Yes, I understand that the peer review and publication process has to be paid for, but restricting access to the fruits of scientific progress -- and therefore also limiting further progress! -- is the wrong way to do it.
My guess is that we'll transition to an "author pays" model. Researchers employed by institutions will have their fees covered by their employer. Researchers who don't have that option are already disadvantaged under the current model, so the fact that they'll still be disadvantaged isn't so terrible. Plus they'll still be able to publish in free online archives that accept non-reviewed and unedited work. Really good work should find it fairly easy to get someone to fund the peer review and editing required to get it into a journal.
Basically, Nissan took it in the shorts for hugely overestimating the residual value of the car when they leased it to me. I think the original lease agreement assumed a residual of over $20K, but the market value had plummeted to barely a third of that by the end of the lease plus a one year extension.
I think the car would have been an okay deal even without their serious miscalculation. With it? What a steal.
Anyone who has substantial amounts of wealth (and a clue what to do with it) has almost all of it invested or tied up in land or other valuable assets.
In the case of Bezos and Gates, essentially all of it is in AMZN and MSFT stock, respectively. Which is why their net worth is tied to the share prices. Were they to try to cash out, their massive stock sales (and the lack of confidence indicated by their decision to cash out), would cause the stock price to drop. So their actual available cash is less than their net worth. OTOH, they can easily borrow billions with their stock as collateral.
Either it is money or it is glorified pokemon cards. You can't have it both ways.
I've pointed that out many times over the years... it's a deep flaw in the basic philosophy of crypto-currency - they want it both ways. They want legitimacy, *and* they want complete freedom with no controls or regulation. They don't grasp that the latter is a necessary consequence of the former.
Some probably do want exactly what you say. Others recognize that as a financial instrument regulation is inevitable, but still see value because although it's subject to regulation, the supply of a cryptocurrency doesn't have to be subject to any central control, unlike fiat currencies.
I don't think that's TCO you're using. Those are operational expenses. The T stands for Total, which would include purchase price, taxes, registration fees, etc...
I did neglect taxes and registration fees, but included purchase price.
Your argument is circular. You assert that someone is an expert if and only if they can write secure C code. My assertion is based not on logic (circular or otherwise) but on data. What can we possibly conclude when security-focused portions of widely-scrutinize, mission-critical operating system kernels; cryptographic libraries; security-focused "trusted" operating systems; secure container systems; and the like, all written by top security-focused software engineers who are doing their level best to make their code airtight, consistently have vulnerabilities which would not be possible in a more secure language?
The only possible conclusion is: writing secure software in C is too hard.
I'm fully aware of the techniques one uses to avoid buffer overflows, stack smashing, heap corruption, etc. They're easy to describe and in the abstract it seems like they should be simple to consistently implement. Heck in many cases correctness can even be automatically validated. So it seems like this should be easy. Massive amounts of real-world experience proves otherwise.
I'll even grant that it's possible that somewhere there are C programmers who are good enough that they never write such bugs. But obviously there aren't enough of them to write all of the security-sensitive software that we need written.
I'm buying a new car, and just ran the numbers for a volt
Don't forget to factor in maintenance costs. EVs are much simpler with fewer moving parts.
A Volt isn't. It's more complex with more moving parts, since it has all the complexity of an ICEV, plus all of a battery EV, plus all of the complexity for transitioning between the two.
A BEV is much simpler, though. And can be *much* cheaper. I expect that my TCO for owning my originally factory-new Nissan LEAF for 200K miles will be about $20K, or $0.10 per mile.
Granted, I kind of lucked out on the cost of the car. I leased it in 2013. At the time, it was a $37K car ($30K after tax credits, which went to the lessor), compared to $40K for a Chevy Volt. But with all the new greater-range EVs coming out now, the value of those older LEAFs has dropped like a rock, so I was able to buy it off lease for $6K. Three years of lease payments added up to just under $7K, so my cost for the vehicle has been just under $13K, including the interest I paid on the lease (I paid cash when I bought it off lease).
I average 4.2 miles per kWh, and pay about $0.11 per kWh, so my energy cost is 2.6 cents per mile. Assuming $2.50 per gallon, I'd have to get 95 miles per gallon in an ICEV to get the same per-mile energy cost.
