I thought this was part of the point of open source software. I assume by the fact that they're giving credit that they're taking actual code, not just feature ideas. The great thing about this is that you can fork without completely splitting your development resources down the middle. The giving of credit is very important though, beyond the legal requirements in the source code attribution. You should credit them as loudly as possible, to avoid political problems, as the other project is a potentially very important resource.
Plausible deniability doesn't mean you can convince someone you didn't say to them what you said to them. It just means they can't prove to someone else that you said it. That they logged your IP just means they know your IP, which proves at most that you talked with them about something, but not what you said.
I'm paying about $700 per hour, and I know for a fact that my school is losing a lot of money on me. Next year, when I'm in-state and paying about $200 per hour, they'll be losing even more. This is why schools have endowments.
I used to be a student at U.Va. in the engineering school, which receives 8% of its budget from the state. If the U.Va. engineering school didn't have other sources of funding, like the university's $1.9 billion general endowment, endowed professorships, special-purpose gifts, etc., and was just going on tuition and state funds, it wouldn't be able to offer a single degree program.
Check out your school's stats. I'm sure you'll find they're losing money hand over fist on you.
Unless you're taking classes from the University of Phoenix Online, you're not buying a product. You're essentially receiving a gift, and paying enough money that they can be sure you're going to take it seriously, so their money wouldn't be better spent on someone else. Tuition is only a big deal for them because it is the most easily controlled source of marginal revenue. Anyway, I did some checking. You're right, 5-50 was wrong. It's more like 2-20. The 2 is if you're an English major at a poorly endowed private college. Most of the slashdot crowd is in the middle or on the 20 end, especially if they're paying in-state tuition at a research institution.
Do you honestly believe your tuition pays for your education? They spend anywhere from $5-$50 per dollar you pay in tuition on you. It's a bargain at twice the price.
That puzzle should be easier than the 15-puzzle...
on
Programming Puzzles
·
· Score: 1
At least for a computer. The scramble counting method used will result in lots of redundant positions. If you do a heuristic search, you'll never encounter redundant board configurations. The variable piece size will actually lower the effective branching factor for heuristic searches, since many moves that are feasible will generate infeasible children, and never again be visited. Also, since the ratio of open to occupied space is higher, you'll have a larger proportion of states where the branching factor is also diminished by the fact that the space is adjacent to a side of the board, meaning that no piece can fill it from that side.
I actually wrote a program that used a variety of searches and heuristics to solve the 15-puzzle (GBFS with the Manhattan Distance heuristic worked fastest, but gave less optimal answers than A*, obivously.) If I weren't in the middle of finals, I'd dust off that code and mangle it to do this to prove it.
My point is, while the puzzle shown may have a deeper and wider solution tree, by being smart, you can prune the width and avoid depth-increasing inefficiencies, so the less uniform structure actually help here.
In a conventional war, with a front separating the territory held by each faction, people who live on it tend to go elsewhere until the war moves along. If we still fought wars like that, autonomous firing systems would be a lot less problematic. Problem is, we spend a month fighting like that, and are left with years of guerilla warfare that follow, effectively making the entire country a battle zone.
The Patriot Missile system fires with no human intervention. It uses an Identification Friend-or-Foe system to track everything in the air, and shoot down anything that shouldn't be there. During the recent Iraq invasion, a glitch in this system caused it to fire upon a British fighter jet, destroying it and killing its pilot. It was about to do the same to a US jet, but that jet was armed with fast-flying radar-seeking missiles designed to take out hostile SAM sites, and was able to take out the radar component of the Patriot system before the missile reached his plane. Notably no one was injured on the ground when he did this, since there was nobody actually sitting in front of the device, or anywhere near it.
I think it'll be a long time before autonomously firing ground systems are in place, because it's hard enough doing IFF in the sky, let alone on the ground. I think the fire-finder system (used in the Balkans to take out mortar positions in the mountains firing upon cities) might do this in some limited capacity, but that's only anti-artillery, rather than telling the difference between a guerilla carrying an RPG and a farmer carrying a section of irrigation pipe. Sure, you could wait until they shoot first for all of these systems, since that's a lot easier to determine automatically, but I think it's quite obvious that waiting for the other guy to shoot first is very far from the policy of the current administration.
