IT is a boom and bust field that will gradually decay in its value as the technologies it is based on mature.
Sure it will, but we still have at least another 50 years to go before it has fully matured. Look around you at all the things that aren't yet computerized or are shoddily computerized. This should tell you how much farther we have to go. Things that ought to be fully computerized but aren't:
Every light switch in your home.
The microwave oven (you should enter the desired temperature as in lukewarm, warm, hot, boiling, not the time).
Driving your car.
Buying your house (enter salary, work place, number of family members, importance of school, yard, etc. computer gives you top three choices).
Every single piece of paperwork that still involves paper.
CS is at best a branch of practical mathematics, but even that would be wrong. Pray tell which part of mathematics studies the interface between a computer and a human?
All of these which are minor compared to the damage we are causing in the oceans. Make no mistake: moving to fully farmed fishing would be great progress from the status quo and a huge step in the right direction, even if still not the 100% perfect solution.
Only ideologues would oppose such a step forward, simply because it doesn't solve the problem completely in a single stroke (what would?)
"Microsoft is under no obligation to produce any service packs, ever," he explains.
This is not quite accurate. As the Ford Pinto case illustrates manufacturers have certain obligations when it comes to the performance of their product. If a major, easily fixable malfunction in WinXP was discovered tomorrow and M$ refused to fix it, it is very likely that a class action lawsuit against it would succeed.
How can you tell a Unix guru apart from a wannabe? The unix guru positively knows that X sucks, but gives in to it in the name for backward compatibility, the wannabe thinks X is good, 'cuz it belongs to Unix.
I applied for a job at Microsoft in which I was supposed to do some specialized, rather advanced research management. The interview boiled down to "do this puzzle". I'm actually good at them, having ranked in the top 99.97% of the population in IQ tests (the ultimate "silly puzzle" test if there was ever one). The problem is that my job had nothing to do with fast problem solving, but all about long term planning, lots of specialized knowledge and experience. I discussed this with my interviewers and expressed my willingness to be tested in things relevant to my job but they wouldn't budge. So I solved the puzzle in a way that was as efficient as any they had ever seen but totally new. They were not impressed.
Then and there I realized that Microsoft shares were not about to move up for a while. This was about two years ago.
My company doesn't buy stuff from Dell anymore, it quit well before this call. They moved to Alienware (although Dell just bought them out, so now they are buying from some local shop near the Omaha office now I think). I haven't bought anything from ATI since, and probably never will. I will be personally buying a laptop in the not-too-distant future. It won't be a Dell.
The last three Dells I bought crapped out way too early. I've stopped buying them. Next computer I buy likely will be an HP.
The Semantic Web is in the old AI tradition of grand overhyped promises with little results to show for them many years later. AI had managed to moved away from this practice that had led to the crisis in funding in the 80s, when people woke up to the fact that AI did not deliver as promised. Here at AAAI there is a sentiment that the semantic web is a step in the wrong direction and Tim Berners-Lee talk here was presented as such. Here's the abstract from the program:
The relationship between AI and the semantic web has been something that has provoked a lot of heated corridor discussion over the years. This talk will try to outline what the semantic web is and is not, at a conference where there may be some anniversary reflection on what AI is and is not. It is not always obvious how to transfer existing AI techniques into a fractal weblike space, or what the effect will be. But it is certainly exciting.
The same chasm appeared during the founders panel, where John McCarthy gave a more sober cautious perspective of where AI is going, while Marvin Minsky issued a call for the old style of over-hyped research such as the "emotion machine" whatever that means.
There was also a feeling that perhaps some of the grand challenges are too ambitious. "We can't make predictions since in some cases we don't even know what the problems are" a famous panelist noted. It is good to have long term goals, but they must be set within the realm of what is at least vaguely foreseeable. Challenges beyond that boundary are in the realm of science fiction not scientific AI.
I've been highly critical of Tim Berners-Lee leadership on the W3C. He established a structure that sidelined individual, mostly-disinterested members and replaced them by corporations interested in log-jam and difficult implementations that keep the small players away. The W3C was from the get go the antithesis of the IETF.
