The balance between speed and accuracy has to depend on your goals: * Goal: sell papers; priority: speed with reasonable accuracy * Goal: get attention; priority: speed with at least a little accuracy * Goal: impeachment; priority: accuracy in time for the election >;)
Seriously, though, different mixtures of speed and accuracy appeal to different audiences. I don't expect that Drudge has a large following among CPA's, for example. And for the masses? Most of us, most of the time, want to be entertained without being fooled -- which means that your profs were right:)
With the importance of the Internet and the fragility of some popular software (*cough*Windows *cough*Outlook), you could make an argument that these apply directly to software.
Does your OS interfere with other computers? Well, if a wide open hole allows whole sections of the net to go down, yeah, this widget interferes with other widgets. It may be difficult to burn down the house right now, but just wait until everybody has their home controlled by Windows Longhorn HVAC edition. Or what if a phreaker brings down 9-1-1 through an open firewall port? Credit card fraud and identity theft are common enough and dangerous enough to, possibly, be worth protecting against.
I'm a developer, I don't look forward to government edict making us even less competitive. But I also have to use the darn things, and from that perspective I'm as frustrated as anyone. Now if we could only get a "Software Quality" approval board *not* controlled by Microsoft, Oracle, etc, etc.
Maybe you and I and other normal people won't use it, but you can bet younger folks will. When the next hot boy band is only available on FingerNailChip (tm), it'll sell. If today's teenagers are like they were 20 years ago, "oldies" music is from two months ago.
There are reports that the first people who heard a phonograph thought there was an actual orchestra. They found the effect *real* enough, even though it would sound scratchy, fuzzy, and fake to us today. As we become more familiar with a technology, our expectations go up. It is possible to spend $10,000 or more on a stereo system and still complain that it just isn't the same as live music.
We tend to project and fill in the details: finding shapes in clouds, seeing a face on the moon and on Mars. Maybe we start by filling in the details and then get more sophisticated (or lazy) and expect the technology to do more of the work.
On the other hand, realism is more than just "making it look real". You could argue, for example, that The Simpsons is more "real" than Leave it to Beaver. It certainly is a more accurate portrayal of modern attitudes.
I know your post is talking specifically of kind of a "Turing Test" for CG: can you tell which portions are CG and which were not. I am just continually fascinated by the way humans work.
As a humorous aside, it is ironic that we continue to raise our standards for CG in movies while our expectations for human actors seem to be in serious decline:)
In this case, they suspected the guy anyway, but couldn't prove it. They used the DNA to decide if he should be arrested. Why didn't they bring the guy in and get a new sample? I think that, if they couldn't legally get that done, they should not have used DNA that was collected for something else.
The real concern (which we didn't see happening here) is when the DNA database is done for searches for possible hits without doing other research first. This gets too close to illegal searches in my opinion.
Also, the error rate on matching is pretty low. But what happens to a low error rate when you have the entire population of the US or the world in a database? And what happens to "innocent until proven guilty" if your DNA matches?
Not trying to be flip... but how do you design an AI without any reference to ethics? Dictionary.com defines ethics like this:
1. a. A set of principles of right conduct.
b. A theory or a system of moral value
Couldn't you stretch and say that any defined goal has some idea of right and wrong? For a stock trading program, profit=right, loss=wrong. For drug testing programs, dead customers=wrong, live customers=right.
Okay, to a logical person it's a ridiculous stretch. But recent history has shown that lawyers, juries, and the US patent office often operate outside of "normal" logic and common sense.
Thanks for the explanations of the age of the universe. But how do they know how old the planet is? I have been looking through the article and the responses and I don't see a complete explanation.
The best I can figure is that it's age is calculated based on (1) its composition and (2) its location/neighbors/star. Is there more or is this it?
Choice (2) seems kind of suspect. I could move into a nursing home but it wouldn't make me old (although I'm uncomfortable close:). It could make decent corroborating/additional evidence, though.
The composition is more solid evidence, but I would think there would be room for interpretation there, too. We have direct physical access to a tiny speck of matter in a huge universe. How do we know that there aren't local deposits of these elements that could have formed a planet much later than they think?
I'm not fighting their calculations, I'd just like to understand them. And to understand how certain they are of their number. Even if it's only 80% sure, it's still an intriguing discovery.
Okay, we've beat to death the fact it was a pre-release version. But look at it this way:
When Open Source software is about the same quality as closed source, the developers consider it unstable and warn people that they may run into problems.
