FWIW, one of the first government tech investments ended up going waaay over-budget, but also ended up saving insurmountable amounts of labor costs; it saved massive amounts of money in the long-run, and ended up setting us up as the early leader in the realm of computing. So, thanks 1890 US Census Bureau.
Exactly. Of course a CEO is going to say that their company is still viable in the market. Who would ever hire a CEO who would say "dangit, you're right, we're done for, good thing I sold all my stock this morning!"?
Right, this is exactly why the founding people at Palm all bailed to form Handspring when 3Com started ruining it. And then bailed once again when Palm acquired them. Palm was driven into the ground hardcore.
Right, Patreon is one of the few things I enjoy paying for in life. I like the ability to help people quit their day-jobs and spend their time working to create something that fascinates me.
Right. A fridge uses around 150 watts when active. And because it is not a theoretical heat-engine, it only going to be able to cool a fraction of that amount of energy. A computer such as OP is imagining uses a few hundred watts. Certainly you could build something that is able to handle the heat generated, but, it's going to need hundreds of watts of power. Compare that to a liquid cooled setup using a similarly sized radiator. That would require a couple watts to run the pump.
Condensation is only an issue when the condensation is so great that electricity is allowed to pass between two electrical contacts. A refrigerator cycles the air within, so if there's no humidity to start within inside a closed fridge, it will not form enough condensate to be a problem. That said, not much condensation is gonna form at all because the PC is generating hundreds of watts of heat that the refrigerator is constantly struggling to cool.
Technically, this can be done. It doesn't work in terms of putting a PC in a commercial fridge. Lots of explanations why that doesn't work already in the comments (by the way, condensation isn't one of them). The way you would make this work is by cycling refrigerant through a heat-conductive plate thermally mounted to the heat producing components. Then allowing the refrigerant to expand in a radiator, compressing it, and continuing the cycle. If you built a large enough system with a beefy enough compressor, and a large enough radiator, it would work to keep your components nice and cool.
The reason why this isn't done is the same reason why it's not done for the engine in your car. Basically, there's no benefit of making your PC run _colder_ than the ambient air temperature. That's what a phase-transition heat-exchanger is good at. If you simply need your chips to not get significantly hotter than ambient air temperature, then it is more efficient to skip the compression, and merely rely on liquid cooling.
That said, phase-transition is not entirely unheard of in the realm of computing. It's merely a very niche application. Some aspects of quantum computers depend on materials behaving differently at very cold temperatures. They will use things like liquid nitrogen to achieve low temperatures. LN is produced by specialty compressors, but in essence, they behave the same as the compressor of a refrigerator. Conceivably, you could hook those compressors directly to the component in a cycle; this just isn't done because it doesn't serve any particular benefit when a lab can just purchase LN, as it's fairly cheap stuff.
As I mentioned in response to another one of your comments, no, you are mistaken. Derivative works are in the realm of copyright law, not trademark law. You can't copy a person's character and write stories about their further adventures and sell those stories. You can't draw a green frog that looks just like pepe, and name him pepe and sell him as an app. No trademark is needed for this protection. It is automatic.
You can draw pepe's on your hardware all you want. Copyright law says that unlicensed derivative works can't be _sold_ unless they fall within fair-use, and this doesn't.
They are, but think about GPUs in the late '90s, when this codec was finalized. I don't think there were any real-time rendered fractals going on in my copy of Unreal Tournament. You could have 2 GPU cores if you really wanted to spend the money, but as I understand it, the drivers weren't really out there to make particularly efficient use of them beyond simple stuff like allowing multiple monitors (I say "simple", that's a laugh, multiple monitor support was atrocious on Windows98, and it took me hours to get it working as expected on Redhat 6; I vaguely recall needing to compile a custom kernel, but don't quote me on that).
