Slashdot Mirror
The Future of Computers
GrokSoup writes: "Great collection of semiconductor where-to-from-here articles in this month's MIT Technology Review. There are articles about molecular computing, quantum computing, DNA computers, and on and on. Fascinating stuff, all pointing to why the current semiconductor hegemony is by no means a "forever thing", as the kids like to say. "
Is anyone else really sick of this crap?
"The problem lies in Heisenberg's Uncertainty Principle, which states that you cannot know the position and the velocity of a particle at the same time...
Almost, Heisenberg's Principle says you cannot know postion and velocity both to within a certain amount of exactitude.
"...looking at the position leaves you with a 'blur of possibilities' for the velocity and vice versa."
Sort of. Measuring the postion of a particle more and more precisely makes the velocity less and less certain (and vice versa). So it's not the looking at the position/velocity, it's the accurate measurement thereof.
--
Have Exchange users? Want to run Linux? Can't afford OpenMail?
Linux MAPI Server!
http://www.openone.com/software/MailOne/
(Exchange Migration HOWTO coming soon)
This argument comes up again and again, almost on an annual basis. However, there's a well know quote that i think applies here:
(Paraphrased) "640k should be enough for anyone..."
The point is, there is always something else that just around the corner. When Bill said the above quote, he hadn't envisioned the wide-spread adaoption of the GUI, 3d 1st person games, digital media (DVD's), or the World Wide Web. And just as Bill couldn't envision these things, neither can we envision what could be the Next Big Thing.
Now matter how fast you think your computer is now, someone, somewhere, is busy writing a peice of software that will make your computer seem like a Sinclair Z80 two years down the line.
Syllable : It's an Operating System
Click the link all you like. Just make sure you have disabled JavaScript, as any security-conscious person would.
-- Ed Avis ed@membled.com
As processor power increases, the difference between good and poorly written software is becoming more dramatic. This is a greater issue that's being dwarfed by pissing games over 1GHz+ speeds. For example, twenty years ago it was common for developers to use VAX-level minicomputers to do development for emergine home computers (Atari, Apple II, early PCs). After all, who could write a decent assembler on a 1MHz system with hardly any RAM? But it turned out that native assemblers on teeny tiny systems were frequently outperforming the minicomputers by a factor of ten or more.
Fast forwarding to today, consider any C++ compiler, say gcc or Visual C++. You can never have enough processor power for such a compiler. Even on top of the line machines you're still talking minutes to rebuild a medium sized project, and seconds for an average link. Now fire up Borland's Object Pascal compiler (buried inside of a RAD tool called Delphi). On a 200MHz machine the compilation time is up near a half million lines per minute. Link time is effectively zero. In general, compilation time doesn't exist. You have to have a pretty big project before you even notice that pressing F9 is taking any time at all. If you're using a 300MHz machine or faster, this is never even an issue. It's never an issue for almost any slower machine either, but I'm playing it safe. Now, yes, Object Pascal is simpler than C++. The compiler maybe doesn't do some of the nutty stuff that gcc does. But in the end, does it matter? A compile time of zero sure does make it easy to go in there and twiddle around with the code, making it go real fast.
This kind of thing is going to be more and more common. Is a bulky application slow because the processor isn't fast enough or because it is bulky? Throwing more processor power at problems like this is a dodge.
Am I the only one who knows how to do URL hacking? :)
If you want to see how the author did it (to see if its CGI or whatever), then go here:
http://hobbiton.org/~zk65/
There is a tar-zipped file there (which I haven't checked out) containing a do-it-yourself kit for ruinning Slashdot discussions.
The author was also kind enough to leave an e-mail address, so you can express how much you love this piece of work (heh).
-----
Obviousness is always the enemy of correctness. -- Bertrand Russell
I have a physics degree, and while I don't remember too much, I do remember this:
;)
Heisenberg's Uncertainty Principle refers to _momentum_ and position, and states that the product of the uncertainty in the momentum and position has a minimum value ( delta P * delta X >= h, where h is Planck's constant, P is a standard physics variable for momentum (no idea why), and X is the position). If you assume the mass is constant (disregarding Einstein's Special Relativity), then the momentum is proportional to the velocity, and what you said about velocity and position is roughly true.
