The answer to that, as always, has been that you need to increase the level of education accordingly. Those who write and maintain AIs are going to have jobs. But a high school diploma isn't going to cut it, any more than finishing 7th grade was going to cut it once we got automation and computers.
That's a nice sentiment, but you seem to forget that not everyone is even capable of getting through high school without significant assistance. For a not insignificant part of the population, there is an upper bounds to where they can go in terms of education that effectively excludes a lot of the higher maths you need to get into CI/ML. As such, we have a lot of people for whom skilled labor is going to be the best they might hope to achieve in terms of employment. If you get rid of those jobs that leaves a lot of people with no prospects for work.
Not quite. The seigniorage is the profit that the government derives from minting money below face value, or if the currency is withdrawn from circulation (ex., by collectors). If the a $100 note is destroyed and replaced by another one then there is no net impact. However, if the note goes into international circulation then it is very profitable since tangible goods and services were exchanged for the note and it likely doesn't return to domestic circulation.
The US Mint does the same thing and part of the reason that the United States has had commemorative quarters since the 1990's is the seigniorage. In fact that seigniorage is one of the avenues of revenue, although not as much as taxes are.
I'm going to push back because what you are saying is extremely dependent upon what you are writing. For your basic business line web-application that is mostly CRUD, there's very little math. Even the parts of a business application related to accounting and finance don't really progress past basic algebra. These also tend to be the jobs that the majority of programmers have.
You step out of that sphere though and math and science starts to play a much bigger role. If you look at the code for something like Unreal Engine there is a ton of math involved (linear algebra, calculus) in making it work. A lot of my background is in bioinformatics and there it's not just a matter of knowing how to code, but you also have to understand biology at more than the basic level. Generally when it comes to software development, the more interesting a project is, the more likely you are going to need some advanced maths and some domain specific scientific knowledge.
Not to be pedantic, but a lot of the research being done in universities is by graduate students. Even national and industry labs tend to have lots of people with BS and MS degrees assisting the PhDs in the work.
I'm still being too lazy to look up the studies, but you are correct in that they didn't have just a random group of people working on something. Usually it was diverse groups of professional peers. So the studies do try and narrow the variable set down to cultural background as opposed to just any background.
I'm too lazy to dig up the links, but most of what I've read argues that diverse groups are better at problem solving. One a basic level this makes sense. Groups with varied backgrounds have more information they can bring to a problem in aggregate. In contrast a group with a very narrow background doesn't have as much information. Given that science is about problem solving, having more information to throw at a problem is always better.
When I heard the announcement I thought of "Forbidden Archeology: The Hidden History of the Human Race" by Michael A. Cremo and Richard L. Thompson (Amazon Link, Condensed Version on Amazon, Wiki Article). The book basically says the same thing: there has been some weak archeological evidence to support the theory that humans (or hominoids) have been around for a lot longer than currently recognized, and in places where people didn't think they were. Obviously the book is controversial and much of the argument is dependent upon scattered artifacts and their interpenetration. A finding like this could booster the books argument though, some interesting stuff to keep an eye on at least.
For now, Rentberry charges users a $25 fee, but in the future, it plans to charge 25 percent of the difference between the asking price and the agreed upon rent. Whoever received the better deal pays the fee -- every month.
This effectively means that the landlord is always incentivized to start the bidding at the highest reasonable price they can. If the bidders push the rent higher, they win. If the bids are below their asking rent then they also win since the tenant has to pay their rent plus a 25% fee.
The current model where you just pay a $25 matching fee isn't unreasonable, it's when you start to collect in perpetuity that it's a problem.
By people who believe a piece of paper is more important than what you can do. Fortunately, their day is coming to an end.
Good luck getting rid of that piece of paper. What most people fail to understand is that there has always been some sort of verification of skills given when people moved from unskilled (i.e. apprentice) to trained (i.e. journeyman) that was recognized throughout a region. The piece of paper might be meaningless after twenty years when you have demonstrable experience, but good luck getting someone to recognize that any training you might have is worth something if you don't have experience to back it up.
Our traffic rules are built with the faults of human drivers in mind. Humans have more blind area than vision so we make rules like, Don't cross into this lane ever, or Don't go when the light is this color. Do these rules ultimately need to apply to autonomous cars?