I used to work tech support at one of the rare well-managed, worker-respecting tech support places, and fairly frequently we'd get calls about things we don't support. Standard procedure was to refer them to the consulting companies we had deals with. Because of these contracts, our users got good deals from them, but only for major projects, since they'd typically have minimum fees that would be rather exorbitant for the small odd jobs they often needed. We'd often get calls back asking if anyone wanted to drop by for a half hour after work and do whatever the odd job in question was for $30 or so. Our manager actually encouraged this practice, since we were still supporting everything we were supposed to and honoring our contracts, and our users were getting the unsupported odd jobs done that were too small for formal consulting. This required our manager paying attention to make sure we were really doing our jobs properly and not trying to screw our customers, but I believe I already mentioned we had good management.
It's about time we had robots that could fix orbiting devices. Two billion is a bargain. Oh, yeah, and it might just save one of the most scientifically energizing pieces of space hardware ever flown.
Some people here seem to think that they'd have to be snooping lots and lots of net traffic in order for this to be any good to them. Not so. If you strongly suspect that the perpetrator comes from some small set, like, say, employees of a certain corporation, students at a certain school, etc., then a 5-minute window of logs will likely show only one hit from that IP range. That, along with what they have that leads them to suspect that IP range in the first place could be enough to execute a warrant.
If you're referring to the quantum effect of coupling, which allows action at a distance for instant communication, I believe that experiments have been able to do it at ranges of a few meters, up to a few seconds after the initial coupling, before it decays. We're still centuries from deploying that technology.
I just got an offer for a co-op at a company I imagine most slashdotters would jump at, or at least those who are 22. For those who aren't in the know, it's when you take a semester off, with the encouragement of your school, and work in industry. They can pay you more than you've ever made in any long term job, but less than they'd pay a true full-time hire, and it's a fairly safe bet you're not going to need a lot of the benefits their average permanent employee needs. You're not there for very long, but since you've just been in school, you pick up the training faster. Ultimately, they get more minions for their expert non-managers, and a good, low-pressure preview of how you'd be as an employee when you get out, and many people do end up working for the companies they co-op with.
Of course, most of you aren't in school. Sucks to be you. Have you thought about a professional master's program? I'm hearing that the mid-career students are doing well with prospects, and in many cases getting promotions and raises at the companies they're currently with, just to keep them around.
While there's not usually a difference under the law, there *is* a huge difference in the courts. Freeloading will generally cost you no more than statutory damages in civil court. Copyright Piracy (the actual definition, not the **AA definition) as in cases like this has a tendency to result in getting the smackdown in the form of very liberal calculations of actual damages, plus punitive damages.
There are actually services that have people read your mail for you and filter it out themselves. These automated filters can actually do better than that, which is good, since I can't imagine any business tolerating the notion of outsiders reading every last one of their email messages.
But their aggregate behavior is quite easily computable. In the human brain, 70% of neuron firings randomly fail to register in their successor. This not only makes our behavior somewhat random, but also implies that there's quite a bit of redundancy and that our brains operate on aggregate behavior of a large neural net, rather than precise behavior of a small one, otherwise we'd be completely unpredictable, rather than just mostly unpredictable. While it's true that you can't model a human brain reliably with a computer, it's also true that you can't even model it reliably with the same human brain. Generally speaking, any simulation that is as good as reality is good enough, even if reality isn't really right.
I'm not really sure what you mean that they haven't been proven. In the sense that they don't give the best answer all the time, this much is obvious. That's why we call it artificial intelligence instead of algorithmics. That said, we know quite well that they work. Most adaptive spam filters are based on Bayesian networks. The best of these are better than humans at identifying spam. We don't typically run the best because the computational load is far too high. Bayesian networks have a delightfully simple evaluation procedure that is basically glorified matrix multiplication. Neural networks are a little more complicated, but not by a whole lot. Recall a recent development that used a neural network inside an 802.11 driver to predictively avoid collisions to improve total network throughput in dense environments. It doesn't reduce collisions to 0, as that would require clairvoyance, but it does a good job. You didn't hear about this 5 years ago because putting a neural net inside an 802.11 driver without killing performance to both network and computer is difficult, particularly without processor instructions dedicated to the task.