Then there's maintenance. Or not. I've put 50K miles on my car and the only maintenance I've had to do is $200 for a set of new tires. Oh, and refill the windshield wiper fluid reservoir. Brakes are still great (EV brakes last longer than ICEV brakes, because much braking is regenerative). Oil changes aren't a thing.
Looking into the future, I can expect to replace the tires at normal intervals, and eventually it'll need brake pads. There is no transmission. No clutch. There are no spark plugs or wires, no catalytic converter or other emissions system, no muffler. At some point I'll need to replace the 12V lead acid battery (normal auto battery).
Of course, eventually the big lithium ion battery will need to be replaced. So far, in 4 years and 50K miles my battery has lost 3% of its capacity. It's warrantied to maintain 80% for 100K miles, but judging by the experience of other LEAF owners in fairly cool climates like mine (very hot climates are a different story!), it's more likely that it'll go 200K miles before it falls to 80%.
When the battery does need to be replaced, I'll be looking at a cost of about $5K, which is pretty comparable to the cost of a new engine for a small car, and quite a bit less than the cost of a new engine & transmission. Plus, by the time I buy a new battery pack, the replacement will almost certainly be higher-capacity and longer-lasting.
Anyway, assuming I drive the car until the battery needs to be replaced at, say, 200K miles, and then junk the car, I'll have spent around $20K, which is a TCO of $0.10 per mile. To go beyond 200K I'll probably looking at CV joints and similar components in addition to a battery, but it seems reasonable that I could get another 200K miles for an additional $7K or so in repairs (incl. battery), plus another $5K in electricity, so about $0.06 per mile.
The downside, of course, is that this is not a road trip vehicle. It's great for running around town, though. And it's a lot more fun to drive than comparable ICEVs.
I do believe it, and here's why: the vast majority of security holes in C code are buffer overruns and failing to sanitize inputs. These are both easily mitigable.
And yet, they so often aren't. Even in code written and reviewed by security experts. You're supporting my point.
having "unique" plug types for particular purposes is a *feature*, not a bug - simply by looking at the plug, we know what the cable and the port does
You're trying to make a virtue of a problem. The right solution is that the cable and port should do everything. There's no need to figure out what it does or doesn't do, because anything the devices in question are capable of doing can be done via that port and plug.
So you see a Type C plug - is it Thunderbolt or not? Is it a DisplayPort?
Yes, it's all of the above, assuming it makes sense that the device can be a source/sink of the relevant data.
What voltages/amps can it provide?
The cable (unless it's non-compliant crap; don't buy non-compliant crap) can handle the full range of voltages and amperages in the specification.
This is incredibly convenient. My laptop, phone and tablet all have USB-C ports, for both data and power. I now travel with only one power adapter, the one for my laptop, which I can also use for my phone and tablet. I also carry a USB-C to USB-C cable, which enables me to charge any of the devices from any other device. Charging my laptop from my phone would be silly, charging it from the tablet is slightly less so and I have actually had occasion to do exactly that. Charging tablet from phone or phone from tablet both make sense sometimes, and I do both often when traveling.
I also occasionally use my phone charger to power my laptop. It doesn't provide enough power to maintain the battery level while I'm using it, but it does stretch battery life, and will recharge the laptop if left connected overnight. I do this mainly because I have a phone charger plugged in next to my bed. I'm waiting for wall outlets like this one to become better and cheaper, and then I'll replace many of my home outlets with them, mostly eliminating the need for wall warts.
That will be a great day. Actually, I hope that before too long we start building DC power distribution networks in homes, the way data centers have gone. At moderate voltages (e.g. 48V) resistive line losses are easily offset by the power savings provided by larger, more efficient central converters. Then each wall outlet should have a couple of simple step-down transformers to provide 20V and 5V over USB, per the power delivery specification.
While we're at it, we should add network over USB as well. So any device plugged into any wall outlet with a simple USB-C to USB-C cable will have both power and data (with 802.1X authentication for security). 20 Gbps should be enough bandwidth for the next couple of decades.
just look at e.g. the many forum posts of people who connected "the wrong type" of USB-C-to-HDMI connectors because they didn't know their USB C wasn't the USB C that they thought USB C was supposed to be.