There are several other pairs of physical properties which are related in this way; energy and time, angular momentum and angular position, etc. I'm not sure whether HUP is used to refer to these other property pairs, or only the momentum/position pair.
I'm sure I'm one of the few who care about such fine points, but I hate to leave misinformation out there unchallenged. I'm sure my fellow anal-retentives will thank me
A quantum computer works by trialling a superposition of wavefunctions or eigenfunctions (solutions of a matrix problem) as solutions of a problem (matrix operation). The resultant eigenvalues (conditions that show that the matrix problem is solved) are measured as observable results of an experiment (with probabilities gathered over a large number of experiments), and thus the eigenfunctions (solutions) are found.
Because every eigenfunction (of an infinte set) can be tested at once, over a certain number of trials, if the set of eigenfunctions (possible solutions) is infinite or very large, but the set of eigenvalues (actual solutions) is finite and smaller the problem is reduced from an infinite to a finite, or large to smaller problem and hence can be solved a lot quicker
That's misleading. Every particle has a single state, described by its wave function. It's just that the wave function doesn't necessarily correspond to a definite value of an observable quantity.
When you aren't looking at it, the state evolves deterministically according to the Schroedinger equation (classically) or Dirac's equation (if you take special relativity into account). For any given measurement you might make, most wave functions don't correspond to a definite value of whatever you are trying to measure: instead the wave function is a sum of states, each corresponding to states which do have definite values of the measurement. Given two observables, the wave function can be decompsed as a sum of states with definite values for either observable---but in general not both at the same time.
To expand on that last bit, many measurements are incommensurable. (i.e. the states that have a definite value for one measurement don't correspond to states that have a definite value of the other---this is the case with position and momentum in Heisenberg's uncertainty principle). Although the wave function evolves deterministically when you aren't looking, when you do look, it randomly collapses into a wave function that has a definite value of whatever you are trying to measure (i.e. into one of the component states for the observable you are looking at). The probability of collapse into a particular state depends on the extent to which that state was a component in the original wave function---dominant components are more likely. This, in a nutshell, is a formal description of how observations affect the state of the system.
Disclaimer: I'm a mathematician and not a physicist. This is my best attempt at explaining Von Neumann's approach to QM as I understand it, but it's simplified and I think physicists prefer different language. If I've messed up, I'm sure any physicists here will correct me.
I am trying to understand this. Is this story a review of techreview.com or an endorsement or even an admission along the lines of "aha! here are some guys talking about some really interesting stuff, you should go and look".
My point is, no disrespect to 'Taco, that this story looks to be the latter, and without any additional commentary or question or review of the linked content I think this is a poor and unstructured way to create a discussion.I mean, I now might have to read a whole other site - all of it - just to see what might be "on topic" or not :-)
In conection with "The end of Moores Law" I want to ask just how much economic value is lost in a system where what you paid for 18 months ago is now worth one quarter, 1/4, 25% of what you paid according to the simple interpretation of that dynamic?
I mean if your salary of 18 months ago were being lost to inflation at that rate, and the effect was spreading throughut the economy, we'd all be in Brazil during the 70's
for the record, Brazil inflated itself to growth on the back of vast loans, much of which originating from Citibank, and all of which collapsed in '82 leading to a massive world wide liquidity crisis.
The net effect of this was deliberately lowered interestrates post - Volker, and contributed to the junk bond (80's) and later equity (90's) booms we have become used toAs the effect of computer obsolescence and the consequent demand for capital and working capital (usually loans) permeates evermore deeply our economy (is this maybe one argument against PC vs big iron trend promoters?) this dependancy becomes more acute.
What I am saying here is that there may be *economic* reasons that bring to a halt technological phoenomena, and that a purely "geek" approach to the issues may yield diverging results from what is happening
Back to my original "complaint" I really think it would be nice is submitted stories came with something at least resembling an editorial viewpoint. Maybe /. is the only place where you can proverbially print your paper and leave the inside pages blank "for reader's notes". Maybe editors here want to avoid infered advocacy.
But surely the editorial point is "We here all have a (personal, financial) interest in the economics and technologlies developed under recent - very interesting and not necessarily fully understood conditions - is this set of linked articles a sign of things to come, or an indicator of impending (local systems) collapse?"