It's true that designing a CPU to *be* a neural or bayesian network is infeasible, but that doesn't mean we can't add instructions to accelerate their evaluation. The evaluation and update of a neural net, traditional or biologically modeled, is a rather simple algorithmic process, though people who have worked with such simulations (see Ben Hocking's post above, he was my quite capable AI TA) will tell you that they make rather obscene optimizations to make it run reasonably fast. I'm talking about things that might sound familiar to graphics people, like removing all multiplications from a program that's supposed to be doing them more than all other operations combined. It's a particularly good candidate for SIMD instructions. Most large neural nets are sparsely connected, so even if your net is substantially larger than your cache, you can beat that with prefetching. Threshold conditional addition is an example of something that can be done very quickly in hardware, and is much more of a pain to code and optimize in software.
If you prefer RISC to CISC, recall that even the original SPARC had special DSP instructions. Putting the sigmoid function and arctan on silicon is really not all that outrageous.
Problem is, off-chip peltier units are separated from the hot parts of the CPU by packaging materials, at least some of which are thermally insulating. This is fine when thermal output is roughly evenly distributed on your CPU, but this is becoming less and less the case. The on-chip peltier helps smooth out the thermal distribution, allowing the external cooling device to keep the chip operating properly without requiring liquid nitrogen.
Of course, even if you adopt the traditional definition of piracy in the scope of intellectual property law, most of what is being decried as piracy is in fact *not* piracy. Piracy in this scope is the mass publication for profit of something you represent to be an authorized distribution of the creator or rights-holder. The term applies to the Win98 CD my dad once bought from a sketchy parts store that had printed on it a warning that copying the CD was "illigal". It certainly doesn't apply in this case though.
Sounds like you've been using PostgreSQL for quite a while, since your criticisms generally apply to rather old versions of MySQL. I hate all databases with equal passion, so I figured I ought to set the record straight.
No data integrity
MySQL has a modular design, allowing you to use any of several database backends, each with different design priorities. If data integrity is a priority for you, RTFM and pick the right backend.
Completely non-standard SQL
This is true of pretty much all DBMS's. MySQL's development cycle is quite delightfully quick, meaning that if people want certain features added, it doesn't take long to get them added.
No extensibility in the engine (functions, stored procedures, etc.)
Ummm... MySQL has those.
No subselects
Yes it does, as of version 4.1. It's not their fault if you're more than a full release behind.
Weird handling of '0' vs. null
The special handling of null forces you to write slightly less braindead code. This kind of handholding is becoming popular in the design of modern programming languages like Java and C#, with the recognition that the later in your development cycle you find a bug, the more it costs to fix it. Of course, there are still times when the programmer needs very explicit control, like when writing an OS kernel or a device driver. If you're writing an OS kernel or a device driver in MySQL, you've got bigger problems.
As a TA, I can attest to this. TAs remember mistakes. Copied code is even more likely to have mistakes, since the person doing the copying doesn't really understand the code. When we see the same mistakes in the span of 2 hours when grading, we notice. Then we mark through their original grade on the grade sheet, write PLAGIARISM in large red letters, and provide them with the URL for the code of student conduct where it says if they do it a second time after receiving a written warning, we get to send them to the committee with the authority to suspend and expel people.
Sometimes I wish corporal punishment were still permitted.
Any time a system breaks because a person does something stupid, a rational analyst asks if there is a way you could change the system to avoid giving the user the opportunity to screw it up. Usually it is. If user-friendliness wasn't in their requirements list, someone responsible for the design of the system screwed up. If it was in their requirements list, it sounds like it probably failed the requirement, so again, someone screwed up. It really sounds like they're just scapegoating the voters here.
This statement implies that the Neopets message boards were not *already* nuts. As a good friend of several Neopets players, I can assure you that this is not the case. While a slight majority of Neopets players are mostly sane, the whole endeavor represents an act of collective insanity, sort of like the Salem Witchcraft Trials, the popular belief that the 9/11 Hijackers were Iraqi, and Britney Spears's fan base, though far less destructive than all 3 of these.
Anyway, poker teaches valuable skills, like quickly estimating probabilities and expectations, economic principles like sunk costs, and discipline. Anything that's fun can be addictive. Given that children don't have much of a concept of their time being worth anything, their valuation of virtual credits on Neopets is likely to be substantially less than adults who do things like purchase Everquest items on eBay. I'm really much more worried about the adults here, but if we as a society don't let people play poker with fake money to feed fake pets; alcohol, premarital sex, and bungie jumping will be the next to go, and that's not a world I want to live in.