It shows a big difference, to me, in the quality standards that OSS developers (and users) expect.
I think there is a lot of potential for professional-grade installations using something like Ardour. The tool itself is open-source and it runs on an open-source OS. Don't think it runs fast enough? Reconfigure your kernel by removing all the junk you don't use. Go through the startup scripts, turning off services and daemons you don't need. Switch to one of the more real-time oriented versions of Linux. Hack the source code, optimize the drivers, add your own features.
Okay, realistically, I'm not going to do this. I will plunk on my guitar in my basement, make a couple of mediocre tracks, and move on with my life.
But's what's stopping you from doing this? Or a technically savvy sound studio owner? Or, cooler than that, you could make a product by combining a rack mounted PC, a custom-configured version of Linux, a custom-configured version of Ardour, and the testing and support that would be needed. Of course, the GPL would require you to make the source code available. But that's okay. You get paid for your expertise without taking anything away from those who provided the expertise to build Ardour or Linux.
Whew, ranting a bit. Sorry. But I am excited by the possibilities.
There are a lot of people, myself included, who think that the problem lies with the way software patents are researched and granted in the US, not necessarily with software patents by themselves. Not to mention the fascination US companies have with taking everything to court.
IANAL, but I believe there are at least two main tests that should be applied before granting a patent: (1) it should be new, and (2) it should not be obvious.
The first test looks at "prior art": is this a common practice or is this really novel and unique? Many software patents have been given for things that are very commonplace: the online "shopping cart", using XOR to draw a cursor on a graphics screen, and now (possibly) caching Internet content. The Patent Office seems to be doing a particularly poor job of making sure that software techniques being patented are actually new and different. It seems to stem from a huge number of applications coupled with a fundamental lack of understanding about software.
The second test, that it is not obvious, is a little more confusing, at least to me. Things that are obvious to programmers probably aren't obvious to, forgive me, normal people.
Having said all that, I'm sure that some people disagree with software patents altogether. I'll let them explain their point of view.
For what it is worth, I treated the SalesDroid respectfully, nodded my head, didn't argue, and said "Thank You". I was actually impressed by his knowledge and the way he explained it (although I admit this isn't clear from my first post). I didn't even come *close* to "Busting the guy's balls".
My fun came at Intel's expense -- they are intentionally watering down one of their own "proofs" that their chips are better than what Apple is using.
I was helping my Father-in-law shop for a laptop for Father's day. Trying to cause trouble, I asked a salesdroid at BestBuy why the new Centrino chip was so much slower than the P4. He gave me a song and dance about optimization, integration, blah, blah, and ended up telling me that clock speed is not that important for overall speed.
It seems like Intel is now playing both sides of the Megahertz gap. This could be a good thing for Apple. If Intel wants to sell the Centrino they will have to convince the public that the "slower" chips are "faster" or "better" and will give away much of their hype advantage. The Centrino tops out at something like 1.6 GHz, well below P4 and G5 clock speeds.
(This info comes from a friend who generally tells me the truth, so I'm giving secondhand info that I'm pretty sure is accurate.)
Small Claims Court is useful for two decent people to get an unbiased answer to a small dispute. The problem is that, in New Hampshire at least, there is no assistance for collecting judgements. It is up to the decency of the losing party to pay up. If they don't, well, you spent half a day in court with nothing to show for it.
In my friend's case, he was ripped off by some contractors building his house. He eventually got some sort of satisfaction by tying up their time in court and hurting their reputation. They eventually paid because he made himself a royal pain in the butt.
I imagine (pardon the metaphor) that Universal Records has a much bigger butt.
I don't see any realistic alternatives, either, but I wish I did. I think that the desktop metaphor helped us immensely at first but is now holding us back.
Some alternatives have been tried and rejected. Examples:
1) IBM (and others) tried to make on-screen objects look and act like real objects.
Real CD player and RealPhone 2) The desktop can be expanded to a virtual office building, with file cabinets in a virtual file room. This doesn't work because it is really the same as a desktop, except with the added advantage of seeming beaurocratic and unfriendly.
3) Virtual reality was destined for greatness as an interface. Maybe there's hope as computers get more and more powerful. Or maybe people don't want to wear goggles and sci-fi wired gloves just to check their email.
I think there are some good reasons that the desktop is still alive:
1) It works pretty well. Not perfect, but good enough for most things most of the time.
2) It is entrenched in our thinking.
3) Inertia and legacy code.
4) Coming up with good, useful, extendable metaphors is difficult.
