right?") instead of dealing with the nitty-gritty ("am I managing the registers cleanly?"); remember, those higher-level aspects still have to be dealt with by that MC program.
CPU cycles are cheap and falling in unit price, but brainpower is expensive and increasingly expensive. Tasks which can be (even only partially) automated are good as they are transferring effort from the expensive resource to the cheap resource.
Bubble sort a.k.a. the world's (second) worst sorting algorithm? You're welcome to it! (The worst is to just keep on randomly permuting the list until it is in order, which has the property of not having an upper bound on the number of operations.) Better to choose something like quicksort, mergesort, shellsort, or insertion sort; you should know the differences between all of those and when to use each.
More seriously, bubble sort has been around in the literature (as an example of how not to sort in algorithms books, and as an example of how to sort in programming books!) for truly ages. In other words, loads of prior art. Good luck with the patenting, but it's a truly stupid way to throw away money.:-)
It's part of accessibility enablement, and we have the same thing on our corporate web-pages. Google also makes use of this sort of thing to good effect in a number of places (I've noticed it in Google Calendar FWIW). What's less clear is what the correct set of keys to use for this is; there should always be some keys that are reserved to managing the browser and not the pages within (and which javascript can't rebind, to prevent obvious attacks).
Actually it's a farm of page servers, probably load-balanced, with some kind of front-end HTTP server. It's not clear whether a dispatcher process is required between the two layers. It's also not particularly innovative given that the idea of load-balanced servers has been around for many years.
Oracle might not be my favouritest guys in the world, but this one probably ought to go their way.
Also, I want to see a study of how much reading while driving impairs your ability.
I don't know for sure, but the last few times I've nearly had an accident have all involved me trying to read a map to figure out where I was. I usually try to keep all that in my head so it isn't a problem, but when you've just got off a long-haul flight to a city you don't know, that map is your best bet for finding your way. Even if you do scare the bejeezus out of the truck driver in the next lane on the freeway when you swerve while looking at the map...
The Web - HTML [...] Who was thinking of screen readers in 1981?
Nitpick: I thought the web was about a decade younger than that. 1981 was back when the IETF was nailing down many core features of the basic internet, and the key user-level apps were email (SMTP was created in 1981, according to what I can glean from the RFC index), telnet and ftp. At that time, even venerable NNTP was 5 years away from publication.
You're making a major mistake here. You're thinking the only way to build a secure core is to define badness, that is to define what must be blocked. That's well known to be a dumb way of doing it because you end up in an arms race to define all possible badness in the world. Crazy.
Far better is to define goodness. That is to say, define exactly when you'll allow something to bypass your protections. For best effect, do this in a way that is user-specific so that the onus in the arms race is wholly on the rootkit author. Let the blackhats do work (or, more likely, pick a different softer target).
[encrypting your net traffic] will slow you up a bit
Not noticably. I've been doing this for years for most traffic (not because of govt admittedly, but because I don't trust all the students round here to not run packet sniffers) and the slowdown is minimal to nothing. CPU time is just so incredibly cheap these days.
Pork: I get the Federal government to build, say, 10 bridges and a couple of federal prisons in my constituency. That's money from the outside flowing in plus new jobs paid for with federal money. The tax base is growing:-)
What about the income of all those prisoners. All those very low income people are going to knock that average down...
But in general, if the actions of a politician are increasing the number of jobs available to their constituents, that politician is doing his/her job. Moreover, pork is less effective at this because it takes jobs and money away from other constituencies (making it harder to attract the support of other politicians for the measure). Supporting real growth is definitely better, especially if the politico intends to get reelected.
Hmm, computing "energy" is interesting, even though it could lead to surprising results (such as highly rating a search engine undergoing catastrophic decline). It also ends up ranking things quite differently:
Google
0.50*0.17*0.17 = 0.01445
Ask
0.06*0.35*0.35 = 0.00735
Yahoo!
0.28*0.16*0.16 = 0.007168 (less than Ask!)Note that I had to pull the figures for the growth rate of Yahoo! out of my butt here, but since we know from the article that they're growing slower than Google, selecting one percentage point less - 16% - and calculating from there should provide a reasonable upper bound on the "energy" value, and the interesting thing is that even with all that, they're still lower than Ask, and the other search engine mentioned (MSN Search) has got to have an even lower "energy" value.
