This assumes that the current leaders in AI development will retain nearly all of the benefits from the use of AI. In particular, it assumes that all of the critical information about how to create AI models will be limited among a small number of software companies.
I see no reason why this should be the case. Creation and use of AI models is already being distributed more widely, e.g. through the open-source TensorFlow software. The basic algorithms of AI models are generally quite simple, and while they do require a bit of expertise to use well, they're usually not all that difficult. If you're willing to invest in the hardware, a relatively small number of experienced engineers, and good training data, it shouldn't be hard for most anybody to build their own AI models for their own applications.
Also, apparently their current demos are running on some severely beefy hardware (e.g. here), and they're only using raytracing for a portion of the scene. This may play a part in gaming eventually, and it's good that the APIs are getting out there, but it probably will be a while before it makes it to the next Call of Duty game.
The short answer is that it isn't brand-new. As the article mentions, nVidia has been doing real-time ray-tracing demos for about a decade. Various tricks and approximations are used to make this a reality. Game developers have largely shied away from it because similar results can be achieved with typically better performance using other methods.
This announcement, particularly with the involvement of Microsoft, indicates that the companies finally feel that the technology is mature enough to actually be used in a game. My guess is that it will likely still be some time before it is put to use. The fact that it was announced at the same time likely indicates that the three companies have been working together for some time behind closed-doors to agree upon the DirectX Raytracing API.
Large crashes are fundamentally unpredictable. If it were possible for anybody to reliably predict such crashes, they would not happen because investors with lots of money on the line would listen to those experts and plan for them, which would avert the crash.
So don't try to look to economists to determine whether or not a large crash is going to happen. What economists are best at is understanding how policy changes impact the economy. Sometimes that means increasing or decreasing the risk of crashes. Sometimes that means predicting how slow or fast a recovery will be given certain policy responses.
Granted, there are some really shit economists. But it's not all that hard to look at their past predictions to see if they're full of shit (e.g. some economists predicted runaway inflation would result from the Fed's quantitative easing policies, inflation which never appeared). Of course, everybody will be wrong sometimes, so it can take some work to separate the wheat from the chaff, but not as much as you might think.
I see the neckbeards are out in force on this thread. Not at all surprised.
Face it: Damore was trying to cloak discriminatory stereotypes in a veneer of science, drawing conclusions which have been widely discredited. This is pseudoscientific nonsense, and given the harm caused to other employees at Google, it more than justified his firing. There is no place for a man like Damore at any company that values its employees.
I am 10000x more worried about governments abusing power and criminals hacking into insecure systems than I am about people, even criminals, having secure communications.
Even in that case I'd have to say the answer is still no. There will be bugs that Microsoft isn't aware of, that game devs either intentionally or unintentionally exploit, which will always lead to discrepancies. The Windows APIs are just far too complicated to do the emulation precisely enough for it to always work.
For a simple kind of example, imagine a game dev writes a multi-threaded algorithm that has a concurrency bug. This concurrency bug remains undetected because in Windows, the various threads always execute in a certain order due to the way thread scheduling and timing work out. But the moment you port it to Linux, the thread scheduler is different, so the threads execute in a different order and the game deadlocks immediately.
This is just one possibility of many that can result in bugs when games run on an emulated Windows stack.
To highly just how absurdly difficult this is, notice that not even Microsoft can get it 100% right when trying to emulate older versions of Windows on modern versions. If you've ever tried to install really old games, you know how those edge cases can bite you. They do an impressively good job, but ultimately there are always edge cases that remain.
The Knowledge is certainly helpful. The problem is that there are idiosyncrasies to city traffic that are virtually impossible for algorithms to capture adequately, and the error rates on them are much too high for a cab driver to tolerate.
I do strongly feel that a good GPS application with up-to-date traffic information will be of tremendous help to a cab driver, but if they don't know the routes themselves, they're going to make mistakes. Sometimes the GPS app will not understand that you can't make a certain turn at a particular intersection. Sometimes it will misread traffic because one lane is far slower than another lane. If you know the streets well, it's often pretty easy to shave a few minutes off of the travel time, and know the best way to avoid big slowdowns if something like a car crash happens.