Yes, there is crap hardware being sold. The USB implementer's forum needs to start enforcing its trademark and shutting down sales of non-compliant crap. In the absence of that, just make sure that the stuff you're buying has been tested by someone who knows what they're doing (on Amazon, check for Benson Leung's comments). No, this state of affairs is not good, but it's an implementation problem, not a problem with the fundamental concept of one plug that does everything.
Yeah, well allow me to say that day looks a long way off to me.
You have no rational basis for that statement. Lacking understanding of what intelligence is, we have no idea what is necessary to create it. We we could hit on the key idea underlying the theory of intelligence today, or it could take a century of slogging incremental progress. We don't know what we don't know.
I don't think it's necessary that we understand in detail how the human brain works. I think it is necessary that we have a solid theory of how intelligence works, though. Given that theory, we will be well-equipped to understand how the brain works, but there would still be a lot of work to map out the details.
Whether we'll discover the theory from work on artificial intelligence or from the study of natural intelligence, or some combination of the two, I don't think anyone can say. My guess it that it will be a combination AI research and study of natural intelligence that will yield the crucial insights. How far we are from obtaining those insights is unknowable.
To rival the human brain (in terms of intuition, not speed or capacity), you need to first be able to duplicate the human brain.
You're implicitly assuming that the way our brains are built is the only way to construct intelligence. I see no rational basis for such an assumption.
I do think that we're not likely to build AI until we understand more deeply what intelligence is, and that that theory will likewise help us understand how our own brains work. But there's no reason to believe that evolution produced the only possible form.
I took an AI course in university and the professor would say that AI has been right around the corner for 40-years....
Yes... that's proof that we have no idea how far we are from creating artificial general intelligence. We never have known how far away it is, because we don't know what intelligence is or how it works.
But "We don't know how far away it is" does not imply "It's many years away". It just means we don't know. We could have the crucial breakthrough this afternoon and find all of our computers controlled by a global Internet-based AI by supper time, or we could be 100 years away. We just can't know until we understand intelligence, and by the time we understand it we'll be able to build it.
But a mad man wouldn't be working on so many "random" things that all lead to the salvation of mankind.
You don't know that any of those things will lead to the salvation of mankind. It seems plausible that all are important to our future, but you can't know.
Maybe, maybe not. Perhaps there's just one crucial idea that needs to be discovered to enable artificial general intelligence (AGI). Perhaps there are a whole bunch of incremental developments in both hardware and software that are required. We don't know, and we won't know until we get there.
Given that, we really should spend considerable time and resources on thinking about how to prepare for the day when we do figure out how to create AGI. Because what does seem quite likely is that the human brain is not the pinnacle of possible intelligence. That in turn means that AGI, once developed, is likely to quickly surpass human ability in the same way and for the same reasons that machines are stronger, faster, tougher, etc., than humans. So, at some point in the future -- and how distant that point is we do not know and cannot guess -- we will be sharing our planet with entities who are orders of magnitude smarter than we are, which means that if our goals conflict with their goals, we'll lose.
AGI poses both boundless opportunity and boundless risk to humanity. If we can convince the superintelligences to work on our behalf and according to what we value, they can help us solve all of our problems and move the human race into a near Utopia. If we can't, we'll probably cease to exist.
Musk is right to be concerned about this. He's wrong to think that we should be making regulations, because we have absolutely no idea what regulations to make.
Zuckerberg is wrong not to be concerned about it. Oh, he's right that our current "AI" technology is so far from being AGI that it's not worth worrying about. But Musk isn't talking about current AI, he's talking about the AGI that will come at some unknown point in the future.
Trust me our understanding of Natural Intelligence is limited.
You mean "our understanding of Intelligence is limited".
Once we understand intelligence, we'll be able to build one. Attempting to build intelligence will likely be an essential element of gaining that understanding, as will the study of "natural" intelligence.
Unrelated question: how would you respond to a tax cut on businesses and the wealthy if it also came with a remediation of the welfare system, strengthening of the middle-class, and a complete elimination of homelessness and hunger through a powerful social safety net? Assume the math works out (it does).
Do I also get a pony? And elimination of the federal debt?