The real question is what these figures mean if anything, other than that the cartoon on the Economist page should not be interpreted as being at all representative of reality! I suppose it does mean that, for the field of search if nothing else, nobody's going to be knocking Google off their post for a while; they're utterly kings of the search hill.
The government workers in france will riot when they find out their paper shuffling jobs have been replaced by a computer.
But computers don't promote efficiency by themselves. They can also codify stupidity, inefficiency and reading slashdot^W^Wslacking. And there will be more paper to shuffle, because everyone will insist on printing out everything several times just to see how things look different if you change the font on the page numbers of the forms by a few pixels. If people don't want to be efficient, they won't be.
It's going to be a revenue strategy. Lots of MS's corporate customers really want the ability to run clustered versions of the back-end apps, so much so that when MS pushed out a half-assed version of their cluster stuff a few years ago, a fair number of their customers jumped on it and started using it to make their data centers more efficient while asking for more. As I understand it, this caught MS somewhat on the hop; they did not expect that sort of takeup at all. But given that lots of larger customers want it, why on earth shouldn't they do it and make money off it?
Of course, the really funny thing about all this is that MS are being good citizens in this space (as long as you accept that you'll be stuck drinking the Redmond KoolAid with them, which seems to be a strategy that is good enough for many businesses, alas) as they're properly engaging with other people in pushing for genuinely better specifications and interoperability in this area. OK, some slashbots will fail to accept this, but MS aren't uniformly evil; they're more like the proverbial Curate's Egg: good in parts...
Sometimes it's easier to come right out and say that you're not giving the client that because it really sucked. Candidates for that sort of thing include any features that prevent other more important features from working: "I know we didn't deliver the flight simulator module for the word-processor, but with the effort we freed up by doing this you've now got an equation editor better than TeX on steroids."
but what if the time and frequency of next transmission is coded in previous message, how are they gonna block that?
That's actually a really bad move, since then if the agent misses one transmission - e.g., through being sick or something - they can't resynchronize (not without contacting their handler, which is much more dangerous). What's worse, they might even start interpreting messages meant for other agents, resulting in completely crazy effects! Time/frequency hopping is just not robust enough.
Racing games and FPS's don't suffer from [getting boring quickly, like comedy].
Speak for yourself. To me, there's not much more boring than a racing game, and the FPS theme of "run along somewhere confusing, and shoot stuff" gets old quickly.
I can't believe this. Amnesty, a major human right organisation with lots of members, money and influence takes up internet freedom as a campaining topic and what do slashdotters do? Moan, complain and try to rip jokes out of it.
You'd better believe it. It's the traditional slashdot response to everything. If they're making jokes (in poor taste, natch) about massive natural disasters and ongoing wars, why would you have any sort of reasonable expectation that an NGO like Amnesty would have some kind of exemption? Come off it!
Slashdot Mirror
The advantage of a committee is not that the committee is better than one man with a vision (I can't think of any instance where that's true, even when the committee includes the original one man) but rather that the committee can keep going even when some of its members leave or are busy. The one-man-band approach can't do that.
There's a physiological limit to our perceptive abilities, and going beyond it (or, frankly, even very close up to it) is a waste of money.
You forget that sometimes programs need access to shitloads of data. Big simulations, big images, big databases, etc. Operating systems have supported files larger than 4GB for a long time now for good reason, and the HPC world has been 64-bit for over a decade, with other sorts of big-iron server not far behind. (More than 4GB of program code is still obscene though. How can you possibly debug anything that large?)
Come one now, that's an easy one! Until arrests are made, things have to look normal. Moreover, there's no real point in him being there anyway. This is all currently being run at an "Operations" level, whereas politicians work more at a "Strategy" level. What earthly good would having him about shoving his oar in do?