GPS apps are also often pretty terrible at the start and end of the journey. At the start, it may not realize which direction you're moving, or how best to reach the road if you're in a parking lot still. At the end of the journey, it may not know the best entrance, and if the best entrance is on a different street than the GPS thinks, it may require a significant detour that would be avoided by simply setting the destination properly. Knowing the city well enough for this can act as a good patch for these inadequacies.
In a busy city like London, it's typically pretty quick and easy to get a new fare, and short fares pay more per mile than long fares. So there isn't really any incentive to shortchange riders like this.
IQ is a largely racist construction that has long been used to "prove" that undesirables don't deserve to be treated as people.
The main problem is that they have an ideology which rejects critical thinking. They see universities as liberal brainwashing centers. Their own ideology has no checks and balances to reject bad ideas, making it virtually impossible to expel bad ideas once they get in. They believe that they are on a particular side in an ideological war, and their opponents are trying to cheat and lie to get them to leave. So they believe people who they see as being on their side (the side of God), and disbelieve everybody else.
So what we've had for decades now are politicians who have learned to convince these voters that they're on their team (typically using some combination of racist dog whistles and evangelical Christian rhetoric). Once they've convinced them that they're on their side, it becomes easy to convince them of all sorts of other things. Such as the idea that global warming is a myth, or the free market is good, or the social safety net is bad. None of these things have anything to do with their voters' core ideology, but they're popular among the Republican party's wealthy donors, and easy to foist off on the larger population of Republicans.
After many decades of this, the Republican party's platform has diverged so far from reality that it is utterly insupportable by facts. When the core of their system is a nest of lies, it becomes impossible to push their ideology with anything but fake news. Leading to stuff like this.
There is nothing physically or mentally wrong with Republicans. It's just that fake news is all they have because their entire political ideology is founded on huge lies.
My bet is it's more a matter of two things: 1) Trump is utterly ignorant of the boundaries of law for the presidency, such that he really doesn't understand that "you scratch my back and I scratch yours" behavior can be severely illegal, and 2) Trump's #1 guiding star is always his ego, and Putin was willing to stroke that ego.
My guess is that there has been a lot of direct collusion with the Russians from some of Trump's subordinates (it's the only explanation for the dismantling of State department I can see), but that Trump's knowledge and understanding of that collusion has been limited. He's probably been fed some oversimplified explanations of what's been going on, and probably supports the effort in the main, but hasn't had much direct interaction. My bet is his subordinates would prefer it that way: it keeps him in power, which keeps them protected (for now).
It will be interesting to see how it all shakes out as Mueller's probe comes to a head. I'd honestly be shocked if his subordinates who did the most collusion didn't take the care and effort to retain some evidence that pointed the finger directly at Trump. Trump's protection only matters as long as they themselves aren't targeted, and I doubt they're strong enough ideologues to continue following him once it no longer serves their interests. If Mueller's probe gets to them, they benefit greatly by pointing the finger up the chain, and can only do that effectively if they've retained some evidence. Trump's only hope is that his subordinates are just as incompetent as he is, or that the Republicans try something dramatic to halt the investigation (though that will only buy him until next year, if the midterm elections turn out like the special elections we had last year).
For not being a Trump supporter, you sure are saying a lot of things in support of Trump.
Trump may be saying things, "that a lot of people have wanted to say for some time," but you failed to specify which people. Those people he's been channeling? White supremacists and xenophobes with absolutely no understanding of how our government works, let alone how other nations work or how to talk to them.
These are "novel approaches" because everybody who had his position before him had ten times his sense, and most were far less racist (at least in the last century).
Intrinsic value, for it to have any meaning at all, should place some limits on the price of the item in question. For example, steel's usefulness as a raw material for a wide variety of goods places limits on its price: if the price rises too much, then manufacturers will seek other resources that can be substituted. If it falls too far, then producers will stop making steel.
Gold has no such limits. Its price is largely determined by how much people are willing to invest in gold. There's nothing that prevents its price from falling precipitously if there's any large-scale move away from buying gold. And its price can grow rapidly if there's a sudden increase in peoples' desire for gold.
Naturally, gold is extremely volatile. It always will be, as long as some significant fraction of people see it as a store of wealth.
Hindsight is easy with this kind of thing. The real problem is that nobody knows when the Bitcoin bubble is going to burst (which it most definitely will). People who make money on bubbles are those who get out before the bubble bursts, not once it's already begun.
