U.S. CitiesCity-Data Forum Index
Go Back   City-Data Forum > General Forums > Economics
 [Register]
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
View detailed profile (Advanced) or search
site with Google Custom Search

Search Forums  (Advanced)
Reply Start New Thread
 
Old 02-23-2015, 04:03 PM
 
Location: Ruidoso, NM
5,170 posts, read 4,733,092 times
Reputation: 4206

Advertisements

Quote:
Originally Posted by ncole1 View Post
Otherwise, without the full range of human behaviors, they will be limited in their range of jobs, and humans will create more jobs that the robots cannot do, to replace the ones that robots are taking in the mean time.
I understand what you are saying. I used to believe the same thing. But after I thought about it some more, I no longer think it's true.

The things that a human can do better than a machine will become less and less, correct? This limits the range of tasks that a human can do that would justify a decent wage. In the past it was relatively easy to move into new jobs as industrialization created greater demand and wealth. Many of these new jobs required *less* skill than the ones replaced, not more... like farmers and craftsmen becoming factory workers. More recently though (last few decades) the only low-skilled growth has been in service. This favors a very different demographic, and is a big part of the reason why female employment is better than male. What is that hypothetical person with a <100 IQ and crappy personality supposed to do? Jobs that require talent in personal service or sales are out, as well as anything needing a high intelligence. Any job that they could get will be squeezed even more as machine sophistication grows.

Again, the problem isn't that we *can't* support everyone at a living standard that is at least as good as we have now and much more advanced, while also needing to work a lot less. In theory this should be easy. The issue is that the rich and powerful will not view this as the optimal outcome for *them*. Their optimal outcome involves a much smaller human population.
Reply With Quote Quick reply to this message

 
Old 02-23-2015, 04:23 PM
 
Location: Ruidoso, NM
5,170 posts, read 4,733,092 times
Reputation: 4206
Quote:
Originally Posted by Adhom View Post
The depth of the inequality is what causes the instability.
There are many societies where there are a few very wealthy and great many poor today. Maybe they aren't that stable, but they do endure. At any rate we are going to experience a time of societal upheaval regardless of the outcome.

Computers, surveillance, drones and many other features of our high tech society make things really different this time. How long will it be before it will be easy to track where someone goes and what they are doing every minute? How long before a crisis occurs that results in the public actually wanting this and putting it into law?

As I stated in the OP, I believe the deterioration in living standards and rights will be gradual for a long time to come, and we will be distracted with real and imagined crisis. Propaganda and manipulation will keep us divided and confused. It will be nearly impossible to organize an effective change, and the longer we wait the less likely it will succeed.
Reply With Quote Quick reply to this message
 
Old 02-23-2015, 07:01 PM
 
24,885 posts, read 11,610,800 times
Reputation: 11621
So people have a hard time grasping what exponential growth looks like. Theres a parable about the inventor of chess being offered a reward, and he asks for one grain of rice on the first chess square, 2 on the second, 4 on the next, etc.

The king says thats much too little. I mean really, thats nothing!

Halfway through the board on the 32nd square its 4 billion grains of rice. Thats a lot of rice!

at the end, the board holds the equivalent of 100 tons of rice for every man, woman and child on Earth.

And thats exactly how computers are working. every 18 months they double in power. we've just finished the first half of the board, and are on the second half. The ability to automate human tasks is increasing exponentially.

Those of us refusing to look at the oncoming train and plan for it are going to make it harder to deal with when it occurs.
Reply With Quote Quick reply to this message
 
Old 02-23-2015, 07:28 PM
 
Location: Ruidoso, NM
5,170 posts, read 4,733,092 times
Reputation: 4206
Quote:
Originally Posted by greywar View Post
And thats exactly how computers are working. every 18 months they double in power.
As ncole1 pointed out earlier, we are experiencing some hardware limitations with that process. Exponential increases do not occur for long in reality for the reasons you mentioned. Eventually something gets in the way.

We are still going to face masses changes though, even if our processing capabilities rise more slowly in the future than they have in the past.
Reply With Quote Quick reply to this message
 
Old 02-23-2015, 07:39 PM
 
12,404 posts, read 9,203,248 times
Reputation: 8863
Quote:
Originally Posted by rruff View Post
I understand what you are saying. I used to believe the same thing. But after I thought about it some more, I no longer think it's true.

The things that a human can do better than a machine will become less and less, correct? This limits the range of tasks that a human can do that would justify a decent wage.
You are assuming that demand in the overall economy stays fixed. I think this view is a result of forgetting how human consumption can grow without bound. People consume more and more, until they start to feel a pinch to the pocketbook.

Demand does not remain fixed as production capacity is enhanced by new producers being added to the system (the robots). If consumption increases by the same amount as production, then there will be the same amount of need for human labor as before.

