Get a neighbour and allocate each other a quota on each other's boxes. Write a script to backup to a file. Encrypt them with gnupg. Transfer the files using any one of half a dozen protocols over the wireless LAN.
John Carmack might be a kick-arse game programmer and a very smart guy, but he is not an expert compiler designer, complexity theorist, or, as is most relevant here, embedded systems programmer for safety-critical systems (though I'm sure he's rapidly learning about it with his rocketry hobby).
Agreed, probably dumb pr person
on
Awari Solved
·
· Score: 2
I don't quite understand why a big lookup table is an important step for AI.
Agreed. It's not important at all. It's probably just some pr flack who was looking for an interesting angle and didn't actually check with the researchers who did the work.
As for what "the AI community" are doing, might I suggest that they *are* trying a whole bunch of different approaches. None seems to offer the holy grail yet.
Not knocking Gentoo, but Debian has been doing a comprehensive volunteer distribution for an extremely long time, and demonstrated its ongoing viability through leadership changes and occasional political turbulence.
Gentoo, though by all reports I've heard it's excellent, is a much newer project.
In my experience, a lot of the people who go into commerce-related degrees are there obtain a piece of paper, and subsequently to obtain well-paying jobs and the power to order other people around in management. In short, they were greedy well before they entered the course. Trying to teach ethics to this kind of person is a waste of time. They'll learn to talk the talk, but that's about it.
However, this doesn't apply to everyone. A fair number of high school graduates come to college with a narrow education and life experiences, and opening their eyes to different perspectives on ethics can only be a good thing.
My other comment is that an ethics course would be a hell of a lot better use of time in a business course than a programming subject would be, and plenty of people try to teach programming to business people.
I presume their forensic people don't have to be crack shots and marathon runners. Why require the same for people doing what are essentially "computer forensics"?
What about tattooing a "reference chart" next to the actual sugar-sensitive tattoo as a rough guide, so that you could monitor major changes by just glancing at the tattoo? Maybe even arrange it so that certain segments glow as threshold levels are reached (kinda like a battery gauge type of thing)?
Or, even, tattoo it to your wrist and have a colour sensor in your watch that started bleeping if your sugar levels changed too much.
I agree that those three episodes were really, really good, but they're getting fewer and further between - just like the Simpsons, which can still be incredibly funny on occasions.
In my opinion, the show is running out of dramatic territory to cover, and it would be nice if it was allowed to die gracefully while we're still watching instead of die a miserable near-ignored death like so many fine shows.
Back in the old days, the only speed limit out there was what speed you were prepared to risk (it wasn't even sealed till the 1970s). Nowadays, whilst the road is quite good and gunbarrel-straight, it is policed by Australian cops who generally have a zero tolerance towards speeding. Backroads, you ask? There aren't any out there. There isn't anything except the road out there. "Nullabor" apparently translates as "Nothing" from the local Aboriginal dialect...
If you *do* want to experiment with warp factor nine, there *are* plenty of backroads in other areas of Australia, or there is the Adelaide-Darwin road. Once you cross the Northern Territory border, there's no speed limit whatsoever:)
There's no government funding for religious activities (except some welfare programs run by churches, which receive government funding as providers of welfare services).
The one major exception is private schooling (often run by religious organisations), which receives government funding here. However, that's based on enrolments to the schools, not census data.
Imagine you're at the pub (perhaps that pub on Glenferrie Road just up from the Swinburne campus). The jukebox is blaring, you're having a great time with some of your mates, and you want to tell somebdy else where you are so they can join you. They're not at home, so you can't ring them there. The bar is pretty noisy, so even if you rang their mobile you'd have to go outside. Instead, you spend 30 seconds sending them an SMS. 22 cents later, they know where you are.
I dunno about you and your friends, but for my friends it's a major way we make sure we meet up on a Friday night.