Tim then jumped into the dubious "semantic web" runaway train, full of inflated promises but bereft of actual results. The "semantic web" is high-risk research best left in the hands of academia. A standards body organization should be focusing on how to make the web better today, by improving on the current protocols, not on day dreaming about HAL-like computers.
you are doing 743 queries per second, great-grand parent post claimed 50,000 queries per second. That is two order of magnitudes larger. Google itself doesn't see that transaction rate.
50,000 inserts per second? What the fsck are you doing that for? Are you logging each packet individually? 50,000 inserts per second means 72 million transactions per day. Only the largest web sites in the world see that type of transaction rates.
Yes, a lot of it is just syntactic sugar to you language people,
Syntactic sugar is a phrase devised by programming language types to justify incompetent HCI design in their pet language. E.g. programmer: "but your choice of using ancient hungarian for keywords makes the language difficult to use in practice". PL-type response: "ah that is just syntactic sugar".
Syntactic sugar is like syrup for pancakes: judicious amounts make them infinetely more palatable, but it does not mean that the more syrup the better.
The laptop works, bleh! Nobody ever doubted that one could build an orange laptop. The question is if one can build a usable laptop today for less than $100. So far all he has produced is a model produced at a cost of tens if not hundreds of thousands of dollars. Even at mass production rates, the cost of the LCD alone remains over $100. Yes, by the cost will go down with time, but by that measure you can already buy my original Toshiba laptop today for less than $100, or even my original IBM PC/AT for $30.
Since like, forever. From the moment C was introduced it was described as "high level assembler" or conversely "low level algol" (and later pascal). You do remember Pascal don't you. High level languages are strongly typechecked and do not provide direct memory access (i.e. &).
What part of theoretically you didn't understand? In practice, low level languages like C are a hell of a lot more bug prone than high level, type checked languages like Java and Pascal.
But you have no control over this. It all happens under the covers. UAC would explicitly ask you: do you authorize apache to 'sudo' listen to port 80 (yes/no)?
Actually the article describes a bug in the implementation of the UAC. However, the UAC modelis superior to the su/sudo unix model. Vista UAC is sudo on steroids. Rather than have the system return an error message saying "you need to execute the whole thing as superuser" you get prompt back it says "this program needs to be superuser to execute the following specific instruction. do you auhorize this?". I'll take the Vista model anyday and the sooner Linux implements something similar (but cleanly) the better the chances of OSS succeeding.
1. User needs a particular application. Depending on company policy, the user may be able to install in their own home folder. If not, they could submit a request to suppot.
Pfffffft. I can't remember when was the last time I could install anything more sophisticated than a shell script without being superuser in Unix. To top it all off, even after you install the.rpg often at the end you still need to manually edit a couple of system files.
Correct. Minix is a mickey mouse academic operating system that contains no paging and no graphics facilities, but aside from that "it's a real operating system". Yeah right.
But in practice Linux 2.6 is 6 million lines of code and a typical microkernel is less than 10k.
You are parroting a meaningless comparison. This is like saying that a car has four wheels while typically the trunk has only one (the spare tire) hence the trunk is a more efficient car.
A real life operating system when put together has over 30M lines of code, some of those are the true kernel-OS functions, some of them are the file system, some are the graphics system, some are the networking facilities.
Microkernels are not smaller OSs, the are smaller code bases running in privileged mode and a larger code base running in unpriviliged mode.
Microkernels are much harder to write and maintain exactly because of this issue. You can do simple things easily - and in particular, you can do things where the information only passes in one direction quite easily, but anythign else is much much harder, because there is no "shared state" (by design).
Andy Tanenbaum gave a talk on this and it was very clear that Minix3 is a typical toy operating system being developed and tested in academia, with no real insight of what is needed out there in the real world. Tanenbaum had his own version of "no one will ever need more than 640K" in claiming that no one will ever need more than 1GB of memory. I've seen that movie, starting with 640K, them 1MB when the EMM came out, then 16MB with the 386 then 2GB with the Pentium and on and on. Every time we rose up to the challenge and used all the memory, yet Andy is still at it, repeating the same platitudes.
The sonic boom prevents any overflight of populated areas and even if significant noise reduction could be achieved the very small constituency for such a service would still see any residual boom noise used as an excuse by the general (and envious) public to restrict or outright ban such overflight
The sonic boom is a redherring thrown by the US airline industry when they lost the SST race. There are many cities in the US where sonic booms are (or used to be) routine. Tucson, Seattle, and anyother city with major military related development areas nearby. Nobody ever complained.