So, the desktop lives on. Maybe the real breakthrough will be actual physical devices that can hide the computer underneath so they're just "smart", not computers.
The worst part of the Windows desktop? It lives in more than one place. It is the "root" or "base" of the system (from the GUI), but it is also a folder buried under the Windows directory on one of the hard drives (from the command line or file browser).
With more than one drive, the drives exist as (1) icons inside "My Computer" inside the desktop, (2) top-level directories (Windows has no real "/" in the Unix sense).
This schizophrenic desktop and drive setup can cause major confusion. It confused me (still does) even with 15+ years of computer and programming experience. Why on earth would they implement it that way? And why did the KDE/Gnome developers copy it?
On a Mac, the "desktop" is your whole computer (like "/" on unix systems). The hard drive icon is just a special kind of folder. Okay, it strains the metaphor a little. It made more sense when the drive icons represented actual floppies.
(as an interesting aside, the early macs allowed you to physically eject a disk but keep the icon on your desktop. It was partially shaded so you could tell the difference. When you tried to access something on it, the floppy in the drive would eject and you would be prompted to put in the correct floppy. It's an interesting way to keep track of the disks you need).
If we are going to use hierarchical filesystems, I completely agree that there needs to be a single hierarchy of information. The Mac (pre-OSX anyway) does this consistently. Unix command-line systems are also consistent. Which one you prefer is kind of a personal taste issue.
This may sound like heresy, but a lot of people don't want the "fun" of configuring a computer as an audio server. I can imagine a lot of people who don't want to mess with some cranky old Windows machine just to listen to music. (It's probably also heresy to say that to most non-enlightened folk, "Computer" means "a PC running Windows")
Some of the features sound neat: 10x recording speed, USB download to MP3 players, CD burning, high-quality D/A converters, etc.
I would be worried about what happens to users and support once the RIAA decides this thing violates motherhood, apple pie, and their stranglehold on the music industry.
Slashdot Mirror
> You could make the machine unreliable, but that wouldn't make for an effective virus distributing machine.
10,000,000 Windows machines can't be wrong!
The balance between speed and accuracy has to depend on your goals:
:)
* Goal: sell papers; priority: speed with reasonable accuracy
* Goal: get attention; priority: speed with at least a little accuracy
* Goal: impeachment; priority: accuracy in time for the election >;)
Seriously, though, different mixtures of speed and accuracy appeal to different audiences. I don't expect that Drudge has a large following among CPA's, for example. And for the masses? Most of us, most of the time, want to be entertained without being fooled -- which means that your profs were right
now that's funny. I've got more!
...as honest as <your least favorite politician>
...as funny as a paper cut.
...as swamped with dates as a geek.
all right, mod me flamebait...
With the importance of the Internet and the fragility of some popular software (*cough*Windows *cough*Outlook), you could make an argument that these apply directly to software.
Does your OS interfere with other computers? Well, if a wide open hole allows whole sections of the net to go down, yeah, this widget interferes with other widgets. It may be difficult to burn down the house right now, but just wait until everybody has their home controlled by Windows Longhorn HVAC edition. Or what if a phreaker brings down 9-1-1 through an open firewall port? Credit card fraud and identity theft are common enough and dangerous enough to, possibly, be worth protecting against.
I'm a developer, I don't look forward to government edict making us even less competitive. But I also have to use the darn things, and from that perspective I'm as frustrated as anyone. Now if we could only get a "Software Quality" approval board *not* controlled by Microsoft, Oracle, etc, etc.
Maybe you and I and other normal people won't use it, but you can bet younger folks will. When the next hot boy band is only available on FingerNailChip (tm), it'll sell. If today's teenagers are like they were 20 years ago, "oldies" music is from two months ago.
Human perception is interesting.
:)
There are reports that the first people who heard a phonograph thought there was an actual orchestra. They found the effect *real* enough, even though it would sound scratchy, fuzzy, and fake to us today. As we become more familiar with a technology, our expectations go up. It is possible to spend $10,000 or more on a stereo system and still complain that it just isn't the same as live music.
We tend to project and fill in the details: finding shapes in clouds, seeing a face on the moon and on Mars. Maybe we start by filling in the details and then get more sophisticated (or lazy) and expect the technology to do more of the work.
On the other hand, realism is more than just "making it look real". You could argue, for example, that The Simpsons is more "real" than Leave it to Beaver. It certainly is a more accurate portrayal of modern attitudes.