That makes perfect sense. OTOH having been fired by Trump may be a sign of qualification in the future.
Well... it also means you got hired by Trump, which may be a sign of disqualification. i suspect that the best option is never to have been associated with Trump in any way.
I hope Spicer goes to CNN or MSNBC and gets to sound off on Trump.
That would be awesome, but Spicer would have to be a complete idiot to do it while Trump is still in office, and maybe even after. Trump has a vindictive streak a mile wide and the sort of "disloyalty" that would constitute in his eyes would provoke a really dangerous reaction.
Yes, exactly the same if you want to ignore the fact that corporations are taxed based on their profits while US citizens are taxed based on their revenue (income). I'd love to be able subtract my mortgage, transportation, food, and utilities costs from my taxable income as operating costs and only be taxed based on what's left.
You don't actually understand your own taxes, I see. You do get to subtract all that stuff, effectively. You absolutely get to subtract your mortgage interest, which is the expense part. The principle payment is net income; it directly increases your net worth. As for the rest of the stuff, that's what standard deductions are for, to account for the basic cost of living, so you pay taxes only on your "extra". And that's actually also part of the justification behind the progressive scale; the less you make the more of your income must go to non-luxury expenses.
What you can't do is decrease your taxable income by choosing to buy more expensive stuff. Corporations can do this, but they have strong motivation not to, because their reason for existence is to generate value for shareholders. Being wasteful, whether it's to reduce tax liability or just to keep employees in unnecessary luxury, is contrary to the shareholders' goals and they'll eventually force the corporation to get more efficient and generate more income.
Correct. And note that it applies only to C corporations, not S corps, LLCs, etc.
Passing costs on to the customer is impossible here, since if the company could make more money by raising prices they would have.
This would be true if all of their competitors weren't subject to the same costs. An individual company can't impose its own costs on customers at will, but when the cost is imposed on the entire industry, competition doesn't prevent it. Foreign competition has an entirely different structure, of course, but countries balance those differences with tariffs.
I'll grant that most economists believe there is enough variation in corporate structure to prevent costs from being pushed on consumers. I'm not sure that's true in all industries, though, especially the capital-intensive industries where C corps are really the only ones able to play.
Exactly the same argument explains how the cost can land on employees. Competition for wages, etc., doesn't prevent it because it applies to the entire industry. There's another anti-labor dynamic as well; if corporate taxes do hit owners of capital, reducing their rate of return, they will move the capital to other industries or other countries, reducing investment in the relevant domestic industries, which eventually reduces competition for workers in those industries, lowering wages.
Profit is what goes to the shareholders, in the form of dividends or increased value of the company, and so if you were going to tax individuals you'd have to tax shareholders. Of course, many shareholders aren't individuals, but rather mutual funds and other companies. Taxing the ultimate recipients gets difficult.
Mutual funds and other companies just push it back a level. At the end of the chain there are always individual people. There is absolutely no disagreement on this among economists. You won't find anyone who seriously argues that corporations ultimately pay taxes.
There's nothing evil or deceptive about corporate income tax. It's a tax on corporate income, which obviously means on direct and indirect shareholders.
Even if it were a tax on shareholders, it would be more transparent and honest to collect it via capital gains taxes. But it's not a tax only on shareholders. It's difficult to tease out where corporate taxes actually land, but the general consensus among economists is that it's primarily split between shareholders and employees; capital and labor. The most common view is that capital bears 80% of the cost and labor 20%, but there are also good models that show that labor bears 70+%.
The bottom line to me is that there's really no point in trying to figure out who actually pays corporate taxes. Just figure out who you actually want to tax, and tax them. If you want it all to land on owners of capital, tax capital gains. If you want to apportion 80% to capital and 20% to labor, do that. If you want to put it all on consumers, do that, via sales taxes.
Corporate taxes are evil because they appear to voters to be taxes on no one, free money, but this is patently false.
Slashdot Mirror
Paywalled research is just a bad idea.
Yes, I understand that the peer review and publication process has to be paid for, but restricting access to the fruits of scientific progress -- and therefore also limiting further progress! -- is the wrong way to do it.