Then you need to look up how the Bush administration refused to reimforce the levies in spite of clear advice that they would fail given a category 3 hurricane.
Slashdot Mirror
IT is a boom and bust field that will gradually decay in its value as the technologies it is based on mature.
Sure it will, but we still have at least another 50 years to go before it has fully matured. Look around you at all the things that aren't yet computerized or are shoddily computerized. This should tell you how much farther we have to go. Things that ought to be fully computerized but aren't:
Every light switch in your home.
The microwave oven (you should enter the desired temperature as in lukewarm, warm, hot, boiling, not the time).
Driving your car.
Buying your house (enter salary, work place, number of family members, importance of school, yard, etc. computer gives you top three choices).
Every single piece of paperwork that still involves paper.
the list goes on and on.
CS is at best a branch of practical mathematics, but even that would be wrong. Pray tell which part of mathematics studies the interface between a computer and a human?
looks like c|net is now hiring highschool students to write their stories:
OMG this is, like, the worst christmas _ever_. That bill gates guy ruins _everything_. It is, li
All of these which are minor compared to the damage we are causing in the oceans. Make no mistake: moving to fully farmed fishing would be great progress from the status quo and a huge step in the right direction, even if still not the 100% perfect solution.
Only ideologues would oppose such a step forward, simply because it doesn't solve the problem completely in a single stroke (what would?)
"Microsoft is under no obligation to produce any service packs, ever," he explains.
This is not quite accurate. As the Ford Pinto case illustrates manufacturers have certain obligations when it comes to the performance of their product. If a major, easily fixable malfunction in WinXP was discovered tomorrow and M$ refused to fix it, it is very likely that a class action lawsuit against it would succeed.
How can you tell a Unix guru apart from a wannabe? The unix guru positively knows that X sucks, but gives in to it in the name for backward compatibility, the wannabe thinks X is good, 'cuz it belongs to Unix.
I applied for a job at Microsoft in which I was supposed to do some specialized, rather advanced research management. The interview boiled down to "do this puzzle". I'm actually good at them, having ranked in the top 99.97% of the population in IQ tests (the ultimate "silly puzzle" test if there was ever one). The problem is that my job had nothing to do with fast problem solving, but all about long term planning, lots of specialized knowledge and experience. I discussed this with my interviewers and expressed my willingness to be tested in things relevant to my job but they wouldn't budge. So I solved the puzzle in a way that was as efficient as any they had ever seen but totally new. They were not impressed.
Then and there I realized that Microsoft shares were not about to move up for a while. This was about two years ago.
My company doesn't buy stuff from Dell anymore, it quit well before this call. They moved to Alienware (although Dell just bought them out, so now they are buying from some local shop near the Omaha office now I think). I haven't bought anything from ATI since, and probably never will. I will be personally buying a laptop in the not-too-distant future. It won't be a Dell.
The last three Dells I bought crapped out way too early. I've stopped buying them. Next computer I buy likely will be an HP.
BS. Most studies have shown no benefit whatsoever from giving every kid a computer.
The same chasm appeared during the founders panel, where John McCarthy gave a more sober cautious perspective of where AI is going, while Marvin Minsky issued a call for the old style of over-hyped research such as the "emotion machine" whatever that means.
There was also a feeling that perhaps some of the grand challenges are too ambitious. "We can't make predictions since in some cases we don't even know what the problems are" a famous panelist noted. It is good to have long term goals, but they must be set within the realm of what is at least vaguely foreseeable. Challenges beyond that boundary are in the realm of science fiction not scientific AI.
I've been highly critical of Tim Berners-Lee leadership on the W3C. He established a structure that sidelined individual, mostly-disinterested members and replaced them by corporations interested in log-jam and difficult implementations that keep the small players away. The W3C was from the get go the antithesis of the IETF.
Tim then jumped into the dubious "semantic web" runaway train, full of inflated promises but bereft of actual results. The "semantic web" is high-risk research best left in the hands of academia. A standards body organization should be focusing on how to make the web better today, by improving on the current protocols, not on day dreaming about HAL-like computers.
you are doing 743 queries per second, great-grand parent post claimed 50,000 queries per second. That is two order of magnitudes larger. Google itself doesn't see that transaction rate.