Slashdot Mirror
I thought this was part of the point of open source software. I assume by the fact that they're giving credit that they're taking actual code, not just feature ideas. The great thing about this is that you can fork without completely splitting your development resources down the middle. The giving of credit is very important though, beyond the legal requirements in the source code attribution. You should credit them as loudly as possible, to avoid political problems, as the other project is a potentially very important resource.
Plausible deniability doesn't mean you can convince someone you didn't say to them what you said to them. It just means they can't prove to someone else that you said it. That they logged your IP just means they know your IP, which proves at most that you talked with them about something, but not what you said.
I'm paying about $700 per hour, and I know for a fact that my school is losing a lot of money on me. Next year, when I'm in-state and paying about $200 per hour, they'll be losing even more. This is why schools have endowments.
I used to be a student at U.Va. in the engineering school, which receives 8% of its budget from the state. If the U.Va. engineering school didn't have other sources of funding, like the university's $1.9 billion general endowment, endowed professorships, special-purpose gifts, etc., and was just going on tuition and state funds, it wouldn't be able to offer a single degree program.
Check out your school's stats. I'm sure you'll find they're losing money hand over fist on you.
Unless you're taking classes from the University of Phoenix Online, you're not buying a product. You're essentially receiving a gift, and paying enough money that they can be sure you're going to take it seriously, so their money wouldn't be better spent on someone else. Tuition is only a big deal for them because it is the most easily controlled source of marginal revenue. Anyway, I did some checking. You're right, 5-50 was wrong. It's more like 2-20. The 2 is if you're an English major at a poorly endowed private college. Most of the slashdot crowd is in the middle or on the 20 end, especially if they're paying in-state tuition at a research institution.
Do you honestly believe your tuition pays for your education? They spend anywhere from $5-$50 per dollar you pay in tuition on you. It's a bargain at twice the price.
At least for a computer. The scramble counting method used will result in lots of redundant positions. If you do a heuristic search, you'll never encounter redundant board configurations. The variable piece size will actually lower the effective branching factor for heuristic searches, since many moves that are feasible will generate infeasible children, and never again be visited. Also, since the ratio of open to occupied space is higher, you'll have a larger proportion of states where the branching factor is also diminished by the fact that the space is adjacent to a side of the board, meaning that no piece can fill it from that side.
I actually wrote a program that used a variety of searches and heuristics to solve the 15-puzzle (GBFS with the Manhattan Distance heuristic worked fastest, but gave less optimal answers than A*, obivously.) If I weren't in the middle of finals, I'd dust off that code and mangle it to do this to prove it.
My point is, while the puzzle shown may have a deeper and wider solution tree, by being smart, you can prune the width and avoid depth-increasing inefficiencies, so the less uniform structure actually help here.
In a conventional war, with a front separating the territory held by each faction, people who live on it tend to go elsewhere until the war moves along. If we still fought wars like that, autonomous firing systems would be a lot less problematic. Problem is, we spend a month fighting like that, and are left with years of guerilla warfare that follow, effectively making the entire country a battle zone.
The Patriot Missile system fires with no human intervention. It uses an Identification Friend-or-Foe system to track everything in the air, and shoot down anything that shouldn't be there. During the recent Iraq invasion, a glitch in this system caused it to fire upon a British fighter jet, destroying it and killing its pilot. It was about to do the same to a US jet, but that jet was armed with fast-flying radar-seeking missiles designed to take out hostile SAM sites, and was able to take out the radar component of the Patriot system before the missile reached his plane. Notably no one was injured on the ground when he did this, since there was nobody actually sitting in front of the device, or anywhere near it.
I think it'll be a long time before autonomously firing ground systems are in place, because it's hard enough doing IFF in the sky, let alone on the ground. I think the fire-finder system (used in the Balkans to take out mortar positions in the mountains firing upon cities) might do this in some limited capacity, but that's only anti-artillery, rather than telling the difference between a guerilla carrying an RPG and a farmer carrying a section of irrigation pipe. Sure, you could wait until they shoot first for all of these systems, since that's a lot easier to determine automatically, but I think it's quite obvious that waiting for the other guy to shoot first is very far from the policy of the current administration.