My guess is that we'll transition to an "author pays" model. Researchers employed by institutions will have their fees covered by their employer. Researchers who don't have that option are already disadvantaged under the current model, so the fact that they'll still be disadvantaged isn't so terrible. Plus they'll still be able to publish in free online archives that accept non-reviewed and unedited work. Really good work should find it fairly easy to get someone to fund the peer review and editing required to get it into a journal.
It was an awesome deal :-)
Basically, Nissan took it in the shorts for hugely overestimating the residual value of the car when they leased it to me. I think the original lease agreement assumed a residual of over $20K, but the market value had plummeted to barely a third of that by the end of the lease plus a one year extension.
I think the car would have been an okay deal even without their serious miscalculation. With it? What a steal.
Anyone who has substantial amounts of wealth (and a clue what to do with it) has almost all of it invested or tied up in land or other valuable assets.
In the case of Bezos and Gates, essentially all of it is in AMZN and MSFT stock, respectively. Which is why their net worth is tied to the share prices. Were they to try to cash out, their massive stock sales (and the lack of confidence indicated by their decision to cash out), would cause the stock price to drop. So their actual available cash is less than their net worth. OTOH, they can easily borrow billions with their stock as collateral.
I've pointed that out many times over the years... it's a deep flaw in the basic philosophy of crypto-currency - they want it both ways. They want legitimacy, *and* they want complete freedom with no controls or regulation. They don't grasp that the latter is a necessary consequence of the former.
Some probably do want exactly what you say. Others recognize that as a financial instrument regulation is inevitable, but still see value because although it's subject to regulation, the supply of a cryptocurrency doesn't have to be subject to any central control, unlike fiat currencies.
I don't think that's TCO you're using. Those are operational expenses. The T stands for Total, which would include purchase price, taxes, registration fees, etc...
I did neglect taxes and registration fees, but included purchase price.
Your argument is circular. You assert that someone is an expert if and only if they can write secure C code. My assertion is based not on logic (circular or otherwise) but on data. What can we possibly conclude when security-focused portions of widely-scrutinize, mission-critical operating system kernels; cryptographic libraries; security-focused "trusted" operating systems; secure container systems; and the like, all written by top security-focused software engineers who are doing their level best to make their code airtight, consistently have vulnerabilities which would not be possible in a more secure language?
The only possible conclusion is: writing secure software in C is too hard.
I'm fully aware of the techniques one uses to avoid buffer overflows, stack smashing, heap corruption, etc. They're easy to describe and in the abstract it seems like they should be simple to consistently implement. Heck in many cases correctness can even be automatically validated. So it seems like this should be easy. Massive amounts of real-world experience proves otherwise.
I'll even grant that it's possible that somewhere there are C programmers who are good enough that they never write such bugs. But obviously there aren't enough of them to write all of the security-sensitive software that we need written.
I'm buying a new car, and just ran the numbers for a volt
Don't forget to factor in maintenance costs. EVs are much simpler with fewer moving parts.
A Volt isn't. It's more complex with more moving parts, since it has all the complexity of an ICEV, plus all of a battery EV, plus all of the complexity for transitioning between the two.
A BEV is much simpler, though. And can be *much* cheaper. I expect that my TCO for owning my originally factory-new Nissan LEAF for 200K miles will be about $20K, or $0.10 per mile.
Granted, I kind of lucked out on the cost of the car. I leased it in 2013. At the time, it was a $37K car ($30K after tax credits, which went to the lessor), compared to $40K for a Chevy Volt. But with all the new greater-range EVs coming out now, the value of those older LEAFs has dropped like a rock, so I was able to buy it off lease for $6K. Three years of lease payments added up to just under $7K, so my cost for the vehicle has been just under $13K, including the interest I paid on the lease (I paid cash when I bought it off lease).
I average 4.2 miles per kWh, and pay about $0.11 per kWh, so my energy cost is 2.6 cents per mile. Assuming $2.50 per gallon, I'd have to get 95 miles per gallon in an ICEV to get the same per-mile energy cost.
Then there's maintenance. Or not. I've put 50K miles on my car and the only maintenance I've had to do is $200 for a set of new tires. Oh, and refill the windshield wiper fluid reservoir. Brakes are still great (EV brakes last longer than ICEV brakes, because much braking is regenerative). Oil changes aren't a thing.
