lclough wrote:
] ] Moore's Law, as chip manufacturers generally refer to it
] ] today, is coming to an end, according to a recent
] ] research paper.
] ...
] ] Resolving these issues is a major goal for the entire
] ] industry. Under Moore's Law, chipmakers can double the
] ] number of transistors on a given chip every two years,
] ] an exponential growth pattern that has allowed computers
] ] to get both cheaper and more powerful at the same time.
]
] Moore's law hits a wall at about 2021, when fundamental
] physics prevent devices from becoming any smaller. After that
] it will
] be ncessary to take other steps (larger chips, 3D chips?) to
] continue the performance increase.

I'm not sure if you've heard of adiabatic computing, but here's a wired.com link about it which might affect moore's law: http://www.wired.com/news/technology/0,1282,61118,00.html
Hugo de Garis called reversible (adiabatic) computing "the greatest of the scientific discoveries of the 20th century."
Actually, they mention adiabatic computing in the link you provided in this paragraph:
"One extremely theoretical potential idea is to reuse electrons. In current architectures, electrons travel from a source to a drain and then are destroyed. With recycling, "you simply transfer the electron to something else," Gargini said. "You can make a lot of calculations without destroying the electrons." "

Hugo De Garis also predicts computers will become "trillions of trillions" of times more powerful than the human brain using molecular electronics, a subfield of nanotechnology. Actually, I think he mentioned using quantum mechanics to do this or something, but I can't make heads or tails of quantum mechanics. When first starting to read about it, I almost dismissed it as some strange meme in science, an idea that was successful at replicating whether or not it was credible.

norfzorf wrote:
] I'm not sure if you've heard of adiabatic computing, but
] here's a wired.com link about it which might affect moore's
] law: http://www.wired.com/news/technology/0,1282,61118,00.html

Very cool technology... (No pun intended. :)

The recent slashdot discussion on this annoyed me. The assumption that this particular scientific vein will be mined for improving power forever is the misconception. Intel has warned and warned about this for years. The date hasn't moved. Moore's law will end. This doesn't mean that computers are going to stop getting faster, but that the rate at which they will get faster will decrease. The engine of Moore's law has had far reaching impacts on the rate of economic growth in the past 20 years. If we wish to sustain such growth we must find a new engine in time. Otherwise the speed of computers is not the only thing thats going to stagnate.

As need is the mother of invention, one can predict with reasonable safety that nothing will be done until the situation is dire. There may be an economic recession around the time that Moore's law stops. If you are ready you can manage your investments appropriately. Conversely, one can expect short "y2k" like bursts in computer/telecom spending around the time that the IPv4 space runs out, and the time when UNIX clocks stop working. All of this is less then 30 years out.

] Hugo De Garis also predicts computers will become "trillions
] of trillions" of times more powerful than the human brain
] using molecular electronics, a subfield of nanotechnology.
] Actually, I think he mentioned using quantum mechanics to do
] this or something, but I can't make heads or tails of quantum
] mechanics.

Its nonlinear. Its not a standard computing architecture. It solves NP in P. Trouble is its extremely unstable. Like a truck driving 50 miles away will destroy it unstable. Whether that is a solvable problem is unknown.

Cool thing is that we are starting so see quantum crypto systems on the market. They really aren't all that interesting from a security standpoint, but improving the technology will eventually allow information to be transmitted across unlimited distances instantly. (Yes, I mean faster then the speed of light.) There are practical limits to the bandwidth of such systems, but there could be some interesting applications. (However, the first applications that come to mind are military... Cache enough tied photons in a remote UAV to allow the final seconds of targeting and weapon firing to be controlled in realtime for increased accuracy.)

Decius wrote: ^

My current understanding and knowledge of computers isn't great enough for me to fully understand the phenomena and mechanisms that occur deep behind the layers of functioinal abstraction. The most interesting things that I have read though are about massively parallel machines which perform many operations simultaneously, like daniel hillis's connection machine, and of course using nature's spectacular magic massive search algorithm, evolution, to design computers.

I remember reading de garis mentioned that the "trillions of trillions" of times smarter computer would need to be about the size of an asteroid. His simile was that these "artilects" would be to us, in terms of advancement, as we are to a single celled organism, possibly indifferent to our pathetic meaningless fate. The only way I can imagine these machines having any "desire" to do anything would be if they were created with artificial evolution and thus their core architecture would be made of replicating entities that were selected just by somehow being successful at replicating. This would possibly make the artilects evolve "curiosity"; maybe curiosity would be fit. What I mean is that it seems if a super intelligent asteroid conglomerate without replicating entities (evolution) at it's core, would just float around not caring about anything. It would have no inherent ambition. The only reason people bother doing stuff is they have evolved a sense of pain and reward, ie for getting injured or mating respectively, which were successfuly replicating geneplexes. It might be extremely immoral if massively intelligent computers created in the future with artificial evolution evolve a sense of pain (as it would maybe make them more fit), perhaps unprecedented pain in proportion to their unprecedented intellects.

An artilect might be built with or evolve it's own phenotypic telescopes to observe the universe but why would it "want" to do anything, such as travel to other planets if it didn't have an innate urge to replicate? It wouldn't have an underlying construction which made it ask questions or care about answers because it wouldn't have evolved a sense of curiosity. If an artilect is just floating around up in empty space though, there is really quite a lack of a stimulating environment. There's no sound in space so all the environmental input of 1's and 0's or whatever would probably be from light from stars and planets. It could be that such artilects would already have an incomprehendably large knowledge base and would just sit there playing around with it in their own "heads". The lack of a stimulating environment wouldn't seem to provide selective pressure... unless the artificial evolutionary mechanisms operate in some bizarre internal way where somehow selection occurs without the need of an environment. It's not like they're swinging around in trees engaging in complex social intercourse. Or maybe the necessary environment is contained w... [ Read More (0.4k in body) ]

norfzorf wrote:
] I remember reading de garis mentioned that the "trillions of
] trillions" of times smarter computer would need to be about
] the size of an asteroid. His simile was that these "artilects"
] would be to us, in terms of advancement, as we are to a single
] celled organism, possibly indifferent to our pathetic
] meaningless fate.

I meant to mention this in my last post. I hate it when AI people get on their high horse and say that with n many transistors xyz computer would be more powerful then the human brain. We don't understand how human brains work, so how can we claim to be building something that does the same stuff? I'm fairly certain that the human brain does not have an ALU at the heart of it. In fact is seems that humans are particularly bad at the sort of "thinking" computers do well, and vice versa.

They only way we can presently be assured of building something that works the way we do is to evolve it, because we don't know what we're trying to make. We'll probably understand how it works before we manage to evolve a replacement in silico, and if we did evolve a replacement, we could never be sure that we really had what we were looking for. (The turing test is an interesting philosophical game but if you consider it a real way to measure AIs you don't understand its meaning. The guy was going insane when he came up with it. Restate it as matrix will be successful when you can't tell the difference between it and the real world. Your inability to distinguish the two doesn't mean the difference is immaterial. It is as much a measure of your blindness.)

] "The “human striving” argument arises from the fact humans
] always seem to want to go beyond what is currently known,
] currently explored, currently achievable. Humans drive
] themselves to climb higher peaks, run faster, cure diseases,
] become stronger and fitter, become more brilliant, etc., etc.
] Why this constant pushing at the barriers? It must be built
] into our genes. Evolution has made us this way... If we have a
] better knowledge of the dangers and delights of the world that
] surrounds us, then we are more likely to survive"

I found your discussion insightful...