Except they are also built for traffic control as well. For some busy streets, the only way you are going to cross it is when the traffic has to stop for red light.
Neural networks have a shaky biological basis at best. More pragmatically, they are a network of perceptrons with sigmoidal output functions. In that cases, yes, more bits of precision can be very relevant. Once you start talking about a deep learning network the updates to individual perceptrons can be very small and 32 bits are needed.
I fully disagree with this statement. So, I will better not get involved in a discussion with you because our positions are too different.
I get the point they were trying to make. Science is concerned with two things:
Reproducible: If I tell you what I did, can you reproduce my experiments?
Understanding: Given the results of those experiments, can we explain what happened?
Right now the EM Drive is on step one. We are trying to isolate all of the sources of error out of the experiment and make it as clear as possible as to what is going on so that the experiment can be reproduced by another lab. If the other lab can't reproduce it, it's not science and the original experiment is likely wrong. If they can reproduce it, then we can really get into part two and figure out why.
It's entirely possible that now that the article out there for scrutiny that it will end up like the Faster-than-light neutrino anomaly, but it is also possible that the drive is using something like the Mach effect to get the thrust. Completely within the bounds of known physics, but not really seen as a useful effect worth exploring.
Why isn't there a similar push to get men into kindergarten education or nursing?
The nursing side of things always catches my eye since, in the United States at least, nursing was considered to be a man's field of work at one point. Prior to the American Civil War it was pretty much impossible for a woman to get involved with nursing but the demand for nurses during the war eased of the restrictions since the men were needed elsewhere. Even then women faced an uphill battle at first since since the doctors didn't think they would be able to handle the "blood and guts" of the battle field, operating room, and recovery areas.
After the war you started to see things quickly swing the other way until the late 1890's when women effectively dominated nursing and the field was established as being "women's work."
Who said you need to model the entire universe? You just need to model the parts of the universe that the agents are interacting with directly. Plus, there is also nothing saying that the modeling is done in real time from the standpoint of the modeler. It could be a very slow model, but from the perspective of the agents, the model would be in real time.
Systems that were written largely by members of Generation X and marketed by Baby Boomers. But no, keep thinking that everything is the fault of which ever generation is the youngest.
...in my opinion, the decline in contributing members is mainly because of today's "gimme-gimme-gimme and I won't give anything back in return" mentality of the young generation of internet users.
So in other-words, everything started going down hill September 1993?
Which thoroughly pisses me off, considering that wikipedia is the biggest free and easily accessible repository of human knowledge (outside of the NSA).
I'm not sure you could call the NSA a repository of human knowledge though. While they might have a lot of human data, the curation (such as it is) at Wikipedia is what actually makes things useful. Just having a bunch of data in and of itself doesn't mean it's actually useful.
Nothing has been fixed, US still has a spending problem.
I'd argue that we have a spending problem and a revenue problem. Think of it this way, if you were trying to pay all of you debts off, would you quit your job and find a lower paying one or would you get a second job to until things are paid off?
I agree with you on all points but one -- this wasn't really a chance to stop the crazy. The budget is too out of control to come up with a fix in a few days. It is going to take a very difficult debate among the entire electorate to decide which sacred cows are going to be slaughtered. It has gotten to the point where no politician is willing to bring the subject up because everyone is going to feel some very real pain in order to solve all of this.
The biggest sacred cow to fix the federal deficits would be to just reform the tax code and actually have more revenue coming in than is being spent. Discretionary spending is only 30% of the budget with 6%, 64% is mandatory spending and the interest on the debt is only 6%. Even if you cut the discretionary spending completely (ignoring the revenue issues due to that workforce being laid off) you still aren't going to close the gap completely. Tax breaks, on the other hand, are currently $1.18 trillion dollars which is more than the discretionary budget.
Once you have the federal deficits taken care of you can just pretty much let the budget cost on cruise control since your principle shouldn't be getting any bigger at that point. Granted between months in the year you might see some nominal increases in the debt due to short term bonds being sold to smooth out the bumps in the revenue caused by how taxes are filed, in the long run even that would go away if a nice âoerainy day fundâ and âoebufferâ account was established.