I know your post is talking specifically of kind of a "Turing Test" for CG: can you tell which portions are CG and which were not. I am just continually fascinated by the way humans work.
As a humorous aside, it is ironic that we continue to raise our standards for CG in movies while our expectations for human actors seem to be in serious decline
No, they're just against it if you have the supply :)
In this case, they suspected the guy anyway, but couldn't prove it. They used the DNA to decide if he should be arrested. Why didn't they bring the guy in and get a new sample? I think that, if they couldn't legally get that done, they should not have used DNA that was collected for something else.
The real concern (which we didn't see happening here) is when the DNA database is done for searches for possible hits without doing other research first. This gets too close to illegal searches in my opinion.
Also, the error rate on matching is pretty low. But what happens to a low error rate when you have the entire population of the US or the world in a database? And what happens to "innocent until proven guilty" if your DNA matches?
Not trying to be flip ... but how do you design an AI without any reference to ethics? Dictionary.com defines ethics like this:
1. a. A set of principles of right conduct.
b. A theory or a system of moral value
Couldn't you stretch and say that any defined goal has some idea of right and wrong? For a stock trading program, profit=right, loss=wrong. For drug testing programs, dead customers=wrong, live customers=right.
Okay, to a logical person it's a ridiculous stretch. But recent history has shown that lawyers, juries, and the US patent office often operate outside of "normal" logic and common sense.
Thanks for the explanations of the age of the universe. But how do they know how old the planet is? I have been looking through the article and the responses and I don't see a complete explanation.
:). It could make decent corroborating/additional evidence, though.
The best I can figure is that it's age is calculated based on (1) its composition and (2) its location/neighbors/star. Is there more or is this it?
Choice (2) seems kind of suspect. I could move into a nursing home but it wouldn't make me old (although I'm uncomfortable close
The composition is more solid evidence, but I would think there would be room for interpretation there, too. We have direct physical access to a tiny speck of matter in a huge universe. How do we know that there aren't local deposits of these elements that could have formed a planet much later than they think?
I'm not fighting their calculations, I'd just like to understand them. And to understand how certain they are of their number. Even if it's only 80% sure, it's still an intriguing discovery.
Okay, we've beat to death the fact it was a pre-release version. But look at it this way:
When Open Source software is about the same quality as closed source, the developers consider it unstable and warn people that they may run into problems.
It shows a big difference, to me, in the quality standards that OSS developers (and users) expect.
I don't know what you're talking about ....
You can play the guitar after 27 beers? I'm impressed!
I think there is a lot of potential for professional-grade installations using something like Ardour. The tool itself is open-source and it runs on an open-source OS. Don't think it runs fast enough? Reconfigure your kernel by removing all the junk you don't use. Go through the startup scripts, turning off services and daemons you don't need. Switch to one of the more real-time oriented versions of Linux. Hack the source code, optimize the drivers, add your own features.
Okay, realistically, I'm not going to do this. I will plunk on my guitar in my basement, make a couple of mediocre tracks, and move on with my life.
But's what's stopping you from doing this? Or a technically savvy sound studio owner? Or, cooler than that, you could make a product by combining a rack mounted PC, a custom-configured version of Linux, a custom-configured version of Ardour, and the testing and support that would be needed. Of course, the GPL would require you to make the source code available. But that's okay. You get paid for your expertise without taking anything away from those who provided the expertise to build Ardour or Linux.
Whew, ranting a bit. Sorry. But I am excited by the possibilities.
Whoa, wait a minute here.
There are a lot of people, myself included, who think that the problem lies with the way software patents are researched and granted in the US, not necessarily with software patents by themselves. Not to mention the fascination US companies have with taking everything to court.
IANAL, but I believe there are at least two main tests that should be applied before granting a patent: (1) it should be new, and (2) it should not be obvious.
The first test looks at "prior art": is this a common practice or is this really novel and unique? Many software patents have been given for things that are very commonplace: the online "shopping cart", using XOR to draw a cursor on a graphics screen, and now (possibly) caching Internet content. The Patent Office seems to be doing a particularly poor job of making sure that software techniques being patented are actually new and different. It seems to stem from a huge number of applications coupled with a fundamental lack of understanding about software.
The second test, that it is not obvious, is a little more confusing, at least to me. Things that are obvious to programmers probably aren't obvious to, forgive me, normal people.
Having said all that, I'm sure that some people disagree with software patents altogether. I'll let them explain their point of view.