50,000 inserts per second? What the fsck are you doing that for? Are you logging each packet individually? 50,000 inserts per second means 72 million transactions per day. Only the largest web sites in the world see that type of transaction rates.
Yes, a lot of it is just syntactic sugar to you language people,
Syntactic sugar is a phrase devised by programming language types to justify incompetent HCI design in their pet language. E.g. programmer: "but your choice of using ancient hungarian for keywords makes the language difficult to use in practice". PL-type response: "ah that is just syntactic sugar".
Syntactic sugar is like syrup for pancakes: judicious amounts make them infinetely more palatable, but it does not mean that the more syrup the better.
The laptop works, bleh! Nobody ever doubted that one could build an orange laptop. The question is if one can build a usable laptop today for less than $100. So far all he has produced is a model produced at a cost of tens if not hundreds of thousands of dollars. Even at mass production rates, the cost of the LCD alone remains over $100. Yes, by the cost will go down with time, but by that measure you can already buy my original Toshiba laptop today for less than $100, or even my original IBM PC/AT for $30.
Since like, forever. From the moment C was introduced it was described as "high level assembler" or conversely "low level algol" (and later pascal). You do remember Pascal don't you. High level languages are strongly typechecked and do not provide direct memory access (i.e. &).
What part of theoretically you didn't understand? In practice, low level languages like C are a hell of a lot more bug prone than high level, type checked languages like Java and Pascal.
But you have no control over this. It all happens under the covers. UAC would explicitly ask you: do you authorize apache to 'sudo' listen to port 80 (yes/no)?
Actually the article describes a bug in the implementation of the UAC. However, the UAC model is superior to the su/sudo unix model. Vista UAC is sudo on steroids. Rather than have the system return an error message saying "you need to execute the whole thing as superuser" you get prompt back it says "this program needs to be superuser to execute the following specific instruction. do you auhorize this?". I'll take the Vista model anyday and the sooner Linux implements something similar (but cleanly) the better the chances of OSS succeeding.
Pfffffft. I can't remember when was the last time I could install anything more sophisticated than a shell script without being superuser in Unix. To top it all off, even after you install the .rpg often at the end you still need to manually edit a couple of system files.
Correct. Minix is a mickey mouse academic operating system that contains no paging and no graphics facilities, but aside from that "it's a real operating system". Yeah right.
But in practice Linux 2.6 is 6 million lines of code and a typical microkernel is less than 10k.
You are parroting a meaningless comparison. This is like saying that a car has four wheels while typically the trunk has only one (the spare tire) hence the trunk is a more efficient car.
A real life operating system when put together has over 30M lines of code, some of those are the true kernel-OS functions, some of them are the file system, some are the graphics system, some are the networking facilities.
Microkernels are not smaller OSs, the are smaller code bases running in privileged mode and a larger code base running in unpriviliged mode.
Microkernels are much harder to write and maintain exactly because of this issue. You can do simple things easily - and in particular, you can do things where the information only passes in one direction quite easily, but anythign else is much much harder, because there is no "shared state" (by design).
Andy Tanenbaum gave a talk on this and it was very clear that Minix3 is a typical toy operating system being developed and tested in academia, with no real insight of what is needed out there in the real world. Tanenbaum had his own version of "no one will ever need more than 640K" in claiming that no one will ever need more than 1GB of memory. I've seen that movie, starting with 640K, them 1MB when the EMM came out, then 16MB with the 386 then 2GB with the Pentium and on and on. Every time we rose up to the challenge and used all the memory, yet Andy is still at it, repeating the same platitudes.
The sonic boom prevents any overflight of populated areas and even if significant noise reduction could be achieved the very small constituency for such a service would still see any residual boom noise used as an excuse by the general (and envious) public to restrict or outright ban such overflight
The sonic boom is a redherring thrown by the US airline industry when they lost the SST race. There are many cities in the US where sonic booms are (or used to be) routine. Tucson, Seattle, and anyother city with major military related development areas nearby. Nobody ever complained.
Then you need to look up how the Bush administration refused to reimforce the levies in spite of clear advice that they would fail given a category 3 hurricane.