Frankly, the only sure-fire way to pay off this debt is via massive spending cuts.
Massive spending cuts are completely useless if you are still running a deficit. If you have a balanced budget then you will eventually payoff all of your debts in full and considering the time frames that governments have to work with, who really cares how long it takes as long as the principle is not increasing and the payments are being made on time?
There are some issues with actually getting a balanced federal budget though due to the inconstant way the money comes in and the unpredictability at times as well. It's not like a household budget where barring job loss you pretty much know exactly how much each paycheck is going to be. The federal budget really does need to be running a slight surplus so that a âoerainy day fundâ can be established but usually as soon as a surplus appears politicians want to use that money to gain political capital as opposed to saving it for later on down the road.
I guess it depends on the type of work though. When it comes to writing software, I find that if I get up, go for a walk, chat with some coworkers, etc. - which is technically "unproductive" - that I can move past a roadblock that I was encountering and sometimes wrap a problem up in about thirty minutes. Other-times where the expectation was that you had to sit and code all day, I could easily spend four times as long working on the same problem due to mental fatigue.
I'm willing to bet your physics, maths and engineering professors don't dick around like that.
Nope, the physics professor I had spent the first day talking Sikhism and the Harmandir Sahib (the Golden Temple) and it would come up from time to time after that, although he did some more time actually talking about physics. Sadly, he had a very thick accent so you really had to pay attention to figure out what he was saying to determine if it was even relevant.
Sure, except for the fact that nobody has been able to explain how the rebels managed to get their hands on chemical weapons in order to launch the attack. Short of manufacturing them they would have had to have gotten them from a Syrian storehouse that fell into their hands in which case it makes sense for the Syrian government to want to wash their hands free of them.
It's entirely possible that the rules have changed since you had your first SSBI, but unless they changed things (again) the only reason that they would talk to elementary school teachers is if you actually given them as one of your references. The last time I looked at DCID 6/4 it basically said something along the lines of verify education and interview sources if education was your primary activity during the past three years. There's always the possibility that an investigation would be expanded, but I like I said a couple posts back, I don't see OPM going and talking to someone's elementary school teachers and asking for a character reference.
Slashdot Mirror
The answer to that, as always, has been that you need to increase the level of education accordingly. Those who write and maintain AIs are going to have jobs. But a high school diploma isn't going to cut it, any more than finishing 7th grade was going to cut it once we got automation and computers.
That's a nice sentiment, but you seem to forget that not everyone is even capable of getting through high school without significant assistance. For a not insignificant part of the population, there is an upper bounds to where they can go in terms of education that effectively excludes a lot of the higher maths you need to get into CI/ML. As such, we have a lot of people for whom skilled labor is going to be the best they might hope to achieve in terms of employment. If you get rid of those jobs that leaves a lot of people with no prospects for work.
The profit is the replacement.
Not quite. The seigniorage is the profit that the government derives from minting money below face value, or if the currency is withdrawn from circulation (ex., by collectors). If the a $100 note is destroyed and replaced by another one then there is no net impact. However, if the note goes into international circulation then it is very profitable since tangible goods and services were exchanged for the note and it likely doesn't return to domestic circulation.
The US Mint does the same thing and part of the reason that the United States has had commemorative quarters since the 1990's is the seigniorage. In fact that seigniorage is one of the avenues of revenue, although not as much as taxes are.
I'm going to push back because what you are saying is extremely dependent upon what you are writing. For your basic business line web-application that is mostly CRUD, there's very little math. Even the parts of a business application related to accounting and finance don't really progress past basic algebra. These also tend to be the jobs that the majority of programmers have.
You step out of that sphere though and math and science starts to play a much bigger role. If you look at the code for something like Unreal Engine there is a ton of math involved (linear algebra, calculus) in making it work. A lot of my background is in bioinformatics and there it's not just a matter of knowing how to code, but you also have to understand biology at more than the basic level. Generally when it comes to software development, the more interesting a project is, the more likely you are going to need some advanced maths and some domain specific scientific knowledge.
Not to be pedantic, but a lot of the research being done in universities is by graduate students. Even national and industry labs tend to have lots of people with BS and MS degrees assisting the PhDs in the work.