Looking into the future, I can expect to replace the tires at normal intervals, and eventually it'll need brake pads. There is no transmission. No clutch. There are no spark plugs or wires, no catalytic converter or other emissions system, no muffler. At some point I'll need to replace the 12V lead acid battery (normal auto battery).
Of course, eventually the big lithium ion battery will need to be replaced. So far, in 4 years and 50K miles my battery has lost 3% of its capacity. It's warrantied to maintain 80% for 100K miles, but judging by the experience of other LEAF owners in fairly cool climates like mine (very hot climates are a different story!), it's more likely that it'll go 200K miles before it falls to 80%.
When the battery does need to be replaced, I'll be looking at a cost of about $5K, which is pretty comparable to the cost of a new engine for a small car, and quite a bit less than the cost of a new engine & transmission. Plus, by the time I buy a new battery pack, the replacement will almost certainly be higher-capacity and longer-lasting.
Anyway, assuming I drive the car until the battery needs to be replaced at, say, 200K miles, and then junk the car, I'll have spent around $20K, which is a TCO of $0.10 per mile. To go beyond 200K I'll probably looking at CV joints and similar components in addition to a battery, but it seems reasonable that I could get another 200K miles for an additional $7K or so in repairs (incl. battery), plus another $5K in electricity, so about $0.06 per mile.
The downside, of course, is that this is not a road trip vehicle. It's great for running around town, though. And it's a lot more fun to drive than comparable ICEVs.
If your point is that these security "experts" are not highly competent, then I support that, yes.
They're the best in the world, in many cases.
I do believe it, and here's why: the vast majority of security holes in C code are buffer overruns and failing to sanitize inputs. These are both easily mitigable.
And yet, they so often aren't. Even in code written and reviewed by security experts. You're supporting my point.
having "unique" plug types for particular purposes is a *feature*, not a bug - simply by looking at the plug, we know what the cable and the port does
You're trying to make a virtue of a problem. The right solution is that the cable and port should do everything. There's no need to figure out what it does or doesn't do, because anything the devices in question are capable of doing can be done via that port and plug.
So you see a Type C plug - is it Thunderbolt or not? Is it a DisplayPort?
Yes, it's all of the above, assuming it makes sense that the device can be a source/sink of the relevant data.
What voltages/amps can it provide?
The cable (unless it's non-compliant crap; don't buy non-compliant crap) can handle the full range of voltages and amperages in the specification.
This is incredibly convenient. My laptop, phone and tablet all have USB-C ports, for both data and power. I now travel with only one power adapter, the one for my laptop, which I can also use for my phone and tablet. I also carry a USB-C to USB-C cable, which enables me to charge any of the devices from any other device. Charging my laptop from my phone would be silly, charging it from the tablet is slightly less so and I have actually had occasion to do exactly that. Charging tablet from phone or phone from tablet both make sense sometimes, and I do both often when traveling.
I also occasionally use my phone charger to power my laptop. It doesn't provide enough power to maintain the battery level while I'm using it, but it does stretch battery life, and will recharge the laptop if left connected overnight. I do this mainly because I have a phone charger plugged in next to my bed. I'm waiting for wall outlets like this one to become better and cheaper, and then I'll replace many of my home outlets with them, mostly eliminating the need for wall warts.
That will be a great day. Actually, I hope that before too long we start building DC power distribution networks in homes, the way data centers have gone. At moderate voltages (e.g. 48V) resistive line losses are easily offset by the power savings provided by larger, more efficient central converters. Then each wall outlet should have a couple of simple step-down transformers to provide 20V and 5V over USB, per the power delivery specification.
While we're at it, we should add network over USB as well. So any device plugged into any wall outlet with a simple USB-C to USB-C cable will have both power and data (with 802.1X authentication for security). 20 Gbps should be enough bandwidth for the next couple of decades.
just look at e.g. the many forum posts of people who connected "the wrong type" of USB-C-to-HDMI connectors because they didn't know their USB C wasn't the USB C that they thought USB C was supposed to be.