For what it is worth, I treated the SalesDroid respectfully, nodded my head, didn't argue, and said "Thank You". I was actually impressed by his knowledge and the way he explained it (although I admit this isn't clear from my first post). I didn't even come *close* to "Busting the guy's balls".
:)
My fun came at Intel's expense -- they are intentionally watering down one of their own "proofs" that their chips are better than what Apple is using.
I'll try to be more clear should I post again
A little off topic, perhaps, but ...
I was helping my Father-in-law shop for a laptop for Father's day. Trying to cause trouble, I asked a salesdroid at BestBuy why the new Centrino chip was so much slower than the P4. He gave me a song and dance about optimization, integration, blah, blah, and ended up telling me that clock speed is not that important for overall speed.
It seems like Intel is now playing both sides of the Megahertz gap. This could be a good thing for Apple. If Intel wants to sell the Centrino they will have to convince the public that the "slower" chips are "faster" or "better" and will give away much of their hype advantage. The Centrino tops out at something like 1.6 GHz, well below P4 and G5 clock speeds.
(This info comes from a friend who generally tells me the truth, so I'm giving secondhand info that I'm pretty sure is accurate.)
Small Claims Court is useful for two decent people to get an unbiased answer to a small dispute. The problem is that, in New Hampshire at least, there is no assistance for collecting judgements. It is up to the decency of the losing party to pay up. If they don't, well, you spent half a day in court with nothing to show for it.
In my friend's case, he was ripped off by some contractors building his house. He eventually got some sort of satisfaction by tying up their time in court and hurting their reputation. They eventually paid because he made himself a royal pain in the butt.
I imagine (pardon the metaphor) that Universal Records has a much bigger butt.
I don't see any realistic alternatives, either, but I wish I did. I think that the desktop metaphor helped us immensely at first but is now holding us back.
Some alternatives have been tried and rejected. Examples:
1) IBM (and others) tried to make on-screen objects look and act like real objects.
Real CD player and RealPhone
2) The desktop can be expanded to a virtual office building, with file cabinets in a virtual file room. This doesn't work because it is really the same as a desktop, except with the added advantage of seeming beaurocratic and unfriendly.
3) Virtual reality was destined for greatness as an interface. Maybe there's hope as computers get more and more powerful. Or maybe people don't want to wear goggles and sci-fi wired gloves just to check their email.
I think there are some good reasons that the desktop is still alive:
1) It works pretty well. Not perfect, but good enough for most things most of the time.
2) It is entrenched in our thinking.
3) Inertia and legacy code.
4) Coming up with good, useful, extendable metaphors is difficult.
So, the desktop lives on. Maybe the real breakthrough will be actual physical devices that can hide the computer underneath so they're just "smart", not computers.
The worst part of the Windows desktop? It lives in more than one place. It is the "root" or "base" of the system (from the GUI), but it is also a folder buried under the Windows directory on one of the hard drives (from the command line or file browser).
With more than one drive, the drives exist as (1) icons inside "My Computer" inside the desktop, (2) top-level directories (Windows has no real "/" in the Unix sense).
This schizophrenic desktop and drive setup can cause major confusion. It confused me (still does) even with 15+ years of computer and programming experience. Why on earth would they implement it that way? And why did the KDE/Gnome developers copy it?
On a Mac, the "desktop" is your whole computer (like "/" on unix systems). The hard drive icon is just a special kind of folder. Okay, it strains the metaphor a little. It made more sense when the drive icons represented actual floppies.
(as an interesting aside, the early macs allowed you to physically eject a disk but keep the icon on your desktop. It was partially shaded so you could tell the difference. When you tried to access something on it, the floppy in the drive would eject and you would be prompted to put in the correct floppy. It's an interesting way to keep track of the disks you need).
If we are going to use hierarchical filesystems, I completely agree that there needs to be a single hierarchy of information. The Mac (pre-OSX anyway) does this consistently. Unix command-line systems are also consistent. Which one you prefer is kind of a personal taste issue.
This may sound like heresy, but a lot of people don't want the "fun" of configuring a computer as an audio server. I can imagine a lot of people who don't want to mess with some cranky old Windows machine just to listen to music. (It's probably also heresy to say that to most non-enlightened folk, "Computer" means "a PC running Windows")
Some of the features sound neat: 10x recording speed, USB download to MP3 players, CD burning, high-quality D/A converters, etc.
I would be worried about what happens to users and support once the RIAA decides this thing violates motherhood, apple pie, and their stranglehold on the music industry.