I'm still being too lazy to look up the studies, but you are correct in that they didn't have just a random group of people working on something. Usually it was diverse groups of professional peers. So the studies do try and narrow the variable set down to cultural background as opposed to just any background.
I'm too lazy to dig up the links, but most of what I've read argues that diverse groups are better at problem solving. One a basic level this makes sense. Groups with varied backgrounds have more information they can bring to a problem in aggregate. In contrast a group with a very narrow background doesn't have as much information. Given that science is about problem solving, having more information to throw at a problem is always better.
When I heard the announcement I thought of "Forbidden Archeology: The Hidden History of the Human Race" by Michael A. Cremo and Richard L. Thompson (Amazon Link, Condensed Version on Amazon, Wiki Article). The book basically says the same thing: there has been some weak archeological evidence to support the theory that humans (or hominoids) have been around for a lot longer than currently recognized, and in places where people didn't think they were. Obviously the book is controversial and much of the argument is dependent upon scattered artifacts and their interpenetration. A finding like this could booster the books argument though, some interesting stuff to keep an eye on at least.
For now, Rentberry charges users a $25 fee, but in the future, it plans to charge 25 percent of the difference between the asking price and the agreed upon rent. Whoever received the better deal pays the fee -- every month.
This effectively means that the landlord is always incentivized to start the bidding at the highest reasonable price they can. If the bidders push the rent higher, they win. If the bids are below their asking rent then they also win since the tenant has to pay their rent plus a 25% fee.
The current model where you just pay a $25 matching fee isn't unreasonable, it's when you start to collect in perpetuity that it's a problem.
By people who believe a piece of paper is more important than what you can do. Fortunately, their day is coming to an end.
Good luck getting rid of that piece of paper. What most people fail to understand is that there has always been some sort of verification of skills given when people moved from unskilled (i.e. apprentice) to trained (i.e. journeyman) that was recognized throughout a region. The piece of paper might be meaningless after twenty years when you have demonstrable experience, but good luck getting someone to recognize that any training you might have is worth something if you don't have experience to back it up.
Our traffic rules are built with the faults of human drivers in mind. Humans have more blind area than vision so we make rules like, Don't cross into this lane ever, or Don't go when the light is this color. Do these rules ultimately need to apply to autonomous cars?
Except they are also built for traffic control as well. For some busy streets, the only way you are going to cross it is when the traffic has to stop for red light.
Neural networks have a shaky biological basis at best. More pragmatically, they are a network of perceptrons with sigmoidal output functions. In that cases, yes, more bits of precision can be very relevant. Once you start talking about a deep learning network the updates to individual perceptrons can be very small and 32 bits are needed.
I get the point they were trying to make. Science is concerned with two things:
Right now the EM Drive is on step one. We are trying to isolate all of the sources of error out of the experiment and make it as clear as possible as to what is going on so that the experiment can be reproduced by another lab. If the other lab can't reproduce it, it's not science and the original experiment is likely wrong. If they can reproduce it, then we can really get into part two and figure out why.
It's entirely possible that now that the article out there for scrutiny that it will end up like the Faster-than-light neutrino anomaly, but it is also possible that the drive is using something like the Mach effect to get the thrust. Completely within the bounds of known physics, but not really seen as a useful effect worth exploring.
Why isn't there a similar push to get men into kindergarten education or nursing?
The nursing side of things always catches my eye since, in the United States at least, nursing was considered to be a man's field of work at one point. Prior to the American Civil War it was pretty much impossible for a woman to get involved with nursing but the demand for nurses during the war eased of the restrictions since the men were needed elsewhere. Even then women faced an uphill battle at first since since the doctors didn't think they would be able to handle the "blood and guts" of the battle field, operating room, and recovery areas.
After the war you started to see things quickly swing the other way until the late 1890's when women effectively dominated nursing and the field was established as being "women's work."
Who said you need to model the entire universe? You just need to model the parts of the universe that the agents are interacting with directly. Plus, there is also nothing saying that the modeling is done in real time from the standpoint of the modeler. It could be a very slow model, but from the perspective of the agents, the model would be in real time.
So none of the above then?
Systems that were written largely by members of Generation X and marketed by Baby Boomers. But no, keep thinking that everything is the fault of which ever generation is the youngest.
