Those might be valid criticisms if his books were about assembly language programming or computer architecture. But they are not. They are about algorithms, and his MMIX programs illustrate how those algorithms are implemented by a computer in terms of the instructions it executes, and the registers, cache, and memory. If a reader wants to learn about real-world CPUs, they can read Hennessy and Patterson.
I find that unit testing helps me to clarify what I want. I usually write the tests before I write the code that will pass them. That's when I start thinking about the boundary cases, and I first think about what the result should be before I think about how I'm going to handle those cases. When I get to the point of writing the code, the tests I wrote help me identify where I need to deal with boundary conditions and how I should do so. The type of programmer that just codes first without thinking, then writes the tests as an afterthought, probably isn't making the most of unit testing.
Remember that MMIX is not designed to be a practical hardware computer architecture. It's designed to illustrate algorithms written in assembly language. It's optimized for humans to read and write, not for computers to execute quickly. I'm glad that he's keeping assembly as part of his books, and that's he's updated them to a 64-bit RISC architecture. Reading MMIX assembly programs is the closest to hardware that some readers will ever get, so he has one chance to show those readers how computers actually work. It had better be as simple as possible for people to understand.
He's still very useful and productive. He's currently churning out new material for The Art of Computer Programming. This includes the new MMIX 64-bit RISC architecture computer that replaces the aging MIX computer and new material on combinatorial algorithms.
I agree with you on unit tests. I find them invaluable to ensure I have a piece of code working properly, and to ensure that I don't introduce errors when I make changes. Probably anyone that didn't write The Art of Computer Programming finds them useful as well.
I guess it could happen, and unleash untold riches in new pharmaceuticals, now that the shackles of patents that are stifling technological development have been released. It could mark the beginning of a new era where diseases are cured instead of treated, of two-hundred-year lifespans, of plentiful cheap medicines for any who are sick. And the hungry could be fed and the cold clothed, by a world elected body dedicated to helping and caring for the citizens of the World!
Certs being free information would completely invalidate their purpose. If you disagree, go ahead and send me your cert. You wouldn't mind me impersonating you, would you?
No, universal coverage would be analogous to everyone having to play and not being able to change their bets according to the cards already shown. No one could unfairly benefit from card counting in that situation.
How about, you make 10 times your dev investment back, you then have to relinquish your patent?
The problem is that most drugs don't even make it to market, because problems are found in the clinical trials. This may be because the drug is not as effective as currently available ones or because of adverse events. Pharmaceutical companies would probably need to recoup more than ten times their investment in drugs that make it to market just to break even on the drugs they made no money on.
It's not that people would suddenly lose interest. It simply wouldn't be economically feasible for private enterprise to do it. Who would develop new drugs knowing that recouping your expenses was impossible? Would you expect the government to finance all drug development?
It's just like counting cards in blackjack. If you the dealer is not allowed to change strategy on knowledge, players that place their bets according to the cards left in the deck can make a killing. Likewise, if the insurance company is not allowed to charge you according to how likely to you are to get a disease, people who buy insurance with full knowledge of their genetic predispositions will tax the insurance system by making sure they are fully insured for the diseases they will likely get.
The proposed solution of universal coverage would remove this problem.
If any company had the freedom to manufacture and sell any drug, that company would simply start producing any drug as soon as it went on the market. They would of course be able to sell that drug for far less than the original company, as they would have almost no R&D overhead. No one would buy the drug from the original company because if would cost far more. That would prevent the company that did the research to create the drug from recouping its expenses. Soon, all companies that make new drugs would go out of business. Then we wouldn't have any new drugs. I hope this illustrates why patents on new drugs (and patents on technology that takes a significant investment to create) are necessary. Just because some patents are bad doesn't mean they all are.
Yes, we will be able to simulate the human brain by 2040 to 2050. We will not be able to do so by 2030, as I said. We will have the a computer equivalent in computational power to a human brain by around 2030. The 10-20 year lag is due to the overhead of emulating hardware that is quite different than a computer. That's why I say the first strong AI will not simulate a human brain, but rather do what a human brain does in different, more computer-like ways.
No, it is the computational power. Last year, one-half of a mouse brain was simulated at one-tenth real-time. Can you imagine how slowly a full human brain would run in 2008? It would take years just to see what happens upon it getting basic sensory stimulus, far too slowly to get useful research out of it.
I know Kurzweil makes similar claims, but I am not merely regurgitating his ideas. The computational power of the brain is massive. In order to have similar computational power in a computer, we will need to wait for many generations of improvements in processors. We will almost certainly need to move away from very powerful cores based on semiconductors timed by a clock towards smaller asynchronous processors. Even with those advancements, there's no way we'll be able to simulate a human brain in real time by 2030. I think the first strong AI will resemble a human brain about as much as the Wright flyer resembled a bird.