It wouldn't be nullifying the patent, but potentially, they could acquire a patent via Eminent Domain, so long as the provide Just Compensation. As far as I can find, there's no precedent for the US government entirely acquiring a patent in this manner. The closest I know of is that they made use of patents without licensing during WWII, then compensated after the fact.
I can't get too furious about this. I'm reading all the things that could go wrong in the comments and why it's a bad idea, and, it's not anything too terrifying. Cost of airfare is always gonna go up one way or another. If it goes up too much, then seats are empty, and airlines lose money; the market adjusts. And if this proposed company screws up, then oversight still drops in and starts making changes. I know it feels weird for Trump to support something that isn't the dumbest idea ever, but, how about we agree to this one, if you give the Democrats something in return for that support...y'know, the way politics used to behave?
Thankfully, that's not the major aspect of his climate-change legacy. His bigger legacy is likely going to be the increased grants and government loans for alternative energy endeavors. He's not the greatest president for environmental efforts (that would probably be Nixon so far), but given the powers granted to a president, Obama did OK.
Nuclear plants are massively expensive and take many years to build; during that time lots of things can halt the effort. This makes it a significantly risky investment. This means it is difficult to secure the money to build them. There are theoretical smaller-scale solutions in development, but they are not ready for fielding (and some may turn out to be impractical). Wind & Solar are safe investments. They can be built quickly, and they come with no public pushback, and very little Not-In-My-Backyard effect. so that's what's being pursued for now. As those realms become more saturated, the push for nuclear can become stronger.
I'm happy to be wrong. Please point me to papers or product-pages, or explain in greater detail? I fully understand that the watermarking algorithm would behave like a compression algorithm, but I'm under the impression that a crack could work on the pixel-level regardless.
Right, it's hard, but I don't believe it's impossible. That's more of what makes it interesting. If someone paid me to design that, I'd have a blast, but if I owned a company that is made or broken based on whether or not such a design worked...Let's just say I'm not buying in on the Screening Room IPO. I've read a couple patents on the problem, but as far as I've been able to find, there's no gold standard algorithm that has been proven in the field on this yet.
It's like a good hash that you have to spread across a video by slightly tweaking values for pixels. You're only allowed a narrow window of entropy, and you can only employ it on a smallish percentage of the total pixels or it becomes distracting. You can either pick the same pixels to change for every user, or you can pick different pixels every time. If it's different pixels every time, then a 3-way diff will reveal the appropriate value to choose to corrupt the plaintext. If it's the same pixels every time, then you isolate those pixels, and you interpolate their appropriate values based on the surrounding pixels. Generally, the place you would like hide this information is in the blacks because that is the least noticeable on anything except extremely high contrast devices. This is also where a lossy compression is going to remove information for exactly the same reason. When it's hidden here, taking the average works extremely well. When it isn't, then either the watermark is already gonna be table-flip distracting, or it's gonna be so uncommon that the average will be Good Enough.
I miss the days when PBS had enough federal funding that it didn't need to air the quasi-informercial junk that it has now. I mean, they always had the aggravating pledge drives, but they'd usually at least have good programming that went along with them.
I used to think that a liberal education program was stupid for technical fields. I thought it was dumb that I had to take all of these core classes that weren't related to Comp-Sci. But on reflection, and having dealt with people who experienced the fully technical schools, people who self-taught and had no formal higher education and others like myself who got a liberal 4-year Bachelor of Science, I really feel that it enriched me, both in my personal and professional life. I always hear people talking about how they never use what they learned in college...I use tons of what I learned in college. I wish the foreign language requirements weren't as harsh; I came in with 3 years of spanish in high school, and they still wanted me to take more. But beyond that, most everything I picked has helped me in life.
I was recently looking into the digital watermarking problem, and it's a fascinating bit of computer science. It's really tricky to sneak into a video with subtly adjusted pixels or audio information in such a way that it reveals the originator of the video. Many of them can be defeated just by doing a lossy compression. Getting something that survives that while not being distracting the viewer; ideally not even being visible is a fun little challenge to tackle.
