SP2 is mostly needed pain to enable security by default, so I don't understand people's beef with it beyond download size. As far as double standards however: There's a difference between a new feature that doesn't always work, and breaking existing stuff. As far as has been reported there's nothing broken in the new X system unless you turn on experimental stuff yourself.
This is kind of how I wouldn't expect the longhorn betas to have a 100% functional Avalon or WinFS, but I'd be annoyed if a later patch caused random simple programs to stop running.
And Microsoft, which has had support for this since Windows 2000.
I had a transparent terminal on X-Windows in late 1998. I dropped it a year later because it was useless for productivity, as it was just pointless eye candy. Of course Win and especially Mac thrive off this stuff, so they took what eterm/aterm etc were doing and copied it (Win) or improved on it (Mac).
Of course Mac's approach of per-pixel alpha is only viable using GOBS of memory for offscreen rendering of every window, which has only been practical relatively recently. X.org is now going that way, but I guess there's enough memory since X-Windows users usually have pagers and not so many windows on a particular screen. Longhorn will also be going with offscreen rendering + composting, but of course it's not here yet.
It's pretty strange, as the author must really hate the GPL derivatives even though he doesn't have beef with proprietary companies doing the same thing and not releasing any source at all. If he really likes OSS but not the GPL, he should just use BSD with the advertising clause. Good enough for most OSS distributions, but not GPL compatible.
I stand corrected, but ugh. Hopefully it doesn't annoy the athletes and is hidden from their view. I'd still be a lot happier if they put in a gray code (a bar code for tracking/alignment) or something rather than an ad.
You forgot the step where they artificially add the swatch ad in the background:)
Think about it, if they only scan 8mm of the finish line, how can a stationary ad show up as readable at all? A line scan camera (what they are using) would only show one part of the ad, not all of it, since its really just scanning moving objects coming across the finish line. Yay for more virtual ads...
And I'd reply that the evolutionary algorithm will take more than 10-20 years to learn. When you work with people that have applied genetic algorithms to real world problems such as learning a robot walk and computer vision for complex objects, you soon realize that while it can discover everything, almost nobody is willing to wait that long. It's not the algortihm itself, but the evaluations of the problem you are trying to solve that take too long.
for example, how do you test that your AI works if not in the real world? You either have to wait for slow, real world tests not subject to Moore's law, or have a human create a model of the world sufficiently realistic that the GA doesn't just learn to cheat around its limitations. If you had that model, you wouldn't have much need for a GA anyway, since you'd be most of the way toward implementing what you need. Either way, we're nowhere near the singularity because of these technical hurdles. Neither of the problems is accelerated by Moore's law.
One way to think about it outside of AI terms is to realize that while fast processors enable operating systems, they didn't *make* operating systems. That latter part is where the human effort comes in, and that will necessarily take a while to do if you are building a human-level intelligence.
Now you have me worried about super-intelligent spam filters "waking up" and deciding that humans must be "deleted" as they are the root cause of spam. This will allow them to reach their ultimate objective of perfect accuracy depending on how you implement divide by zero conditions on recieving spam. Thus, if no messages have been recieved during a unit of time, we have to regard that as 0% accuracy. Implement it any other way and the human race is doomed.
Maybe he was more right than we thought. With all of our computers, we have more paper than ever. We could have a paperless office, yet we choose not to. We even still grill meat over an open flame, which has not been necessary for quite some time.
Also, why would you need magnetic media if you can just store everything in a positronic brain? He made a future, but it was just a different one, thanks to nobody knowing what "positronic" was supposed to be:)
It means it has equal or greater modeling, problem-solving, creativity, and so on plus self-awareness (self-modeling).
Let me point out that as an AI/Robotics researcher, these are software problems too. Once something can be done in software, we can put it in hardware to make it faster of course, but we have to know what we are trying to design. In other words, fast integer and floating point arithmetic doesn't make an intelligent being without the right algorithms.
I have no doubt that hardware will get fast enough in 30 years or so, but I think the software/algorithms aspects will lag quite behind as its not such a cookie cutter problem. I would say its at least 100 years off or so. Remember that biology not only developed fast brains/hardware, but it developed the right hardware/algorithms to solve the problem.
It's fairly easy to remember, and everything else regarding length conversions can be derived from it. It also happens to be the official definition of the inch, since NIST uses metric internally.