Yes, there is crap hardware being sold. The USB implementer's forum needs to start enforcing its trademark and shutting down sales of non-compliant crap. In the absence of that, just make sure that the stuff you're buying has been tested by someone who knows what they're doing (on Amazon, check for Benson Leung's comments). No, this state of affairs is not good, but it's an implementation problem, not a problem with the fundamental concept of one plug that does everything.
That fundamental concept is awesome.
Yeah, well allow me to say that day looks a long way off to me.
You have no rational basis for that statement. Lacking understanding of what intelligence is, we have no idea what is necessary to create it. We we could hit on the key idea underlying the theory of intelligence today, or it could take a century of slogging incremental progress. We don't know what we don't know.
Whether we'll discover the theory from work on artificial intelligence or from the study of natural intelligence, or some combination of the two, I don't think anyone can say. My guess it that it will be a combination AI research and study of natural intelligence that will yield the crucial insights. How far we are from obtaining those insights is unknowable.
To rival the human brain (in terms of intuition, not speed or capacity), you need to first be able to duplicate the human brain.
You're implicitly assuming that the way our brains are built is the only way to construct intelligence. I see no rational basis for such an assumption.
I do think that we're not likely to build AI until we understand more deeply what intelligence is, and that that theory will likewise help us understand how our own brains work. But there's no reason to believe that evolution produced the only possible form.
I took an AI course in university and the professor would say that AI has been right around the corner for 40-years....
Yes... that's proof that we have no idea how far we are from creating artificial general intelligence. We never have known how far away it is, because we don't know what intelligence is or how it works.
But "We don't know how far away it is" does not imply "It's many years away". It just means we don't know. We could have the crucial breakthrough this afternoon and find all of our computers controlled by a global Internet-based AI by supper time, or we could be 100 years away. We just can't know until we understand intelligence, and by the time we understand it we'll be able to build it.
But a mad man wouldn't be working on so many "random" things that all lead to the salvation of mankind.
You don't know that any of those things will lead to the salvation of mankind. It seems plausible that all are important to our future, but you can't know.
Unless you're the time traveller.
General AI is likely still very far away.
Maybe, maybe not. Perhaps there's just one crucial idea that needs to be discovered to enable artificial general intelligence (AGI). Perhaps there are a whole bunch of incremental developments in both hardware and software that are required. We don't know, and we won't know until we get there.
Given that, we really should spend considerable time and resources on thinking about how to prepare for the day when we do figure out how to create AGI. Because what does seem quite likely is that the human brain is not the pinnacle of possible intelligence. That in turn means that AGI, once developed, is likely to quickly surpass human ability in the same way and for the same reasons that machines are stronger, faster, tougher, etc., than humans. So, at some point in the future -- and how distant that point is we do not know and cannot guess -- we will be sharing our planet with entities who are orders of magnitude smarter than we are, which means that if our goals conflict with their goals, we'll lose.
AGI poses both boundless opportunity and boundless risk to humanity. If we can convince the superintelligences to work on our behalf and according to what we value, they can help us solve all of our problems and move the human race into a near Utopia. If we can't, we'll probably cease to exist.
Musk is right to be concerned about this. He's wrong to think that we should be making regulations, because we have absolutely no idea what regulations to make.
Zuckerberg is wrong not to be concerned about it. Oh, he's right that our current "AI" technology is so far from being AGI that it's not worth worrying about. But Musk isn't talking about current AI, he's talking about the AGI that will come at some unknown point in the future.
Everyone's understanding of AI is limited.
Trust me our understanding of Natural Intelligence is limited.
You mean "our understanding of Intelligence is limited".
Once we understand intelligence, we'll be able to build one. Attempting to build intelligence will likely be an essential element of gaining that understanding, as will the study of "natural" intelligence.
Come on, it's not impossible to write secure C code if you know what you're doing.
If you believe this, you don't pay much attention to CVEs.
> The point is that it's quite clear that no one is highly competent at writing security-critical software in C.
Huh? You literally just said there is a plethora of security defects written by highly competent C programmers. So which is it?
That's what I said. It's so hard to write secure software in C that basically no one can do it.
Unrelated question: how would you respond to a tax cut on businesses and the wealthy if it also came with a remediation of the welfare system, strengthening of the middle-class, and a complete elimination of homelessness and hunger through a powerful social safety net? Assume the math works out (it does).