I used to work tech support at one of the rare well-managed, worker-respecting tech support places, and fairly frequently we'd get calls about things we don't support. Standard procedure was to refer them to the consulting companies we had deals with. Because of these contracts, our users got good deals from them, but only for major projects, since they'd typically have minimum fees that would be rather exorbitant for the small odd jobs they often needed. We'd often get calls back asking if anyone wanted to drop by for a half hour after work and do whatever the odd job in question was for $30 or so. Our manager actually encouraged this practice, since we were still supporting everything we were supposed to and honoring our contracts, and our users were getting the unsupported odd jobs done that were too small for formal consulting. This required our manager paying attention to make sure we were really doing our jobs properly and not trying to screw our customers, but I believe I already mentioned we had good management.
It's about time we had robots that could fix orbiting devices. Two billion is a bargain. Oh, yeah, and it might just save one of the most scientifically energizing pieces of space hardware ever flown.
Some people here seem to think that they'd have to be snooping lots and lots of net traffic in order for this to be any good to them. Not so. If you strongly suspect that the perpetrator comes from some small set, like, say, employees of a certain corporation, students at a certain school, etc., then a 5-minute window of logs will likely show only one hit from that IP range. That, along with what they have that leads them to suspect that IP range in the first place could be enough to execute a warrant.
If you're referring to the quantum effect of coupling, which allows action at a distance for instant communication, I believe that experiments have been able to do it at ranges of a few meters, up to a few seconds after the initial coupling, before it decays. We're still centuries from deploying that technology.
I just got an offer for a co-op at a company I imagine most slashdotters would jump at, or at least those who are 22. For those who aren't in the know, it's when you take a semester off, with the encouragement of your school, and work in industry. They can pay you more than you've ever made in any long term job, but less than they'd pay a true full-time hire, and it's a fairly safe bet you're not going to need a lot of the benefits their average permanent employee needs. You're not there for very long, but since you've just been in school, you pick up the training faster. Ultimately, they get more minions for their expert non-managers, and a good, low-pressure preview of how you'd be as an employee when you get out, and many people do end up working for the companies they co-op with.
Of course, most of you aren't in school. Sucks to be you. Have you thought about a professional master's program? I'm hearing that the mid-career students are doing well with prospects, and in many cases getting promotions and raises at the companies they're currently with, just to keep them around.
The state that gave us Silicon Valley is leading the fight for medical marijuana. Coincidence?
I think not.
While there's not usually a difference under the law, there *is* a huge difference in the courts. Freeloading will generally cost you no more than statutory damages in civil court. Copyright Piracy (the actual definition, not the **AA definition) as in cases like this has a tendency to result in getting the smackdown in the form of very liberal calculations of actual damages, plus punitive damages.
There are actually services that have people read your mail for you and filter it out themselves. These automated filters can actually do better than that, which is good, since I can't imagine any business tolerating the notion of outsiders reading every last one of their email messages.
But their aggregate behavior is quite easily computable. In the human brain, 70% of neuron firings randomly fail to register in their successor. This not only makes our behavior somewhat random, but also implies that there's quite a bit of redundancy and that our brains operate on aggregate behavior of a large neural net, rather than precise behavior of a small one, otherwise we'd be completely unpredictable, rather than just mostly unpredictable. While it's true that you can't model a human brain reliably with a computer, it's also true that you can't even model it reliably with the same human brain. Generally speaking, any simulation that is as good as reality is good enough, even if reality isn't really right.
I'm not really sure what you mean that they haven't been proven. In the sense that they don't give the best answer all the time, this much is obvious. That's why we call it artificial intelligence instead of algorithmics. That said, we know quite well that they work. Most adaptive spam filters are based on Bayesian networks. The best of these are better than humans at identifying spam. We don't typically run the best because the computational load is far too high. Bayesian networks have a delightfully simple evaluation procedure that is basically glorified matrix multiplication. Neural networks are a little more complicated, but not by a whole lot. Recall a recent development that used a neural network inside an 802.11 driver to predictively avoid collisions to improve total network throughput in dense environments. It doesn't reduce collisions to 0, as that would require clairvoyance, but it does a good job. You didn't hear about this 5 years ago because putting a neural net inside an 802.11 driver without killing performance to both network and computer is difficult, particularly without processor instructions dedicated to the task.