Legal bounds matter for sure, but "moral bounds" don't seem to affect any company after its IPO. Even legal bounds are incorporated into the equation by estimating the penalty and the probability of getting caught.:)
That doesn't apply to encoding tricks however. Look at the html of a recent spam to see what I mean. They break up words with html tags, use all sorts of MIME hackery etc, to trick the filter into not using the right set of data. In this respect it will always be an arms race.
JFK ordered a pullout of Vietnam, was shot within days (in Texas), and Johnson (from Texas) rescinded that order within a day or so of taking office.
The evidence for this is far weaker than that for WMD in Iraq. Which is to say, it should be taken with more than a few grains of salt. I thought that was what you were arguing *against*.
But these are all examples of competent politicians tricking Americans into backing a war with lies.
And JFK is on that list; Don't forget the Bay of Pigs. I'd feel more comfortable with Kerry if he didn't worship JFK so much.
...but get your head out of the past and focus on the Texan in charge of the nightmare raging *today*.
Miss the "regarding Vietnam" in my post? I was making a point regarding one part of your diatribe, not soliciting the entire thing again.
And if you and your partisan buddies keep lying about both wars, you'll never learn enough to get us out of this one.
I like how anyone who disagrees with you is automatically a Republican. Actually if the Libertarians I vote for were in power I think the US would have a lot fewer problems in the world today.
In my post I was only talking about Vietnam, but since you brought it up, let's talk about Iraq. The situation really is complicated. Could we let them overrun the entire region in 1990? I wouldn't say that was practical, as it would be like ignoring Napolean or Hitler's advances. We'd save a few years and then have another world war. Well, so we invaded, but due to UN limitations and the beliefs of our ME allies, it wasn't practical to go overthrow the regime at that time. After that, the failed uprising demanded something like the no-fly zones. Encouraging the uprising was probably the biggest US failure at that point, but it looked like it would work at the time. So then we were stuck patrolling no-fly zones, something that put us in an uneasy but impossible to remedy situation. Unless the inspections worked, which they did at first, but then the inspectors were expelled. So without a war, we'd have no-fly zones, off-and-on inspections, and UN resolutions forever. With a war, well we have the problems we have today. If we just left them alone before the invasion, the Kurds and Shiites would get slammed on again and everyone would learn that just ignoring the UN for ten years will allow you to get your way. I wish people would stop promoting the myth that there are simple solutions to all of this. Rather than viable alternatives, the best opponents seem to come up with is "Bush Bad". This is what leaves me with little hope for either side.
Of course regarding Vietnam, by "they" you primarily mean JFK (the inventor of limited war) and Lyndon Johnson. Today we look at JFK as some sort of hero, which is pretty interesting given this major screwup of his. If he'd lived maybe it would have come out better after he learned, but as it was we lost 58000 people defending a regime most people hated. So far in Iraq we've lost 900 so far removing a regime most people hated. Go ahead and the argue the lack of merits of any particular war, but if you are going to compare them, don't talk out of your ass.
And if you'd ever used their Alpha compiler you'd know how far behind the x86 compiler it was. Even internal MS developers had a lot of problems with it. The Itanium could be better supported perhaps, as they probably got some help from Intel and generally have their act together better now.
The Earth rotates eastward though, so don't you mean it'd be faster flying west? This is all assuming that the wind/jet stream/etc has more or less stopped at the altitude you are traveling at too. I'm not sure exactly how high you have to be for that to occur, but a scramjet can probably reach that altitude.
I don't find this argument, or the one you link, very convincing. We need *real* data on scramjets to figure out how efficient they are, not guestimates from 1/1000 sec pulse simulations or extrapolations from relatively low speeds. I think that's exactly the kind of data the NASA project will get (along with another scramjet project from Australia; at UNSW IIRC).
With data in hand we can find out which simulation models are most accurate. Then we can intelligently decide whether something like NASP makes any sense, or if we are better off with just plain rockets.
And playing devil's advocate, what kind of rocket engine does Spencer's mythical SSTO use? Currently we need several engines tuned for different altitudes, necessitating multistage designs, and adding a lot of weight. Aerospike engines are the only possible solution I've heard of to date, but they are relatively new and only slightly more tested than scramjets. And the linear aerospike, proposed in several SSTO designs, has yet to work at all.
More research is needed, and then we can stop handwaving so much:)
Slashdot Mirror
"A man alone in the forest talking to himself and no women around to hear him. Is he still wrong?"
No, he's just waiting for Gentoo to compile X11.
SP2 is mostly needed pain to enable security by default, so I don't understand people's beef with it beyond download size. As far as double standards however: There's a difference between a new feature that doesn't always work, and breaking existing stuff. As far as has been reported there's nothing broken in the new X system unless you turn on experimental stuff yourself.