...in my opinion, the decline in contributing members is mainly because of today's "gimme-gimme-gimme and I won't give anything back in return" mentality of the young generation of internet users.
So in other-words, everything started going down hill September 1993?
Which thoroughly pisses me off, considering that wikipedia is the biggest free and easily accessible repository of human knowledge (outside of the NSA).
I'm not sure you could call the NSA a repository of human knowledge though. While they might have a lot of human data, the curation (such as it is) at Wikipedia is what actually makes things useful. Just having a bunch of data in and of itself doesn't mean it's actually useful.
Nothing has been fixed, US still has a spending problem.
I'd argue that we have a spending problem and a revenue problem. Think of it this way, if you were trying to pay all of you debts off, would you quit your job and find a lower paying one or would you get a second job to until things are paid off?
I agree with you on all points but one -- this wasn't really a chance to stop the crazy. The budget is too out of control to come up with a fix in a few days. It is going to take a very difficult debate among the entire electorate to decide which sacred cows are going to be slaughtered. It has gotten to the point where no politician is willing to bring the subject up because everyone is going to feel some very real pain in order to solve all of this.
The biggest sacred cow to fix the federal deficits would be to just reform the tax code and actually have more revenue coming in than is being spent. Discretionary spending is only 30% of the budget with 6%, 64% is mandatory spending and the interest on the debt is only 6%. Even if you cut the discretionary spending completely (ignoring the revenue issues due to that workforce being laid off) you still aren't going to close the gap completely. Tax breaks, on the other hand, are currently $1.18 trillion dollars which is more than the discretionary budget.
Once you have the federal deficits taken care of you can just pretty much let the budget cost on cruise control since your principle shouldn't be getting any bigger at that point. Granted between months in the year you might see some nominal increases in the debt due to short term bonds being sold to smooth out the bumps in the revenue caused by how taxes are filed, in the long run even that would go away if a nice âoerainy day fundâ and âoebufferâ account was established.
Frankly, the only sure-fire way to pay off this debt is via massive spending cuts.
Massive spending cuts are completely useless if you are still running a deficit. If you have a balanced budget then you will eventually payoff all of your debts in full and considering the time frames that governments have to work with, who really cares how long it takes as long as the principle is not increasing and the payments are being made on time?
There are some issues with actually getting a balanced federal budget though due to the inconstant way the money comes in and the unpredictability at times as well. It's not like a household budget where barring job loss you pretty much know exactly how much each paycheck is going to be. The federal budget really does need to be running a slight surplus so that a âoerainy day fundâ can be established but usually as soon as a surplus appears politicians want to use that money to gain political capital as opposed to saving it for later on down the road.
I guess it depends on the type of work though. When it comes to writing software, I find that if I get up, go for a walk, chat with some coworkers, etc. - which is technically "unproductive" - that I can move past a roadblock that I was encountering and sometimes wrap a problem up in about thirty minutes. Other-times where the expectation was that you had to sit and code all day, I could easily spend four times as long working on the same problem due to mental fatigue.
Back in 1995 Dilbert highlighted some of the issues with development and engineering type positions and what is considered work - namely, the time spent at home thinking about a problem isn't considered "work."
I'm willing to bet your physics, maths and engineering professors don't dick around like that.
Nope, the physics professor I had spent the first day talking Sikhism and the Harmandir Sahib (the Golden Temple) and it would come up from time to time after that, although he did some more time actually talking about physics. Sadly, he had a very thick accent so you really had to pay attention to figure out what he was saying to determine if it was even relevant.
Sure, except for the fact that nobody has been able to explain how the rebels managed to get their hands on chemical weapons in order to launch the attack. Short of manufacturing them they would have had to have gotten them from a Syrian storehouse that fell into their hands in which case it makes sense for the Syrian government to want to wash their hands free of them.
It's entirely possible that the rules have changed since you had your first SSBI, but unless they changed things (again) the only reason that they would talk to elementary school teachers is if you actually given them as one of your references. The last time I looked at DCID 6/4 it basically said something along the lines of verify education and interview sources if education was your primary activity during the past three years. There's always the possibility that an investigation would be expanded, but I like I said a couple posts back, I don't see OPM going and talking to someone's elementary school teachers and asking for a character reference.