It may or may not be useful for a disabled person, but (even after practice) I think it's slower than predictive text on a mobile phone keypad and handwriting recognition on a PDA. Therefore, I'm not sure whether there will be any useful applications for the able-bodied.
The OpenBSD project's FTP server doesn't run on OpenBSD, so the details of how the hack happened aren't that relevant to OpenBSD's security.
Read the OpenBSD FAQ for the details of why the FTP server isn't an OpenBSD box, but IIRC it's basically because it's a donated box and bandwidth from a university, and beggars can't be choosers.
In most of the world, it's much cheaper to send an SMS message than it is to make a call - here in Oz sending an SMS costs you 20c, whereas calls cost a minimum of 40c for a 30-second call (and most plans are more expensive).
You can send and receive SMS messages in a noisy environment such as on public transport or at a bar. This is rather useful to the young, urban and single:)
In Australia (and I gather everywhere else where mobiles are popular except the US) SMS is regarded as an essential feature by just about anyone under 30.
Voyager is not travelling all that fast, and we could go faster with sufficient time and engineering effort.
First cab off the rank is probably the Orion drive. Build a really big plate, attach it with really big springs and dampers to a heavily radiation-shielded spacecraft, and detonate atom bombs behind the plate. The basic technology exists right now, all you need is a pile of cash and be prepared to violate the space weapons treaty. Maximum speed is about 1-2% of the speed of light, so you're still taking a couple of centuries to Proxima Centauri.
Next option is a fusion engine. We can't generate power with controlled fusion yet, but ITER probably will if and when it gets built. ITER is, er, rather large and heavy, and doesn't really produce much net power, so a practical space fusion power plant is a fair bit of engineering development down the road. Anyway, the idea is quite simple. Release the "exhaust" of the reaction out the back of the engine, just like a normal rocket except the exhuast is a hell of a lot hotter and travelling a lot faster. Maximum speed maybe 10-12% of the speed of light.
Alternatively, use a light sail powered by a really big laser. All you need is to scale up laser and telescope technology a crapload (so, again, considerable engineering development required). Maximum speed? Somewhere between 10 and maybe 30% of the speed of light, depending on just how big you can make your mirror (and consequently how far you can keep accelerating).
The other big issue with interstellar spacecraft is the question of how much debris is out there. If there's a lot, as you go faster you'll need one hell of a shield to protect you.
Finally, there's there's also the possibility of using antimatter-matter reactions to power a ship. Antimatter is kinda powerful stuff to have around, and you could theoretically use it to power a ship to near the speed of light. However, there is no known natural source, and manufacturing it requires milllions of times more energy put in than you get back when you "burn" it. It, therefore, is a really long-term option from when humanity has such astounding energy generation capacity it can afford to use it to power antimatter-powered spaceships.
All in all, there are some possibilities, but most are still a fair bit of technological development away. Let's get to the rest of the solar system first:)
As others have pointed out, it's 51% government owned. The current government wants to sell it, and is trying to persuade its coalition partners and non-aligned parties in parliament to support the sale.
In practice, it's operated like a private monopolist for about a decade now.
I read the article, and I'm wondering how Personal Telco can afford to provide access to two T1's for free. Last I heard, that kind of high-quality bandwidth still doesn't come cheap.
I'm not familiar with the minefield problem. As far as the TSP goes, no, quantum computers of the types proposed so far haven't been any help. As the TSP is an NP-complete problem, it would be extremely big news if quantum computers could help.
I have heard suggestions that alternative designs for quantum computers would theoretically be able to tackle the TSP (or indeed any NP-complete problems) but from what little I know about the area I don't believe anybody's come up with a vaguely plausible way such a computer might be constructed.
Aside from Shor's factoring algorithm, there is also Grover's searching algorithm that lets you do linear search in slightly less than linear time, and apparently an algorithm for doing quantum physics simulations (surprise surprise). That's it. Three algorithms which work better on a quantum computer. None of them seem much use for Cyc at first glance.