Even if you could replace a large amount of labor with robots, then consumer demand would still increase enough to keep full employment.

It would be vaguely analogous to the infamous Jevons Paradox ( Jevons paradox - Wikipedia, the free encyclopedia ) - except instead of energy efficiency, it would be labor efficiency. Increasing production per human-hour would lead to increased labor demand, not less.

Quote:
Originally Posted by rruff View Post
In the past it was relatively easy to move into new jobs as industrialization created greater demand and wealth. Many of these new jobs required *less* skill than the ones replaced, not more... like farmers and craftsmen becoming factory workers. More recently though (last few decades) the only low-skilled growth has been in service.
No, virtually every sector of the economy has grown.

Quote:
Originally Posted by rruff View Post
This favors a very different demographic, and is a big part of the reason why female employment is better than male. What is that hypothetical person with a <100 IQ and crappy personality supposed to do? Jobs that require talent in personal service or sales are out, as well as anything needing a high intelligence. Any job that they could get will be squeezed even more as machine sophistication grows.
You're leaping to conclusions. Until we discuss exactly which jobs would be replaced by robots and which jobs would not, we are not in a position to make this type of claim.

And there is no reason to bring IQ scores into the discussion - in fact, there is no evidence that there is even a relation between the "IQ score" you need for a job and its automatability. Plenty of things we think of as requiring "more intelligence" or "less intelligence" are counter-intuitively not so. Are you familiar with Moravec's paradox? http://en.wikipedia.org/wiki/Moravec%27s_paradox

Essentially the hardest things to get a machine to do turn out to be the things everyone can do that require little to no skill. Although the examples there largely draw upon motor skills, a similar argument about face recognition, among other things can be made. Again, even many clinically mentally retarded people with IQ scores below 70 are far, far better at these things than ANY computer program in existence.




Quote:
Originally Posted by rruff View Post
Again, the problem isn't that we *can't* support everyone at a living standard that is at least as good as we have now and much more advanced, while also needing to work a lot less. In theory this should be easy. The issue is that the rich and powerful will not view this as the optimal outcome for *them*. Their optimal outcome involves a much smaller human population.
This would lead not to automation, but to war. As we know from history, increasing concentration of power and wealth, if kept unchecked, inevitably leads to war. And this is true regardless of whether it is a result of supply and demand, or is a result of changing political conditions.
Reply With Quote Quick reply to this message
 
Old 02-23-2015, 07:40 PM
 
24,885 posts, read 11,610,800 times
Reputation: 11621
Quote:
Originally Posted by rruff View Post
As ncole1 pointed out earlier, we are experiencing some hardware limitations with that process. Exponential increases do not occur for long in reality for the reasons you mentioned. Eventually something gets in the way.

We are still going to face masses changes though, even if our processing capabilities rise more slowly in the future than they have in the past.
Im involved in the field more directly then ncole1 is I suspect..... I assure you, we're not going to be hitting any barriers for quite a while. Even her blog caps the speed at 12,100,000 GHz. And it doesnt take into account changes in parallel processing, and quantum computing. Nor all the things surrounding it.

I assure you a million cores running at 12,100,000 Ghz is a mind boggling idea.
Reply With Quote Quick reply to this message
 
Old 02-23-2015, 07:53 PM
 
12,404 posts, read 9,203,248 times
Reputation: 8863
Quote:
Originally Posted by greywar View Post
Im involved in the field more directly then ncole1 is I suspect..... I assure you, we're not going to be hitting any barriers for quite a while. Even her blog caps the speed at 12,100,000 GHz. And it doesnt take into account changes in parallel processing, and quantum computing. Nor all the things surrounding it.

I assure you a million cores running at 12,100,000 Ghz is a mind boggling idea.
You cannot simply apply any algorithm on a quantum or parallel computer and get the same enhancement. Again, Amdahl's Law comes into play and some algorithms are simply not parallelizable, particularly when it comes to simulations.

Quantum computing is not simply an enhancement of classical computing; it is a different beast entirely. For some things such as factoring large products of prime numbers, it can do very well. However it is far from clear that such problems will expand to cover anything relevant to job automation on a large scale, especially after you consider the effects of quantum decoherence.

The temperature a system must be cooled to in order to manage QD gets close to absolute zero very quickly in most systems with the types of qubits in wide use today. Before you got to the complexity of computer you need, the decoherence time scale would be far too short to be useful.
Reply With Quote Quick reply to this message
 
Old 02-23-2015, 08:29 PM
 
24,885 posts, read 11,610,800 times
Reputation: 11621
Quote:
Originally Posted by ncole1 View Post
You cannot simply apply any algorithm on a quantum or parallel computer and get the same enhancement. Again, Amdahl's Law comes into play and some algorithms are simply not parallelizable, particularly when it comes to simulations.
Cool. now think about reality. you're discussing a subset, and applying it to the whole. Turns out parallalization works for a ton of things. additionally a LOT of other things are going on alongside. hard drives, power, and more.