DO NOT CLICK THE "THIS IS MORE INFORMATIVE LINK"!
It's a link that sends a message from yourself to this page.....
Hmmm...I heard of this a while back, and thought it would have been fixed by now...
dylan_-
--
Igor Presnyakov stole my hat
First, IWAPMBHFMN (I was a physics major but have forgotten much now)
... in which an atomic nucleus can be spinning clockwise and counterclockwise at the same time
This is really a misnomer. It's a misinterpretation of intrinsic spin. Each particle in the nucleus possesses it, but the physical spinning (clockwise/counterclockwise) of the nucleus itself is not a quantum state. If it's a hydrogen atom, the nucleus is just a proton, so you can't really define a physical spin (in the rotational sense) although it will have an intrinsic spin. I think you can have an effective spin for a multiparticle nucleus, by summing up all the spins for the constituent particles, but I just made that up and it could be wrong. Labelling this peculiar quantity "spin" doesn't actually mean anything is spinning, in simply arises from the fact that particles appear to have angular momentum.
It is a bizarre world in which matter itself dissolves into a ghostly blur of possibilities as soon as you try to look at it.
Er... the Heisenberg Uncertainty Principle says that the error in position times the error in energy must always be greater than some minimum value. This means that the more precisely you measure ("look at") one quantity, the less precision you have in the other quantity at the same time. By that equation, if the error in one quantity approaches zero (meaning you're measuring near absolute precision), the error in the other quantity approaches infinity.
The leads to the old Quantum Physics homework excuse: "Professor, I calculated the energy in problem #3 so precisely that now I have no idea where in the universe my homework is!"
-CausticPuppy "Of all the people I know, you're certainly one of them." -Somebody I don't know
Quantum mechanics describes the behavior of particles exclusively in terms of a "psi function". The square of the psi function (imagine something like p2(x,y,z,t)) is the probability that a particle is a certain place at a certain time. It's referred to as the psi squared function.
But the theory only provides the probability. Obviously, when you look at the thing, you figure out where it is.
There are various interpretations of this. One is wierder than the next - my favorite is the idea that a new parallel universe is created for each possible observation, and each universe is identical except that the observer saw something else. But this is regarded as a little uneconomical, as you end up with a LOT of inherently unobservable universes (whatever a universe is). I swear I'm not inventing this.
But anyway, once you've seen the thing, the probability that it is where it is is one, and the probability that it is where it ain't is zero, so your psi squared function "collapses" to a boring function that is zero everywhere but one point, where it's one.
Please note that changing the psi squared function (obviously) changes the psi function too, and (as I already said) that quantum mechanics describes particles exclusively in terms of their psi functions. This means that every observation inevitably changes the particle itself.
If you can figure out exactly what makes the psi function collapse you get a Nobel Prize. Unless of course you're a grad student - then your professor does.
The Heisenberg Uncertainty Principle is a related but different idea.
I bragged about my Karma at a job interview but I didn't get the job.
'The Fabric of Reality' by David Deutsch. It doesn't cover up-to-the-moment technology, naturally, but it provides an ample explanation of how they work and a great deal of food for thought on their potential. It's well written too, but because of the concepts it tackles it will take most readers quite a while (well, it took me quite a while anyway).
"What is freedom of expression? Without the freedom to offend, it ceases to exist." Salman Rushdie
...at the same time, computers have reached a point where they can do everything we need.
While my old 80 MHz PowerMac lacks in voice recognition (it does have it, though) it makes a fine webserver and mailserver. I can still use it to browse the web and write papers. My roommate still uses a 200 MHz Pentium with Win98 and some newer games with no complaints.
One of the articles discusses that Moore's law is finally coming to an end. This might not be because it is harder and harder to make computers faster and faster as the article states. It may be because computers are finally doing everything we need them to. I know that having a smarter, better interaction with the computer will drive the industry for faster machines, but the proliferation of PDA's shows that we've passed the critical point of having enough computing power and not needing more. I mean that the computers are powerful enough now, that we are focusing on making them smaller and smaller.
Hell, today's PDA can wip me at chess. That's as powerful as I need!
--
I'm not a physics student, but I seem to remember that the possibilities only matter when you're NOT looking at the system. Does anyone else know whether this is mistaken or not?