Penrose's assertion that human intelligence is non-algorithmic may or may not be true; however, most mathematicians and logicians who have studied his proof think it's hopelessly flawed.
As for the use of quantum computers in AI - at present, nobody has provided an example of a vaguely AI-related problem that quantum computers of the type currently being studied would be useful for. Somebody may do so in the future, of course. In any case, anything that can be done on a quantum computer can be simulated on a normal one (in a theoretical sense, it may take till the end of the universe to do so). They don't give you the ability to compute anything "non-algorithmic".
Maybe you're joking, but in case you aren't, gigaflops are next-to-useless also. Firstly, it's talking about floating point operations, which are highly relevant if you're doing 3D stuff, weather forecasting, or the like, but totally irrelevant if you're, say, using GCC to compile Mozilla.
However, even if you are doing floating-point work, just because one machine has a higher quoted gflops rating than another doesn't mean it'll be faster on the relevant app. That's because quoted gflop figures are usually "peak" figures. In practice, to get that peak figure, your program needs to be able to keep each FP unit of the CPU active at all times, which rarely occurs because of the limitations of the problem, the compiler, and the memory bandwidth of the system. So, in many cases a machine with a lower peak figure may actually be much better in practice because it can use the capacity it has a higher proportion of the time.
As always, the only benchmark that really matters is the performance on the app you run on your own data.
Of course, people are quite correct to point out that this kind of tactic is common in business. It's still stupid for governments to fall for it. Public tendering is the way to go for any big government purchase - heck, for any big purchase.
Slashdot Mirror
Get a neighbour and allocate each other a quota on each other's boxes. Write a script to backup to a file. Encrypt them with gnupg. Transfer the files using any one of half a dozen protocols over the wireless LAN.
John Carmack might be a kick-arse game programmer and a very smart guy, but he is not an expert compiler designer, complexity theorist, or, as is most relevant here, embedded systems programmer for safety-critical systems (though I'm sure he's rapidly learning about it with his rocketry hobby).
Agreed. It's not important at all. It's probably just some pr flack who was looking for an interesting angle and didn't actually check with the researchers who did the work.
As for what "the AI community" are doing, might I suggest that they *are* trying a whole bunch of different approaches. None seems to offer the holy grail yet.
Gentoo, though by all reports I've heard it's excellent, is a much newer project.
However, this doesn't apply to everyone. A fair number of high school graduates come to college with a narrow education and life experiences, and opening their eyes to different perspectives on ethics can only be a good thing.
My other comment is that an ethics course would be a hell of a lot better use of time in a business course than a programming subject would be, and plenty of people try to teach programming to business people.
For those of us not in the states, could you please provide some mor details on what this system does?
I presume their forensic people don't have to be crack shots and marathon runners. Why require the same for people doing what are essentially "computer forensics"?
Or, even, tattoo it to your wrist and have a colour sensor in your watch that started bleeping if your sugar levels changed too much.
In my opinion, the show is running out of dramatic territory to cover, and it would be nice if it was allowed to die gracefully while we're still watching instead of die a miserable near-ignored death like so many fine shows.
If you *do* want to experiment with warp factor nine, there *are* plenty of backroads in other areas of Australia, or there is the Adelaide-Darwin road. Once you cross the Northern Territory border, there's no speed limit whatsoever :)
The one major exception is private schooling (often run by religious organisations), which receives government funding here. However, that's based on enrolments to the schools, not census data.
I dunno about you and your friends, but for my friends it's a major way we make sure we meet up on a Friday night.
:BTW, where's the astronomy department?
It may or may not be useful for a disabled person, but (even after practice) I think it's slower than predictive text on a mobile phone keypad and handwriting recognition on a PDA. Therefore, I'm not sure whether there will be any useful applications for the able-bodied.
Read the OpenBSD FAQ for the details of why the FTP server isn't an OpenBSD box, but IIRC it's basically because it's a donated box and bandwidth from a university, and beggars can't be choosers.