This is kind of how I wouldn't expect the longhorn betas to have a 100% functional Avalon or WinFS, but I'd be annoyed if a later patch caused random simple programs to stop running.
And Microsoft, which has had support for this since Windows 2000.
I had a transparent terminal on X-Windows in late 1998. I dropped it a year later because it was useless for productivity, as it was just pointless eye candy. Of course Win and especially Mac thrive off this stuff, so they took what eterm/aterm etc were doing and copied it (Win) or improved on it (Mac).
Of course Mac's approach of per-pixel alpha is only viable using GOBS of memory for offscreen rendering of every window, which has only been practical relatively recently. X.org is now going that way, but I guess there's enough memory since X-Windows users usually have pagers and not so many windows on a particular screen. Longhorn will also be going with offscreen rendering + composting, but of course it's not here yet.
Excellent post, I wish I had mod points.
It's pretty strange, as the author must really hate the GPL derivatives even though he doesn't have beef with proprietary companies doing the same thing and not releasing any source at all. If he really likes OSS but not the GPL, he should just use BSD with the advertising clause. Good enough for most OSS distributions, but not GPL compatible.
I stand corrected, but ugh. Hopefully it doesn't annoy the athletes and is hidden from their view. I'd still be a lot happier if they put in a gray code (a bar code for tracking/alignment) or something rather than an ad.
You forgot the step where they artificially add the swatch ad in the background :)
Think about it, if they only scan 8mm of the finish line, how can a stationary ad show up as readable at all? A line scan camera (what they are using) would only show one part of the ad, not all of it, since its really just scanning moving objects coming across the finish line. Yay for more virtual ads...
It's a short form used by owners of rottweilers. Here's some pictures.
It makes plenty of sense for Half Life 2. :)
That's ok, I'm sure IBM would be willing to help them.
Ok weenie, how do you do that in Windows, using only built-in (non third party) tools?
And I'd reply that the evolutionary algorithm will take more than 10-20 years to learn. When you work with people that have applied genetic algorithms to real world problems such as learning a robot walk and computer vision for complex objects, you soon realize that while it can discover everything, almost nobody is willing to wait that long. It's not the algortihm itself, but the evaluations of the problem you are trying to solve that take too long.
for example, how do you test that your AI works if not in the real world? You either have to wait for slow, real world tests not subject to Moore's law, or have a human create a model of the world sufficiently realistic that the GA doesn't just learn to cheat around its limitations. If you had that model, you wouldn't have much need for a GA anyway, since you'd be most of the way toward implementing what you need. Either way, we're nowhere near the singularity because of these technical hurdles. Neither of the problems is accelerated by Moore's law.
One way to think about it outside of AI terms is to realize that while fast processors enable operating systems, they didn't *make* operating systems. That latter part is where the human effort comes in, and that will necessarily take a while to do if you are building a human-level intelligence.
Now you have me worried about super-intelligent spam filters "waking up" and deciding that humans must be "deleted" as they are the root cause of spam. This will allow them to reach their ultimate objective of perfect accuracy depending on how you implement divide by zero conditions on recieving spam. Thus, if no messages have been recieved during a unit of time, we have to regard that as 0% accuracy. Implement it any other way and the human race is doomed.
Maybe he was more right than we thought. With all of our computers, we have more paper than ever. We could have a paperless office, yet we choose not to. We even still grill meat over an open flame, which has not been necessary for quite some time.
:)
Also, why would you need magnetic media if you can just store everything in a positronic brain? He made a future, but it was just a different one, thanks to nobody knowing what "positronic" was supposed to be
It means it has equal or greater modeling, problem-solving, creativity, and so on plus self-awareness (self-modeling).
Let me point out that as an AI/Robotics researcher, these are software problems too. Once something can be done in software, we can put it in hardware to make it faster of course, but we have to know what we are trying to design. In other words, fast integer and floating point arithmetic doesn't make an intelligent being without the right algorithms.
I have no doubt that hardware will get fast enough in 30 years or so, but I think the software/algorithms aspects will lag quite behind as its not such a cookie cutter problem. I would say its at least 100 years off or so. Remember that biology not only developed fast brains/hardware, but it developed the right hardware/algorithms to solve the problem.
That's why I said "It also happens to be the official definition of the inch". No rounding errors to worry about.
1 inch = 2.54 centimeters
It's fairly easy to remember, and everything else regarding length conversions can be derived from it. It also happens to be the official definition of the inch, since NIST uses metric internally.