So yes, you could have made a lot of money. But Bitcoin could also have burst not long after it hit $700 and you could have lost it all. The bubble could burst tomorrow. Or it could linger for another year and burst at $32,000. Or it could partially-burst, then grow again even bigger, then have a crash of monumental proportions. Ultimately, nobody knows, and it's almost impossible to tie Bitcoin to any fundamentals that could be used to judge whether its value is reasonable or not.
So by all means, invest in Bitcoin. But only do it with money you can afford to lose.
I do think this is most interesting if we don't allow the AI to adjust the game clock at all: if there's too large a delay between input and output, it will simply fail.
There have been AI's designed explicitly to play Super Mario World (see here, for instance). Where this becomes cool is with an AI that doesn't get an abstract representation of the level, but has to interpret the pixels that are displayed. That's a far more complicated problem, as the information displayed on even a simple game like Super Mario World is orders of magnitude richer than the information used to specify the state of a game of chess or go.
As for cars, I may be off, but I believe that they don't attempt to do this sort of "from scratch" learning, but rather break the problem up into pieces: train the neural net to detect things like cars or people or lines on the road, and then determine whether or not it expects those things to move and how. I imagine it's particularly difficult to train the neural net to deal with rare obstacles (such as bouncing debris from a wreck or a falling tree).
1. A heat pump is also an AC, and isn't much more expensive than a straight AC unit.
2. The heat pump is lower-cost for a good range of temperatures (mine is cheaper than a gas furnace for heating for anything above about 20F, if I recall).
That's precisely the way my system works. It combines a heat pump that doubles as an AC, and has a furnace for when temperatures become too low for the heat pump to operate efficiently. It also has a calculator on the thermostat to set the temperature where the system switches between gas furnace and electric heat pump based upon my gas/electricity costs.
Yeah, burying pipes is great if you have large temperature swings. Even a few feet down, the temperature swings throughout the year are dampened dramatically (a few feet below the ground it's much cooler in summer and warmer in winter than the surface). But if you live in a more temperate clime, there just isn't a need for it.
Right now I pay roughly $100/mo average to heat/cool my house. It would probably cost something along the lines of $10,000 to get a system with heat exchange through buried pipes installed. If it cut my heating/cooling bill in half, then it would take some 17 years to repay, by which time the system might need expensive maintenance or replacement. And given that the temperatures don't swing all that dramatically here, cutting my bill in half might be a very generous estimate.
Heat pumps, in the right conditions, can be far more efficient than furnaces. With a furnace, your efficiency will typically be around 90%-95%, depending upon how well the system is insulated. With a heat pump, you can get efficiency that is ten times that under the right conditions, meaning you may only need 10BTU of energy to transfer 100BTU of heat. The exact efficiency of the heat pump system varies greatly depending upon how it's designed, the temperature of the system it's exchanging heat with, and how efficient the transfer of heat is (e.g. if it exchanges heat with a reservoir of water it'll be more efficient than if it exchanges heat with the air).
Last year I had a heat pump system installed in my house which cut by half the power cost of heating the house in winter. There's still a furnace for backup if the temperatures dip too low (which has happened), and there's never been a situation where the heating system lacked sufficient power to keep the house heated.
The point is that Bitcoin doesn't solve any pressing problem (currencies already exist), but it's soaking up more and more computing power as miners jump on the bandwagon to make more money, aided and abetted by speculators who continue to drive the price up. Bitcoin transactions are already far more expensive in terms of their raw cost than most traditional transaction schemes.
If Bitcoin transactions were on par in terms of resource cost to more traditional transaction systems, such as credit card payments or bank account transfers, then it might have a place in financial transactions. But we're not living in that world. Instead, if the enthusiasm for Bitcoin continues, we're looking at a world where it eats up a huge fraction of total power usage to no benefit for humanity.
Gold-based currencies had this problem, only Bitcoin's issue is far worse. Typically it's not so much that the currency fails as the economy it's based upon collapses due to deflation.
This is part of the reason why the Great Depression was so bad, for instance. Japan has been struggling with deflation over the last couple of decades as well, resulting in a long period of economic stagnation. The smaller, poorer countries that share the Euro currency are also dealing with a deflationary problem right now.
But in all of these cases, the magnitude of the deflation is nowhere remotely close to the deflation with Bitcoin. Real-world factors tend to prevent national currencies from deflating nearly that rapidly. If Bitcoin were actually used as the national currency of a country, that country's economy would be experiencing unprecedented economic pain.