1. It was the academic robotics teams that did well in the 2005 DARPA grand challenge (the top finishers were from Stanford and Carnegie-Mellon). Note that this also incorporated your #6, computer vision.
5. AI continues to improve at a steady pace. It's not that "nothing ever happened". For example, the DARPA Grand Challenge was won and Kasparov was beaten by Deep Blue. I think you might be referring to "strong AI" which we won't get until about 2030 because computers simply won't be fast enough until then.
I think if anything, people aren't realizing how quickly technology is advancing. This is the nature of exponential growth. I'm sure people will be dismissing AI and robotic research right up to when everything changes, seemingly out of the blue to most people.
I suppose there may be some slight modification of local climate as heat is transferred from the desert to locations where electricity is used. Is that really worse than burning fossil fuels and chopping down rain forests though? Besides, there will be no global cooling effect, as the energy remains on the Earth, except for perhaps any generated light that escapes into space.
Yeah, I was sitting here on my ThinkPad T61 just now wondering what the issue is. No wonder I couldn't figure out what people were complaining about. Then I saw the screenshots of a glossy screen, and I have to say that totally sucks. Yeah, ThinkPad T61 may be the way to go. Mine is 15.4 inches widescreen at 1280x800 resolution (WXGA), but they go up to 1680x1050 (WSXGA).
For you to have two possible choices, you would have to know ahead of time which door Monty was going to open. If a stagehand whispered to you right before you went on "Psst... don't pick door 2 because Monty will open it to reveal a goat!" then you would always have a 50% chance of picking the car and there would be no advantage in switching. Your chances of winning increase in the original Monty Hall problem because if you're allowed to pick any of the three doors, and you pick a goat on your first choice (which you will do 2/3 of the time), you have forced Monty to open the only other door that has a goat, thus revealing to you which door has the car. That's why switching gets you the car 2/3 of the time.
The rational decision would be to take the check. The door I picked has an estimated value of $10000, less than the $12000 check. Monty didn't even have to open a door for me to make that decision.
If the goats and car do not move, then if you had a 1/3 chance of picking the car at first, how can merely opening a door change that 1/3 chance? It's only by rearranging what's behind the doors or picking another door than can change your chances of winning.
There's no debate. Of course when you first choose among three equally likely options, your chances of picking the car are 1/3. If you can't figure that out, you're completely lost before Monty even shows the goat!
Slashdot Mirror
It's a joke, son. Itanium. Titanic. Titanium. Get it?
Those might be valid criticisms if his books were about assembly language programming or computer architecture. But they are not. They are about algorithms, and his MMIX programs illustrate how those algorithms are implemented by a computer in terms of the instructions it executes, and the registers, cache, and memory. If a reader wants to learn about real-world CPUs, they can read Hennessy and Patterson.
I find that unit testing helps me to clarify what I want. I usually write the tests before I write the code that will pass them. That's when I start thinking about the boundary cases, and I first think about what the result should be before I think about how I'm going to handle those cases. When I get to the point of writing the code, the tests I wrote help me identify where I need to deal with boundary conditions and how I should do so. The type of programmer that just codes first without thinking, then writes the tests as an afterthought, probably isn't making the most of unit testing.
Remember that MMIX is not designed to be a practical hardware computer architecture. It's designed to illustrate algorithms written in assembly language. It's optimized for humans to read and write, not for computers to execute quickly. I'm glad that he's keeping assembly as part of his books, and that's he's updated them to a 64-bit RISC architecture. Reading MMIX assembly programs is the closest to hardware that some readers will ever get, so he has one chance to show those readers how computers actually work. It had better be as simple as possible for people to understand.
He's still very useful and productive. He's currently churning out new material for The Art of Computer Programming. This includes the new MMIX 64-bit RISC architecture computer that replaces the aging MIX computer and new material on combinatorial algorithms.
I agree with you on unit tests. I find them invaluable to ensure I have a piece of code working properly, and to ensure that I don't introduce errors when I make changes. Probably anyone that didn't write The Art of Computer Programming finds them useful as well.
I guess it could happen, and unleash untold riches in new pharmaceuticals, now that the shackles of patents that are stifling technological development have been released. It could mark the beginning of a new era where diseases are cured instead of treated, of two-hundred-year lifespans, of plentiful cheap medicines for any who are sick. And the hungry could be fed and the cold clothed, by a world elected body dedicated to helping and caring for the citizens of the World!
Nah!
Certs being free information would completely invalidate their purpose. If you disagree, go ahead and send me your cert. You wouldn't mind me impersonating you, would you?
No, universal coverage would be analogous to everyone having to play and not being able to change their bets according to the cards already shown. No one could unfairly benefit from card counting in that situation.