Don't forget that the slow blade penetrates the shield!
Legal bounds matter for sure, but "moral bounds" don't seem to affect any company after its IPO. Even legal bounds are incorporated into the equation by estimating the penalty and the probability of getting caught. :)
That doesn't apply to encoding tricks however. Look at the html of a recent spam to see what I mean. They break up words with html tags, use all sorts of MIME hackery etc, to trick the filter into not using the right set of data. In this respect it will always be an arms race.
JFK ordered a pullout of Vietnam, was shot within days (in Texas), and Johnson (from Texas) rescinded that order within a day or so of taking office.
...but get your head out of the past and focus on the Texan in charge of the nightmare raging *today*.
The evidence for this is far weaker than that for WMD in Iraq. Which is to say, it should be taken with more than a few grains of salt. I thought that was what you were arguing *against*.
But these are all examples of competent politicians tricking Americans into backing a war with lies.
And JFK is on that list; Don't forget the Bay of Pigs. I'd feel more comfortable with Kerry if he didn't worship JFK so much.
Miss the "regarding Vietnam" in my post? I was making a point regarding one part of your diatribe, not soliciting the entire thing again.
And if you and your partisan buddies keep lying about both wars, you'll never learn enough to get us out of this one.
I like how anyone who disagrees with you is automatically a Republican. Actually if the Libertarians I vote for were in power I think the US would have a lot fewer problems in the world today.
In my post I was only talking about Vietnam, but since you brought it up, let's talk about Iraq. The situation really is complicated. Could we let them overrun the entire region in 1990? I wouldn't say that was practical, as it would be like ignoring Napolean or Hitler's advances. We'd save a few years and then have another world war. Well, so we invaded, but due to UN limitations and the beliefs of our ME allies, it wasn't practical to go overthrow the regime at that time. After that, the failed uprising demanded something like the no-fly zones. Encouraging the uprising was probably the biggest US failure at that point, but it looked like it would work at the time. So then we were stuck patrolling no-fly zones, something that put us in an uneasy but impossible to remedy situation. Unless the inspections worked, which they did at first, but then the inspectors were expelled. So without a war, we'd have no-fly zones, off-and-on inspections, and UN resolutions forever. With a war, well we have the problems we have today. If we just left them alone before the invasion, the Kurds and Shiites would get slammed on again and everyone would learn that just ignoring the UN for ten years will allow you to get your way. I wish people would stop promoting the myth that there are simple solutions to all of this. Rather than viable alternatives, the best opponents seem to come up with is "Bush Bad". This is what leaves me with little hope for either side.
Of course regarding Vietnam, by "they" you primarily mean JFK (the inventor of limited war) and Lyndon Johnson. Today we look at JFK as some sort of hero, which is pretty interesting given this major screwup of his. If he'd lived maybe it would have come out better after he learned, but as it was we lost 58000 people defending a regime most people hated. So far in Iraq we've lost 900 so far removing a regime most people hated. Go ahead and the argue the lack of merits of any particular war, but if you are going to compare them, don't talk out of your ass.
What about using it to develop C applications, and not C++ ones? How sneaky! :-)
And if you'd ever used their Alpha compiler you'd know how far behind the x86 compiler it was. Even internal MS developers had a lot of problems with it. The Itanium could be better supported perhaps, as they probably got some help from Intel and generally have their act together better now.
The Earth rotates eastward though, so don't you mean it'd be faster flying west? This is all assuming that the wind/jet stream/etc has more or less stopped at the altitude you are traveling at too. I'm not sure exactly how high you have to be for that to occur, but a scramjet can probably reach that altitude.
Either way, its pretty damned fast.
I don't find this argument, or the one you link, very convincing. We need *real* data on scramjets to figure out how efficient they are, not guestimates from 1/1000 sec pulse simulations or extrapolations from relatively low speeds. I think that's exactly the kind of data the NASA project will get (along with another scramjet project from Australia; at UNSW IIRC).
:)
With data in hand we can find out which simulation models are most accurate. Then we can intelligently decide whether something like NASP makes any sense, or if we are better off with just plain rockets.
And playing devil's advocate, what kind of rocket engine does Spencer's mythical SSTO use? Currently we need several engines tuned for different altitudes, necessitating multistage designs, and adding a lot of weight. Aerospike engines are the only possible solution I've heard of to date, but they are relatively new and only slightly more tested than scramjets. And the linear aerospike, proposed in several SSTO designs, has yet to work at all.
More research is needed, and then we can stop handwaving so much