Umm, that's really rather off what most people observe in reality. While in the limit the directly-written machine code (MC for short) might be better than anything out of a compiler, the level of guru-dom required to do this on modern architectures is really ridiculously high and the error rate (per kilobyte of object code) with direct writing of MC is through the roof as the real problem is that the error rate per line of input code (however expressed) is probably close to constant anyway. It's way cheaper to throw a few extra processor cycles at a problem than it is to train people to write everything in MC. Especially if you are targetting multiple architectures; MC is utterly non-portable you know. All that abstraction means you can spend most of your time dealing with the higher-level aspects of the problem ("is the size of the window right?") instead of dealing with the nitty-gritty ("am I managing the registers cleanly?"); remember, those higher-level aspects still have to be dealt with by that MC program.
CPU cycles are cheap and falling in unit price, but brainpower is expensive and increasingly expensive. Tasks which can be (even only partially) automated are good as they are transferring effort from the expensive resource to the cheap resource.
Bubble sort a.k.a. the world's (second) worst sorting algorithm? You're welcome to it! (The worst is to just keep on randomly permuting the list until it is in order, which has the property of not having an upper bound on the number of operations.) Better to choose something like quicksort, mergesort, shellsort, or insertion sort; you should know the differences between all of those and when to use each.
:-)
More seriously, bubble sort has been around in the literature (as an example of how not to sort in algorithms books, and as an example of how to sort in programming books!) for truly ages. In other words, loads of prior art. Good luck with the patenting, but it's a truly stupid way to throw away money.
It's part of accessibility enablement, and we have the same thing on our corporate web-pages. Google also makes use of this sort of thing to good effect in a number of places (I've noticed it in Google Calendar FWIW). What's less clear is what the correct set of keys to use for this is; there should always be some keys that are reserved to managing the browser and not the pages within (and which javascript can't rebind, to prevent obvious attacks).
Actually it's a farm of page servers, probably load-balanced, with some kind of front-end HTTP server. It's not clear whether a dispatcher process is required between the two layers. It's also not particularly innovative given that the idea of load-balanced servers has been around for many years.
Oracle might not be my favouritest guys in the world, but this one probably ought to go their way.
Also, I want to see a study of how much reading while driving impairs your ability.
I don't know for sure, but the last few times I've nearly had an accident have all involved me trying to read a map to figure out where I was. I usually try to keep all that in my head so it isn't a problem, but when you've just got off a long-haul flight to a city you don't know, that map is your best bet for finding your way. Even if you do scare the bejeezus out of the truck driver in the next lane on the freeway when you swerve while looking at the map...
Ahem.
Nitpick: I thought the web was about a decade younger than that. 1981 was back when the IETF was nailing down many core features of the basic internet, and the key user-level apps were email (SMTP was created in 1981, according to what I can glean from the RFC index), telnet and ftp. At that time, even venerable NNTP was 5 years away from publication.
You're making a major mistake here. You're thinking the only way to build a secure core is to define badness, that is to define what must be blocked. That's well known to be a dumb way of doing it because you end up in an arms race to define all possible badness in the world. Crazy.
Far better is to define goodness. That is to say, define exactly when you'll allow something to bypass your protections. For best effect, do this in a way that is user-specific so that the onus in the arms race is wholly on the rootkit author. Let the blackhats do work (or, more likely, pick a different softer target).
But in general, if the actions of a politician are increasing the number of jobs available to their constituents, that politician is doing his/her job. Moreover, pork is less effective at this because it takes jobs and money away from other constituencies (making it harder to attract the support of other politicians for the measure). Supporting real growth is definitely better, especially if the politico intends to get reelected.
The real question is what these figures mean if anything, other than that the cartoon on the Economist page should not be interpreted as being at all representative of reality! I suppose it does mean that, for the field of search if nothing else, nobody's going to be knocking Google off their post for a while; they're utterly kings of the search hill.
And anyway, why is the federal government so unpatriotic as to not use .gov.us instead of .gov? You'd think they loved their country more than that...
Of course, the really funny thing about all this is that MS are being good citizens in this space (as long as you accept that you'll be stuck drinking the Redmond KoolAid with them, which seems to be a strategy that is good enough for many businesses, alas) as they're properly engaging with other people in pushing for genuinely better specifications and interoperability in this area. OK, some slashbots will fail to accept this, but MS aren't uniformly evil; they're more like the proverbial Curate's Egg: good in parts...
I read that article title and immediately wondered if the researchers were using BabyBio or MiracleGro...