AFAICT, most of the slowdown is doing net lookups of spam listings.
In Australia (and I gather everywhere else where mobiles are popular except the US) SMS is regarded as an essential feature by just about anyone under 30.
First cab off the rank is probably the Orion drive. Build a really big plate, attach it with really big springs and dampers to a heavily radiation-shielded spacecraft, and detonate atom bombs behind the plate. The basic technology exists right now, all you need is a pile of cash and be prepared to violate the space weapons treaty. Maximum speed is about 1-2% of the speed of light, so you're still taking a couple of centuries to Proxima Centauri.
Next option is a fusion engine. We can't generate power with controlled fusion yet, but ITER probably will if and when it gets built. ITER is, er, rather large and heavy, and doesn't really produce much net power, so a practical space fusion power plant is a fair bit of engineering development down the road. Anyway, the idea is quite simple. Release the "exhaust" of the reaction out the back of the engine, just like a normal rocket except the exhuast is a hell of a lot hotter and travelling a lot faster. Maximum speed maybe 10-12% of the speed of light.
Alternatively, use a light sail powered by a really big laser. All you need is to scale up laser and telescope technology a crapload (so, again, considerable engineering development required). Maximum speed? Somewhere between 10 and maybe 30% of the speed of light, depending on just how big you can make your mirror (and consequently how far you can keep accelerating).
The other big issue with interstellar spacecraft is the question of how much debris is out there. If there's a lot, as you go faster you'll need one hell of a shield to protect you.
Finally, there's there's also the possibility of using antimatter-matter reactions to power a ship. Antimatter is kinda powerful stuff to have around, and you could theoretically use it to power a ship to near the speed of light. However, there is no known natural source, and manufacturing it requires milllions of times more energy put in than you get back when you "burn" it. It, therefore, is a really long-term option from when humanity has such astounding energy generation capacity it can afford to use it to power antimatter-powered spaceships.
All in all, there are some possibilities, but most are still a fair bit of technological development away. Let's get to the rest of the solar system first :)
In practice, it's operated like a private monopolist for about a decade now.
I read the article, and I'm wondering how Personal Telco can afford to provide access to two T1's for free. Last I heard, that kind of high-quality bandwidth still doesn't come cheap.
I have heard suggestions that alternative designs for quantum computers would theoretically be able to tackle the TSP (or indeed any NP-complete problems) but from what little I know about the area I don't believe anybody's come up with a vaguely plausible way such a computer might be constructed.
Aside from Shor's factoring algorithm, there is also Grover's searching algorithm that lets you do linear search in slightly less than linear time, and apparently an algorithm for doing quantum physics simulations (surprise surprise). That's it. Three algorithms which work better on a quantum computer. None of them seem much use for Cyc at first glance.
As for the use of quantum computers in AI - at present, nobody has provided an example of a vaguely AI-related problem that quantum computers of the type currently being studied would be useful for. Somebody may do so in the future, of course. In any case, anything that can be done on a quantum computer can be simulated on a normal one (in a theoretical sense, it may take till the end of the universe to do so). They don't give you the ability to compute anything "non-algorithmic".
However, even if you are doing floating-point work, just because one machine has a higher quoted gflops rating than another doesn't mean it'll be faster on the relevant app. That's because quoted gflop figures are usually "peak" figures. In practice, to get that peak figure, your program needs to be able to keep each FP unit of the CPU active at all times, which rarely occurs because of the limitations of the problem, the compiler, and the memory bandwidth of the system. So, in many cases a machine with a lower peak figure may actually be much better in practice because it can use the capacity it has a higher proportion of the time.
As always, the only benchmark that really matters is the performance on the app you run on your own data.
Of course, people are quite correct to point out that this kind of tactic is common in business. It's still stupid for governments to fall for it. Public tendering is the way to go for any big government purchase - heck, for any big purchase.
here. Still have grave reservations on using it on anybody competent enough to understand what it is.