National currencies usually are fine with modest amounts of inflation, but cause economic disruption if inflation is too low (or negative) or much too high. As long as inflation is stable and of a modest amount, the currency works pretty well.
Bitcoin miners perform the processing required to process transactions. When a bitcoin transaction is submitted, it is submitted along with a "bid" for how much the person is willing to pay for the transaction. Miners then prioritize incoming transactions based upon how much is bid. If the bid is too low, it will end up at the end of a long queue and might take hours to finish (if ever).
The optimal bid amount, then, is determined by the computing power of the Bitcoin network combined with how much processing power it actually takes to process each transaction and how many transactions there are. Right now, bids above about 150 Satoshis per byte (which works out to just under $5 per transaction on average at current prices) finish relatively quickly (typically under 30 minutes). Bids below that take an increasingly long time to commit.
This high transaction cost is a function of the poor scaling of the Bitcoin algorithm to large numbers of transactions. It's one reason among many why I think Bitcoins are an absolutely abysmal medium of exchange, and why I question the entire concept of a blockchain-based currency system.
I'd love to see if this AI can learn to play a more complicated game like Super Mario World given only:
1) The pixels displayed as input.
2) Fail conditions (when a life is lost).
3) Basic map navigation rules (bonus if these can be eliminated and the game can be judged only on whether or not it gets a game over or completes the final level).
4) Valid controller inputs.
I do wonder how this AI would translate from the turn-based world of Chess and Go to realtime.
I'd be wiling to bet that over a decade of languishing economy is the most significant factor in suicide rates, not their fastidiousness. If you look at their suicide rate over time, there was a huge jump in 1998, which was when Japan's GDP hit bottom after the 1996 crash (their economy has languished in a low-growth state ever since).
Social pressures might have influenced the strength of the link between the economy and suicides, but the driving factor really was the economy, for the most part. Note that the US has shown increases in deaths related to drugs, alcohol, and suicide since 2000, which is likely due to persistent economic weakness for middle and lower-class citizens over this period.
Slashdot Mirror
This assumes that the current leaders in AI development will retain nearly all of the benefits from the use of AI. In particular, it assumes that all of the critical information about how to create AI models will be limited among a small number of software companies.
I see no reason why this should be the case. Creation and use of AI models is already being distributed more widely, e.g. through the open-source TensorFlow software. The basic algorithms of AI models are generally quite simple, and while they do require a bit of expertise to use well, they're usually not all that difficult. If you're willing to invest in the hardware, a relatively small number of experienced engineers, and good training data, it shouldn't be hard for most anybody to build their own AI models for their own applications.
Also, apparently their current demos are running on some severely beefy hardware (e.g. here), and they're only using raytracing for a portion of the scene. This may play a part in gaming eventually, and it's good that the APIs are getting out there, but it probably will be a while before it makes it to the next Call of Duty game.
The short answer is that it isn't brand-new. As the article mentions, nVidia has been doing real-time ray-tracing demos for about a decade. Various tricks and approximations are used to make this a reality. Game developers have largely shied away from it because similar results can be achieved with typically better performance using other methods.
This announcement, particularly with the involvement of Microsoft, indicates that the companies finally feel that the technology is mature enough to actually be used in a game. My guess is that it will likely still be some time before it is put to use. The fact that it was announced at the same time likely indicates that the three companies have been working together for some time behind closed-doors to agree upon the DirectX Raytracing API.
Large crashes are fundamentally unpredictable. If it were possible for anybody to reliably predict such crashes, they would not happen because investors with lots of money on the line would listen to those experts and plan for them, which would avert the crash.
So don't try to look to economists to determine whether or not a large crash is going to happen. What economists are best at is understanding how policy changes impact the economy. Sometimes that means increasing or decreasing the risk of crashes. Sometimes that means predicting how slow or fast a recovery will be given certain policy responses.
Granted, there are some really shit economists. But it's not all that hard to look at their past predictions to see if they're full of shit (e.g. some economists predicted runaway inflation would result from the Fed's quantitative easing policies, inflation which never appeared). Of course, everybody will be wrong sometimes, so it can take some work to separate the wheat from the chaff, but not as much as you might think.
I see the neckbeards are out in force on this thread. Not at all surprised.