Quote:
Quantum computing is not simply an enhancement of classical computing; it is a different beast entirely. For some things such as factoring large products of prime numbers, it can do very well. However it is far from clear that such problems will expand to cover anything relevant to job automation on a large scale, especially after you consider the effects of quantum decoherence.
On this I partially agree. Quantum computing is a very challenging topic, that we are just starting to advance into, and the software development for it is considerably different then I am used to thinking. But the things it CAN do....represent the opposite of the earlier comment I replied to. These do a subset amazingly, and some of them will be applicable.

Quote:
The temperature a system must be cooled to in order to manage QD gets close to absolute zero very quickly in most systems with the types of qubits in wide use today. Before you got to the complexity of computer you need, the decoherence time scale would be far too short to be useful.
The complexity of computer I need for what? No seriously. If its 1,000,000 times faster, and 1,000,000 more capable then a human brain....and I make a billion of them, all networked together.....

See what I mean? This is a exponential problem, the water is 1 foot deep, and your argument is that we're all fine because its only got 10 more doublings to go before it runs out of water.
1.5 months from now..."its only 2 feet deep, no worries!"
3 years from now "OK so its 4 feet, its not that bad"
..... finally 15 years from now the water is 1,024 feet deep......
Reply With Quote Quick reply to this message
 
Old 02-23-2015, 08:43 PM
 
12,404 posts, read 9,203,248 times
Reputation: 8863
Quote:
Originally Posted by greywar View Post
Cool. now think about reality. you're discussing a subset, and applying it to the whole. Turns out parallalization works for a ton of things. additionally a LOT of other things are going on alongside. hard drives, power, and more.
So you need to show that the algorithms that are relevant for robotic automation have a very low serial computation fraction and are thus not very limited by Amdahl's Law. Until this is shown, the parallel argument is inconclusive.

Quote:
Originally Posted by greywar View Post

On this I partially agree. Quantum computing is a very challenging topic, that we are just starting to advance into, and the software development for it is considerably different then I am used to thinking. But the things it CAN do....represent the opposite of the earlier comment I replied to. These do a subset amazingly, and some of them will be applicable.


The complexity of computer I need for what? No seriously. If its 1,000,000 times faster, and 1,000,000 more capable then a human brain....and I make a billion of them, all networked together.....
Right , but the question is whether decoherence will prevent you from getting it that complex and fast in the first place, not what happens afterward.

Quote:
Originally Posted by greywar View Post
See what I mean? This is a exponential problem, the water is 1 foot deep, and your argument is that we're all fine because its only got 10 more doublings to go before it runs out of water.
1.5 months from now..."its only 2 feet deep, no worries!"
3 years from now "OK so its 4 feet, its not that bad"
..... finally 15 years from now the water is 1,024 feet deep......
You are pre-supposing continued exponential growth of the parameters relevant to automation, which you have neither given a reason to believe, nor have you even specified what parameters you believe we should use and why.

You have thus failed to make a case for extensive automation in the coming decades.
Reply With Quote Quick reply to this message
 
Old 02-23-2015, 09:00 PM
 
24,885 posts, read 11,610,800 times
Reputation: 11621
Quote:
Originally Posted by ncole1 View Post
So you need to show that the algorithms that are relevant for robotic automation have a very low serial computation fraction and are thus not very limited by Amdahl's Law. Until this is shown, the parallel argument is inconclusive.



Right , but the question is whether decoherence will prevent you from getting it that complex and fast in the first place, not what happens afterward.



You are pre-supposing continued exponential growth of the parameters relevant to automation, which you have neither given a reason to believe, nor have you even specified what parameters you believe we should use and why.

You have thus failed to make a case for extensive automation in the coming decades.
Human beings can do this. Why in the world would you believe that for some reason theres a scientific reason to stop automation from exceeding human capabilities?

So think about automation, there is a ton of parallel things going on. Vision and decision making with neural networks for example.

On the face of it your argument is foolish. It fails the sniff test. We have a very flawed system right now that does this sort of thing and much much more, its called human beings. And we're slow....very very slow highly parallel system.

You've constructed your argument very poorly.
Reply With Quote Quick reply to this message
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.

Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.


Reply
Please update this thread with any new information or opinions. This open thread is still read by thousands of people, so we encourage all additional points of view.

Quick Reply
Message:


Over $104,000 in prizes was already given out to active posters on our forum and additional giveaways are planned!

Go Back   City-Data Forum > General Forums > Economics
Similar Threads
Follow City-Data.com founder on our Forum or

All times are GMT -6.

2005-2018, Advameg, Inc.

City-Data.com - Archive 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35 - Top