Slashdot Mirror
FWIW, one of the first government tech investments ended up going waaay over-budget, but also ended up saving insurmountable amounts of labor costs; it saved massive amounts of money in the long-run, and ended up setting us up as the early leader in the realm of computing. So, thanks 1890 US Census Bureau.
Exactly. Of course a CEO is going to say that their company is still viable in the market. Who would ever hire a CEO who would say "dangit, you're right, we're done for, good thing I sold all my stock this morning!"?
Right, this is exactly why the founding people at Palm all bailed to form Handspring when 3Com started ruining it. And then bailed once again when Palm acquired them. Palm was driven into the ground hardcore.
Right, Patreon is one of the few things I enjoy paying for in life. I like the ability to help people quit their day-jobs and spend their time working to create something that fascinates me.
Right. A fridge uses around 150 watts when active. And because it is not a theoretical heat-engine, it only going to be able to cool a fraction of that amount of energy. A computer such as OP is imagining uses a few hundred watts. Certainly you could build something that is able to handle the heat generated, but, it's going to need hundreds of watts of power. Compare that to a liquid cooled setup using a similarly sized radiator. That would require a couple watts to run the pump.
Condensation is only an issue when the condensation is so great that electricity is allowed to pass between two electrical contacts. A refrigerator cycles the air within, so if there's no humidity to start within inside a closed fridge, it will not form enough condensate to be a problem. That said, not much condensation is gonna form at all because the PC is generating hundreds of watts of heat that the refrigerator is constantly struggling to cool.
The reason why this isn't done is the same reason why it's not done for the engine in your car. Basically, there's no benefit of making your PC run _colder_ than the ambient air temperature. That's what a phase-transition heat-exchanger is good at. If you simply need your chips to not get significantly hotter than ambient air temperature, then it is more efficient to skip the compression, and merely rely on liquid cooling.
That said, phase-transition is not entirely unheard of in the realm of computing. It's merely a very niche application. Some aspects of quantum computers depend on materials behaving differently at very cold temperatures. They will use things like liquid nitrogen to achieve low temperatures. LN is produced by specialty compressors, but in essence, they behave the same as the compressor of a refrigerator. Conceivably, you could hook those compressors directly to the component in a cycle; this just isn't done because it doesn't serve any particular benefit when a lab can just purchase LN, as it's fairly cheap stuff.
Yeah, that was the line that hurt me too.
As I mentioned in response to another one of your comments, no, you are mistaken. Derivative works are in the realm of copyright law, not trademark law. You can't copy a person's character and write stories about their further adventures and sell those stories. You can't draw a green frog that looks just like pepe, and name him pepe and sell him as an app. No trademark is needed for this protection. It is automatic.
You can draw pepe's on your hardware all you want. Copyright law says that unlicensed derivative works can't be _sold_ unless they fall within fair-use, and this doesn't.
Why do I get the feeling you enjoy the writings of Ayn Rand?
That's be a really weird codec to get that down to one byte. I'd recommend using a couple more than that.
They are, but think about GPUs in the late '90s, when this codec was finalized. I don't think there were any real-time rendered fractals going on in my copy of Unreal Tournament. You could have 2 GPU cores if you really wanted to spend the money, but as I understand it, the drivers weren't really out there to make particularly efficient use of them beyond simple stuff like allowing multiple monitors (I say "simple", that's a laugh, multiple monitor support was atrocious on Windows98, and it took me hours to get it working as expected on Redhat 6; I vaguely recall needing to compile a custom kernel, but don't quote me on that).
It wouldn't be nullifying the patent, but potentially, they could acquire a patent via Eminent Domain, so long as the provide Just Compensation. As far as I can find, there's no precedent for the US government entirely acquiring a patent in this manner. The closest I know of is that they made use of patents without licensing during WWII, then compensated after the fact.