Face it: Damore was trying to cloak discriminatory stereotypes in a veneer of science, drawing conclusions which have been widely discredited. This is pseudoscientific nonsense, and given the harm caused to other employees at Google, it more than justified his firing. There is no place for a man like Damore at any company that values its employees.
I am 10000x more worried about governments abusing power and criminals hacking into insecure systems than I am about people, even criminals, having secure communications.
Even in that case I'd have to say the answer is still no. There will be bugs that Microsoft isn't aware of, that game devs either intentionally or unintentionally exploit, which will always lead to discrepancies. The Windows APIs are just far too complicated to do the emulation precisely enough for it to always work.
For a simple kind of example, imagine a game dev writes a multi-threaded algorithm that has a concurrency bug. This concurrency bug remains undetected because in Windows, the various threads always execute in a certain order due to the way thread scheduling and timing work out. But the moment you port it to Linux, the thread scheduler is different, so the threads execute in a different order and the game deadlocks immediately.
This is just one possibility of many that can result in bugs when games run on an emulated Windows stack.
To highly just how absurdly difficult this is, notice that not even Microsoft can get it 100% right when trying to emulate older versions of Windows on modern versions. If you've ever tried to install really old games, you know how those edge cases can bite you. They do an impressively good job, but ultimately there are always edge cases that remain.
The Knowledge is certainly helpful. The problem is that there are idiosyncrasies to city traffic that are virtually impossible for algorithms to capture adequately, and the error rates on them are much too high for a cab driver to tolerate.
I do strongly feel that a good GPS application with up-to-date traffic information will be of tremendous help to a cab driver, but if they don't know the routes themselves, they're going to make mistakes. Sometimes the GPS app will not understand that you can't make a certain turn at a particular intersection. Sometimes it will misread traffic because one lane is far slower than another lane. If you know the streets well, it's often pretty easy to shave a few minutes off of the travel time, and know the best way to avoid big slowdowns if something like a car crash happens.
GPS apps are also often pretty terrible at the start and end of the journey. At the start, it may not realize which direction you're moving, or how best to reach the road if you're in a parking lot still. At the end of the journey, it may not know the best entrance, and if the best entrance is on a different street than the GPS thinks, it may require a significant detour that would be avoided by simply setting the destination properly. Knowing the city well enough for this can act as a good patch for these inadequacies.
In a busy city like London, it's typically pretty quick and easy to get a new fare, and short fares pay more per mile than long fares. So there isn't really any incentive to shortchange riders like this.
No, no it is not.
IQ is a largely racist construction that has long been used to "prove" that undesirables don't deserve to be treated as people.
The main problem is that they have an ideology which rejects critical thinking. They see universities as liberal brainwashing centers. Their own ideology has no checks and balances to reject bad ideas, making it virtually impossible to expel bad ideas once they get in. They believe that they are on a particular side in an ideological war, and their opponents are trying to cheat and lie to get them to leave. So they believe people who they see as being on their side (the side of God), and disbelieve everybody else.
So what we've had for decades now are politicians who have learned to convince these voters that they're on their team (typically using some combination of racist dog whistles and evangelical Christian rhetoric). Once they've convinced them that they're on their side, it becomes easy to convince them of all sorts of other things. Such as the idea that global warming is a myth, or the free market is good, or the social safety net is bad. None of these things have anything to do with their voters' core ideology, but they're popular among the Republican party's wealthy donors, and easy to foist off on the larger population of Republicans.
After many decades of this, the Republican party's platform has diverged so far from reality that it is utterly insupportable by facts. When the core of their system is a nest of lies, it becomes impossible to push their ideology with anything but fake news. Leading to stuff like this.
There is nothing physically or mentally wrong with Republicans. It's just that fake news is all they have because their entire political ideology is founded on huge lies.
I believe we called you racists because you kept claiming white people are better than everybody else. Funny, that.
I doubt that's the main inroad the Russians had.
My bet is it's more a matter of two things: 1) Trump is utterly ignorant of the boundaries of law for the presidency, such that he really doesn't understand that "you scratch my back and I scratch yours" behavior can be severely illegal, and 2) Trump's #1 guiding star is always his ego, and Putin was willing to stroke that ego.