Do I also get a pony? And elimination of the federal debt?
You may been in the job when the administration changed.
And stayed... demonstrating extreme lack of foresight.
:P
That makes perfect sense. OTOH having been fired by Trump may be a sign of qualification in the future.
Well... it also means you got hired by Trump, which may be a sign of disqualification. i suspect that the best option is never to have been associated with Trump in any way.
I hope Spicer goes to CNN or MSNBC and gets to sound off on Trump.
That would be awesome, but Spicer would have to be a complete idiot to do it while Trump is still in office, and maybe even after. Trump has a vindictive streak a mile wide and the sort of "disloyalty" that would constitute in his eyes would provoke a really dangerous reaction.
It's exactly the same thing.
Yes, exactly the same if you want to ignore the fact that corporations are taxed based on their profits while US citizens are taxed based on their revenue (income). I'd love to be able subtract my mortgage, transportation, food, and utilities costs from my taxable income as operating costs and only be taxed based on what's left.
You don't actually understand your own taxes, I see. You do get to subtract all that stuff, effectively. You absolutely get to subtract your mortgage interest, which is the expense part. The principle payment is net income; it directly increases your net worth. As for the rest of the stuff, that's what standard deductions are for, to account for the basic cost of living, so you pay taxes only on your "extra". And that's actually also part of the justification behind the progressive scale; the less you make the more of your income must go to non-luxury expenses.
What you can't do is decrease your taxable income by choosing to buy more expensive stuff. Corporations can do this, but they have strong motivation not to, because their reason for existence is to generate value for shareholders. Being wasteful, whether it's to reduce tax liability or just to keep employees in unnecessary luxury, is contrary to the shareholders' goals and they'll eventually force the corporation to get more efficient and generate more income.
IIRC, corporate income taxes are paid on profits.
Correct. And note that it applies only to C corporations, not S corps, LLCs, etc.
Passing costs on to the customer is impossible here, since if the company could make more money by raising prices they would have.
This would be true if all of their competitors weren't subject to the same costs. An individual company can't impose its own costs on customers at will, but when the cost is imposed on the entire industry, competition doesn't prevent it. Foreign competition has an entirely different structure, of course, but countries balance those differences with tariffs.
I'll grant that most economists believe there is enough variation in corporate structure to prevent costs from being pushed on consumers. I'm not sure that's true in all industries, though, especially the capital-intensive industries where C corps are really the only ones able to play.
Exactly the same argument explains how the cost can land on employees. Competition for wages, etc., doesn't prevent it because it applies to the entire industry. There's another anti-labor dynamic as well; if corporate taxes do hit owners of capital, reducing their rate of return, they will move the capital to other industries or other countries, reducing investment in the relevant domestic industries, which eventually reduces competition for workers in those industries, lowering wages.
Profit is what goes to the shareholders, in the form of dividends or increased value of the company, and so if you were going to tax individuals you'd have to tax shareholders. Of course, many shareholders aren't individuals, but rather mutual funds and other companies. Taxing the ultimate recipients gets difficult.
Mutual funds and other companies just push it back a level. At the end of the chain there are always individual people. There is absolutely no disagreement on this among economists. You won't find anyone who seriously argues that corporations ultimately pay taxes.
There's nothing evil or deceptive about corporate income tax. It's a tax on corporate income, which obviously means on direct and indirect shareholders.
Even if it were a tax on shareholders, it would be more transparent and honest to collect it via capital gains taxes. But it's not a tax only on shareholders. It's difficult to tease out where corporate taxes actually land, but the general consensus among economists is that it's primarily split between shareholders and employees; capital and labor. The most common view is that capital bears 80% of the cost and labor 20%, but there are also good models that show that labor bears 70+%.
The bottom line to me is that there's really no point in trying to figure out who actually pays corporate taxes. Just figure out who you actually want to tax, and tax them. If you want it all to land on owners of capital, tax capital gains. If you want to apportion 80% to capital and 20% to labor, do that. If you want to put it all on consumers, do that, via sales taxes.
Corporate taxes are evil because they appear to voters to be taxes on no one, free money, but this is patently false.