I can't get too furious about this. I'm reading all the things that could go wrong in the comments and why it's a bad idea, and, it's not anything too terrifying. Cost of airfare is always gonna go up one way or another. If it goes up too much, then seats are empty, and airlines lose money; the market adjusts. And if this proposed company screws up, then oversight still drops in and starts making changes. I know it feels weird for Trump to support something that isn't the dumbest idea ever, but, how about we agree to this one, if you give the Democrats something in return for that support...y'know, the way politics used to behave?
Last I've seen, the ATCA hasn't decided if they like the idea or not.
It is possible to have strong concerns about carbon markets while still supporting the Paris Agreement.
Thankfully, that's not the major aspect of his climate-change legacy. His bigger legacy is likely going to be the increased grants and government loans for alternative energy endeavors. He's not the greatest president for environmental efforts (that would probably be Nixon so far), but given the powers granted to a president, Obama did OK.
Nuclear plants are massively expensive and take many years to build; during that time lots of things can halt the effort. This makes it a significantly risky investment. This means it is difficult to secure the money to build them. There are theoretical smaller-scale solutions in development, but they are not ready for fielding (and some may turn out to be impractical). Wind & Solar are safe investments. They can be built quickly, and they come with no public pushback, and very little Not-In-My-Backyard effect. so that's what's being pursued for now. As those realms become more saturated, the push for nuclear can become stronger.
I'm happy to be wrong. Please point me to papers or product-pages, or explain in greater detail? I fully understand that the watermarking algorithm would behave like a compression algorithm, but I'm under the impression that a crack could work on the pixel-level regardless.
Right, it's hard, but I don't believe it's impossible. That's more of what makes it interesting. If someone paid me to design that, I'd have a blast, but if I owned a company that is made or broken based on whether or not such a design worked...Let's just say I'm not buying in on the Screening Room IPO. I've read a couple patents on the problem, but as far as I've been able to find, there's no gold standard algorithm that has been proven in the field on this yet.
It's like a good hash that you have to spread across a video by slightly tweaking values for pixels. You're only allowed a narrow window of entropy, and you can only employ it on a smallish percentage of the total pixels or it becomes distracting. You can either pick the same pixels to change for every user, or you can pick different pixels every time. If it's different pixels every time, then a 3-way diff will reveal the appropriate value to choose to corrupt the plaintext. If it's the same pixels every time, then you isolate those pixels, and you interpolate their appropriate values based on the surrounding pixels. Generally, the place you would like hide this information is in the blacks because that is the least noticeable on anything except extremely high contrast devices. This is also where a lossy compression is going to remove information for exactly the same reason. When it's hidden here, taking the average works extremely well. When it isn't, then either the watermark is already gonna be table-flip distracting, or it's gonna be so uncommon that the average will be Good Enough.
I miss the days when PBS had enough federal funding that it didn't need to air the quasi-informercial junk that it has now. I mean, they always had the aggravating pledge drives, but they'd usually at least have good programming that went along with them.
I used to think that a liberal education program was stupid for technical fields. I thought it was dumb that I had to take all of these core classes that weren't related to Comp-Sci. But on reflection, and having dealt with people who experienced the fully technical schools, people who self-taught and had no formal higher education and others like myself who got a liberal 4-year Bachelor of Science, I really feel that it enriched me, both in my personal and professional life. I always hear people talking about how they never use what they learned in college...I use tons of what I learned in college. I wish the foreign language requirements weren't as harsh; I came in with 3 years of spanish in high school, and they still wanted me to take more. But beyond that, most everything I picked has helped me in life.
I was recently looking into the digital watermarking problem, and it's a fascinating bit of computer science. It's really tricky to sneak into a video with subtly adjusted pixels or audio information in such a way that it reveals the originator of the video. Many of them can be defeated just by doing a lossy compression. Getting something that survives that while not being distracting the viewer; ideally not even being visible is a fun little challenge to tackle.