My guess is that there has been a lot of direct collusion with the Russians from some of Trump's subordinates (it's the only explanation for the dismantling of State department I can see), but that Trump's knowledge and understanding of that collusion has been limited. He's probably been fed some oversimplified explanations of what's been going on, and probably supports the effort in the main, but hasn't had much direct interaction. My bet is his subordinates would prefer it that way: it keeps him in power, which keeps them protected (for now).
It will be interesting to see how it all shakes out as Mueller's probe comes to a head. I'd honestly be shocked if his subordinates who did the most collusion didn't take the care and effort to retain some evidence that pointed the finger directly at Trump. Trump's protection only matters as long as they themselves aren't targeted, and I doubt they're strong enough ideologues to continue following him once it no longer serves their interests. If Mueller's probe gets to them, they benefit greatly by pointing the finger up the chain, and can only do that effectively if they've retained some evidence. Trump's only hope is that his subordinates are just as incompetent as he is, or that the Republicans try something dramatic to halt the investigation (though that will only buy him until next year, if the midterm elections turn out like the special elections we had last year).
For not being a Trump supporter, you sure are saying a lot of things in support of Trump.
Trump may be saying things, "that a lot of people have wanted to say for some time," but you failed to specify which people. Those people he's been channeling? White supremacists and xenophobes with absolutely no understanding of how our government works, let alone how other nations work or how to talk to them.
These are "novel approaches" because everybody who had his position before him had ten times his sense, and most were far less racist (at least in the last century).
That's not what intrinsic value means.
Intrinsic value, for it to have any meaning at all, should place some limits on the price of the item in question. For example, steel's usefulness as a raw material for a wide variety of goods places limits on its price: if the price rises too much, then manufacturers will seek other resources that can be substituted. If it falls too far, then producers will stop making steel.
Gold has no such limits. Its price is largely determined by how much people are willing to invest in gold. There's nothing that prevents its price from falling precipitously if there's any large-scale move away from buying gold. And its price can grow rapidly if there's a sudden increase in peoples' desire for gold.
Naturally, gold is extremely volatile. It always will be, as long as some significant fraction of people see it as a store of wealth.
Hindsight is easy with this kind of thing. The real problem is that nobody knows when the Bitcoin bubble is going to burst (which it most definitely will). People who make money on bubbles are those who get out before the bubble bursts, not once it's already begun.
So yes, you could have made a lot of money. But Bitcoin could also have burst not long after it hit $700 and you could have lost it all. The bubble could burst tomorrow. Or it could linger for another year and burst at $32,000. Or it could partially-burst, then grow again even bigger, then have a crash of monumental proportions. Ultimately, nobody knows, and it's almost impossible to tie Bitcoin to any fundamentals that could be used to judge whether its value is reasonable or not.
So by all means, invest in Bitcoin. But only do it with money you can afford to lose.
I do think this is most interesting if we don't allow the AI to adjust the game clock at all: if there's too large a delay between input and output, it will simply fail.
There have been AI's designed explicitly to play Super Mario World (see here, for instance). Where this becomes cool is with an AI that doesn't get an abstract representation of the level, but has to interpret the pixels that are displayed. That's a far more complicated problem, as the information displayed on even a simple game like Super Mario World is orders of magnitude richer than the information used to specify the state of a game of chess or go.
As for cars, I may be off, but I believe that they don't attempt to do this sort of "from scratch" learning, but rather break the problem up into pieces: train the neural net to detect things like cars or people or lines on the road, and then determine whether or not it expects those things to move and how. I imagine it's particularly difficult to train the neural net to deal with rare obstacles (such as bouncing debris from a wreck or a falling tree).
1. A heat pump is also an AC, and isn't much more expensive than a straight AC unit.
2. The heat pump is lower-cost for a good range of temperatures (mine is cheaper than a gas furnace for heating for anything above about 20F, if I recall).
That's precisely the way my system works. It combines a heat pump that doubles as an AC, and has a furnace for when temperatures become too low for the heat pump to operate efficiently. It also has a calculator on the thermostat to set the temperature where the system switches between gas furnace and electric heat pump based upon my gas/electricity costs.
Yeah, burying pipes is great if you have large temperature swings. Even a few feet down, the temperature swings throughout the year are dampened dramatically (a few feet below the ground it's much cooler in summer and warmer in winter than the surface). But if you live in a more temperate clime, there just isn't a need for it.