It's true that designing a CPU to *be* a neural or bayesian network is infeasible, but that doesn't mean we can't add instructions to accelerate their evaluation. The evaluation and update of a neural net, traditional or biologically modeled, is a rather simple algorithmic process, though people who have worked with such simulations (see Ben Hocking's post above, he was my quite capable AI TA) will tell you that they make rather obscene optimizations to make it run reasonably fast. I'm talking about things that might sound familiar to graphics people, like removing all multiplications from a program that's supposed to be doing them more than all other operations combined. It's a particularly good candidate for SIMD instructions. Most large neural nets are sparsely connected, so even if your net is substantially larger than your cache, you can beat that with prefetching. Threshold conditional addition is an example of something that can be done very quickly in hardware, and is much more of a pain to code and optimize in software.
If you prefer RISC to CISC, recall that even the original SPARC had special DSP instructions. Putting the sigmoid function and arctan on silicon is really not all that outrageous.
Problem is, off-chip peltier units are separated from the hot parts of the CPU by packaging materials, at least some of which are thermally insulating. This is fine when thermal output is roughly evenly distributed on your CPU, but this is becoming less and less the case. The on-chip peltier helps smooth out the thermal distribution, allowing the external cooling device to keep the chip operating properly without requiring liquid nitrogen.
Of course, even if you adopt the traditional definition of piracy in the scope of intellectual property law, most of what is being decried as piracy is in fact *not* piracy. Piracy in this scope is the mass publication for profit of something you represent to be an authorized distribution of the creator or rights-holder. The term applies to the Win98 CD my dad once bought from a sketchy parts store that had printed on it a warning that copying the CD was "illigal". It certainly doesn't apply in this case though.
Sounds like you've been using PostgreSQL for quite a while, since your criticisms generally apply to rather old versions of MySQL. I hate all databases with equal passion, so I figured I ought to set the record straight.
No data integrity
MySQL has a modular design, allowing you to use any of several database backends, each with different design priorities. If data integrity is a priority for you, RTFM and pick the right backend.
Completely non-standard SQL
This is true of pretty much all DBMS's. MySQL's development cycle is quite delightfully quick, meaning that if people want certain features added, it doesn't take long to get them added.
No extensibility in the engine (functions, stored procedures, etc.)
Ummm... MySQL has those.
No subselects
Yes it does, as of version 4.1. It's not their fault if you're more than a full release behind.
Weird handling of '0' vs. null
The special handling of null forces you to write slightly less braindead code. This kind of handholding is becoming popular in the design of modern programming languages like Java and C#, with the recognition that the later in your development cycle you find a bug, the more it costs to fix it. Of course, there are still times when the programmer needs very explicit control, like when writing an OS kernel or a device driver. If you're writing an OS kernel or a device driver in MySQL, you've got bigger problems.
As a TA, I can attest to this. TAs remember mistakes. Copied code is even more likely to have mistakes, since the person doing the copying doesn't really understand the code. When we see the same mistakes in the span of 2 hours when grading, we notice. Then we mark through their original grade on the grade sheet, write PLAGIARISM in large red letters, and provide them with the URL for the code of student conduct where it says if they do it a second time after receiving a written warning, we get to send them to the committee with the authority to suspend and expel people.
Sometimes I wish corporal punishment were still permitted.
Any time a system breaks because a person does something stupid, a rational analyst asks if there is a way you could change the system to avoid giving the user the opportunity to screw it up. Usually it is. If user-friendliness wasn't in their requirements list, someone responsible for the design of the system screwed up. If it was in their requirements list, it sounds like it probably failed the requirement, so again, someone screwed up. It really sounds like they're just scapegoating the voters here.
...the Neopets message boards went nuts...
This statement implies that the Neopets message boards were not *already* nuts. As a good friend of several Neopets players, I can assure you that this is not the case. While a slight majority of Neopets players are mostly sane, the whole endeavor represents an act of collective insanity, sort of like the Salem Witchcraft Trials, the popular belief that the 9/11 Hijackers were Iraqi, and Britney Spears's fan base, though far less destructive than all 3 of these.
Anyway, poker teaches valuable skills, like quickly estimating probabilities and expectations, economic principles like sunk costs, and discipline. Anything that's fun can be addictive. Given that children don't have much of a concept of their time being worth anything, their valuation of virtual credits on Neopets is likely to be substantially less than adults who do things like purchase Everquest items on eBay. I'm really much more worried about the adults here, but if we as a society don't let people play poker with fake money to feed fake pets; alcohol, premarital sex, and bungie jumping will be the next to go, and that's not a world I want to live in.
That only 1 in 400 computer users doesn't get popups.