It's not that people would suddenly lose interest. It simply wouldn't be economically feasible for private enterprise to do it. Who would develop new drugs knowing that recouping your expenses was impossible? Would you expect the government to finance all drug development?
It's just like counting cards in blackjack. If you the dealer is not allowed to change strategy on knowledge, players that place their bets according to the cards left in the deck can make a killing. Likewise, if the insurance company is not allowed to charge you according to how likely to you are to get a disease, people who buy insurance with full knowledge of their genetic predispositions will tax the insurance system by making sure they are fully insured for the diseases they will likely get.
The proposed solution of universal coverage would remove this problem.
If any company had the freedom to manufacture and sell any drug, that company would simply start producing any drug as soon as it went on the market. They would of course be able to sell that drug for far less than the original company, as they would have almost no R&D overhead. No one would buy the drug from the original company because if would cost far more. That would prevent the company that did the research to create the drug from recouping its expenses. Soon, all companies that make new drugs would go out of business. Then we wouldn't have any new drugs. I hope this illustrates why patents on new drugs (and patents on technology that takes a significant investment to create) are necessary. Just because some patents are bad doesn't mean they all are.
and even NetApplications reports that users are already switching to Firefox 3.
Yes, we will be able to simulate the human brain by 2040 to 2050. We will not be able to do so by 2030, as I said. We will have the a computer equivalent in computational power to a human brain by around 2030. The 10-20 year lag is due to the overhead of emulating hardware that is quite different than a computer. That's why I say the first strong AI will not simulate a human brain, but rather do what a human brain does in different, more computer-like ways.
No, it is the computational power. Last year, one-half of a mouse brain was simulated at one-tenth real-time. Can you imagine how slowly a full human brain would run in 2008? It would take years just to see what happens upon it getting basic sensory stimulus, far too slowly to get useful research out of it.
I know Kurzweil makes similar claims, but I am not merely regurgitating his ideas. The computational power of the brain is massive. In order to have similar computational power in a computer, we will need to wait for many generations of improvements in processors. We will almost certainly need to move away from very powerful cores based on semiconductors timed by a clock towards smaller asynchronous processors. Even with those advancements, there's no way we'll be able to simulate a human brain in real time by 2030. I think the first strong AI will resemble a human brain about as much as the Wright flyer resembled a bird.
1. It was the academic robotics teams that did well in the 2005 DARPA grand challenge (the top finishers were from Stanford and Carnegie-Mellon). Note that this also incorporated your #6, computer vision.
5. AI continues to improve at a steady pace. It's not that "nothing ever happened". For example, the DARPA Grand Challenge was won and Kasparov was beaten by Deep Blue. I think you might be referring to "strong AI" which we won't get until about 2030 because computers simply won't be fast enough until then.
I think if anything, people aren't realizing how quickly technology is advancing. This is the nature of exponential growth. I'm sure people will be dismissing AI and robotic research right up to when everything changes, seemingly out of the blue to most people.
Wikipedia has an article on the Ariane crash. If you think something in the article is inaccurate, you can always check the sources.
I suppose there may be some slight modification of local climate as heat is transferred from the desert to locations where electricity is used. Is that really worse than burning fossil fuels and chopping down rain forests though? Besides, there will be no global cooling effect, as the energy remains on the Earth, except for perhaps any generated light that escapes into space.
Yeah, I was sitting here on my ThinkPad T61 just now wondering what the issue is. No wonder I couldn't figure out what people were complaining about. Then I saw the screenshots of a glossy screen, and I have to say that totally sucks. Yeah, ThinkPad T61 may be the way to go. Mine is 15.4 inches widescreen at 1280x800 resolution (WXGA), but they go up to 1680x1050 (WSXGA).
Here's a simulator that doesn't require free registration yada yada yada.
For you to have two possible choices, you would have to know ahead of time which door Monty was going to open. If a stagehand whispered to you right before you went on "Psst... don't pick door 2 because Monty will open it to reveal a goat!" then you would always have a 50% chance of picking the car and there would be no advantage in switching. Your chances of winning increase in the original Monty Hall problem because if you're allowed to pick any of the three doors, and you pick a goat on your first choice (which you will do 2/3 of the time), you have forced Monty to open the only other door that has a goat, thus revealing to you which door has the car. That's why switching gets you the car 2/3 of the time.
The rational decision would be to take the check. The door I picked has an estimated value of $10000, less than the $12000 check. Monty didn't even have to open a door for me to make that decision.
If the goats and car do not move, then if you had a 1/3 chance of picking the car at first, how can merely opening a door change that 1/3 chance? It's only by rearranging what's behind the doors or picking another door than can change your chances of winning.
There's no debate. Of course when you first choose among three equally likely options, your chances of picking the car are 1/3. If you can't figure that out, you're completely lost before Monty even shows the goat!