Right now I pay roughly $100/mo average to heat/cool my house. It would probably cost something along the lines of $10,000 to get a system with heat exchange through buried pipes installed. If it cut my heating/cooling bill in half, then it would take some 17 years to repay, by which time the system might need expensive maintenance or replacement. And given that the temperatures don't swing all that dramatically here, cutting my bill in half might be a very generous estimate.
Heat pumps, in the right conditions, can be far more efficient than furnaces. With a furnace, your efficiency will typically be around 90%-95%, depending upon how well the system is insulated. With a heat pump, you can get efficiency that is ten times that under the right conditions, meaning you may only need 10BTU of energy to transfer 100BTU of heat. The exact efficiency of the heat pump system varies greatly depending upon how it's designed, the temperature of the system it's exchanging heat with, and how efficient the transfer of heat is (e.g. if it exchanges heat with a reservoir of water it'll be more efficient than if it exchanges heat with the air).
Last year I had a heat pump system installed in my house which cut by half the power cost of heating the house in winter. There's still a furnace for backup if the temperatures dip too low (which has happened), and there's never been a situation where the heating system lacked sufficient power to keep the house heated.
The point is that Bitcoin doesn't solve any pressing problem (currencies already exist), but it's soaking up more and more computing power as miners jump on the bandwagon to make more money, aided and abetted by speculators who continue to drive the price up. Bitcoin transactions are already far more expensive in terms of their raw cost than most traditional transaction schemes.
If Bitcoin transactions were on par in terms of resource cost to more traditional transaction systems, such as credit card payments or bank account transfers, then it might have a place in financial transactions. But we're not living in that world. Instead, if the enthusiasm for Bitcoin continues, we're looking at a world where it eats up a huge fraction of total power usage to no benefit for humanity.
Gold-based currencies had this problem, only Bitcoin's issue is far worse. Typically it's not so much that the currency fails as the economy it's based upon collapses due to deflation.
This is part of the reason why the Great Depression was so bad, for instance. Japan has been struggling with deflation over the last couple of decades as well, resulting in a long period of economic stagnation. The smaller, poorer countries that share the Euro currency are also dealing with a deflationary problem right now. But in all of these cases, the magnitude of the deflation is nowhere remotely close to the deflation with Bitcoin. Real-world factors tend to prevent national currencies from deflating nearly that rapidly. If Bitcoin were actually used as the national currency of a country, that country's economy would be experiencing unprecedented economic pain.
National currencies usually are fine with modest amounts of inflation, but cause economic disruption if inflation is too low (or negative) or much too high. As long as inflation is stable and of a modest amount, the currency works pretty well.
Bitcoin miners perform the processing required to process transactions. When a bitcoin transaction is submitted, it is submitted along with a "bid" for how much the person is willing to pay for the transaction. Miners then prioritize incoming transactions based upon how much is bid. If the bid is too low, it will end up at the end of a long queue and might take hours to finish (if ever).
The optimal bid amount, then, is determined by the computing power of the Bitcoin network combined with how much processing power it actually takes to process each transaction and how many transactions there are. Right now, bids above about 150 Satoshis per byte (which works out to just under $5 per transaction on average at current prices) finish relatively quickly (typically under 30 minutes). Bids below that take an increasingly long time to commit.
This high transaction cost is a function of the poor scaling of the Bitcoin algorithm to large numbers of transactions. It's one reason among many why I think Bitcoins are an absolutely abysmal medium of exchange, and why I question the entire concept of a blockchain-based currency system.
I'd love to see if this AI can learn to play a more complicated game like Super Mario World given only: 1) The pixels displayed as input. 2) Fail conditions (when a life is lost). 3) Basic map navigation rules (bonus if these can be eliminated and the game can be judged only on whether or not it gets a game over or completes the final level). 4) Valid controller inputs. I do wonder how this AI would translate from the turn-based world of Chess and Go to realtime.
I'd be wiling to bet that over a decade of languishing economy is the most significant factor in suicide rates, not their fastidiousness. If you look at their suicide rate over time, there was a huge jump in 1998, which was when Japan's GDP hit bottom after the 1996 crash (their economy has languished in a low-growth state ever since).
Social pressures might have influenced the strength of the link between the economy and suicides, but the driving factor really was the economy, for the most part. Note that the US has shown increases in deaths related to drugs, alcohol, and suicide since 2000, which is likely due to persistent economic weakness for middle and lower-class citizens over this period.