The core question with free trade turns out NOT to be about free trade, but rather to be about productivity. Here's the essence:
Suppose I am Toyota, and I build a auto-manufacturing system, which allows me to build cars at a cost of 80% of what it costs at GM, by spending 5% extra on parts, and 25% less on labor (absolute, not relative). Is this a good thing? If this approach spreads, it will cause 5 million people from current jobs to become unemployed, as manufacturing in general gets cheaper. Only benefit is that everyone using manufactured items is happier, and saves (about) 20% of what they were spending. And so they spend it on more manufactured stuff (5%), and on services (15%).
Suppose I am a software methodologist, and I can take your software company and improve it's efficiency at delivering software, using Agile methods done well, by 10x. My team of 15 can do what your large company software team takes 150 to do...and faster with fewer bugs...and my people aren't IQ-exceptional. Is this a good thing? Supposing (falsely) that we're writing all the software we want written, spreading the software methodology that I use would result in 90% unemployment among programmers. Should you ban my approach? Is the cost of software dropping by 10x worth putting millions of folks out of work?
Is it worthwhile for the automated industrial looms that constituted the second wave of the industrial revolution to allow 3 low-paid weavers to do what 30 high-paid weavers used to do? Shouldn't we ban that? We didn't. We're now 10-10,000 times richer, depending on how you calculate.
IF we agree that it's a legitimate choice for me (really, this is my job) to advocate switching software methodology to doing things for 10x cheaper...then we have a discussion to have. If you don't...I'd recommend a move back to primitive agriculture. Feel free to work 16 hours a day scraping dirt, in order to have only enough to eat that 1/2 of your children die of malnutrition. That is the result of your policy.
------------------
Harder:
Suppose Yudkowsky's applied Epistemology work (great recently, this October. Rather unusually, I don't have anything to fight about) result in a low-grade AGI that caps somewhere near 2x human intelligence, and which has particular facility with 3D printers. Eli can produce anything made of plastic using his secret slave AGI at a cost of 1/100th what anyone else does. And it takes no workers. Because he can print his own printers. And he can synthesize plastic from the seawater near his house in Berkeley. Is it a net win to destroy every plastics-producing job in the country, just in order to get free plastic stuff?
That's what happened to Agriculture in the rich world. One dude with a tractor and industrial fertilizers can now do the work of 49 dudes without tractors or industrial fertilizers. Win for everyone? For sure. But it was a painful transition. Honestly, we're in the middle of that same transition in manufacturing as well. Robots are cheaper than humans (partly because of government, but mostly because technological improvements). What used to take 50 people to do, currently takes 3, and we're well on the path to 1, or even less.
---------------------
Harder:
Suppose that the Russian government announces that not only are they sitting on a quadrillion dollars worth of diamonds, but also that oilfield they found in the Arctic circle contains about a bajillion dollars worth of oil that it costs $1/barrell to extract from the ground. Furthermore, because we so clearly all want the same form of centrally managed communist/fascist/crony capitalist government, they are going to give us as much oil as we can use for free. Take it or leave it?
What if the Klingons dropped a trillion barrels of oil into a newly dug (very deep) lake in Utah, and then flew off? Should we use it?
What if something suddenly becomes (near) free that used to take work to get?
--------------------------
What is the proper response to productivity improvements. My claim is that every single one of the examples I give above gives us a better life...and puts a whole buttload of people out of work. And the better life for people is always the right answer.
---------------------------
From the point of view of folks in this country, sending a pile of steel and $1000 to japan, and getting back a finished car is 100% identical to doing the same trade with yudkowsky's AGI.
Do we like productivity improvements? Or do we advocate a life of poverty, with less stuff and more work? As far as I can tell, that's the only real question. I think it's easy.
Subscribe to:
Post Comments (Atom)
Posts
17 comments:
My claim is that every single one of the examples I give above gives us a better life
Are you sure? I'm not. There are always trade-offs, and if a productivity increase leaves the population restless, indolent, and hooked on health destroying recreational drugs and eating, then maybe the long run benefit is to forego the easier lifestyle for a little more constructive discomfort.
1. "Are you sure?" I don't believe in certainty. Hard question to answer.
I think we have very strong evidence that increased wealth, through cheaper stuff and higher productivity, is the single greatest impactable factor that impacts human happiness (statistically).
I think that falling wealth is incredibly bad juju.
There's a qualitative difference between "We can do these kinds of jobs more cheaply with automation" and "We can do any kind of jobs more cheaply with automation", no? In the former case, most hapless farmers have to retrain to become factory workers. In the latter case, most hapless everybody-with-inadequate-savings have to become welfare dependants or looters or dead.
Roy...
My quick estimation is that as of today...IQ 70 means that you're not competent to work. Hiring you in the USA is more costly than hiring a robot to do the work...and since there are no jobs that you can do that a robot can't...you're screwed. This from my personal experience with such folks..and their work history.
Furthermore...that IQ number for which robots/software agents are universally better than you is climbing by something near 1 IQ point every 1-3 years. We have fewer than 100 years...maybe as few as 30 before 1/2 of the US population is 100% unemployable ... and the total GDP will be substantially higher than what we have now.
The quesiton is not if. The quesiton is when. And how do we adjust?
Imagine if the space aliens left in your kitchen each and every day all of the food, clothing, and other goods you might ever need that day.
You did not produce them, know how to produce them, or control or understand the flow of raw materials that brought them to you.
Should you use them? What are the risks? Could you even stop the population from using them?
I am struck by the idea that the risks of using these resource are huge, and that the urge is irresistible. In fact, perhaps in a single growing season you no longer will even have control of your caloric destiny as a people.
And in terms of goods, in just a few decades the knowledge required to produce almost all of your goods will be lost.
Everything will be controlled by a few nameless, faceless aliens. Better yet, say these nameless, faceless aliens truly are benevolent. Only want to help. Think this is a good way to give everyone what they want and make them happy. What is the result?
Perhaps time to start a screenplay of the easy way to invade and take over a planet. My own version of War of the Worlds.
Note: I am not actually asking any questions here.
HomeSchoolAdvantage -- makes me think of the chimpanzees, raised by humans, which then could not be re-introduced into the wild (successfully, anyway).
We have fewer than 100 years...maybe as few as 30 before 1/2 of the US population is 100% unemployable ... and the total GDP will be substantially higher than what we have now.
The quesiton is not if. The quesiton is when. And how do we adjust?
Interesting. I find this a bit pessimistic from the optimistic Aretae.
What claim do the unproductive have upon the productive? In general, none: think of all the past farmers, their entire lineage in agriculture, now off the land, doing something else, or nothing.
Another example: American Indian reservations, dismally unproductive places, unless they have a casino. They live off past treaty claims, but what is the encumbrance on future generations to honor an archaic claim?
--
We have fewer than 100 years...maybe as few as 30 before 1/2 of the US population is 100% unemployable ... and the total GDP will be substantially higher than what we have now.
--
I'm suspicious of this claim. If GDP is that high, it seems like there would be a lot of money available for jobs that, right now, have too low an ROI to be practical. If wage preferences aren't too sticky, the ROI shifts further in favour of those jobs existing. And if we're in such a high-productivity, high-efficiency economy, transaction costs have probably dropped like a homesick rock and made it easier to get people into those jobs.
For example, right now I'm making a big pot of meat slop for the week. I'd rather pay someone else to gather the ingredients, cook the slop, and pack it into Pyrex containers. High transaction costs make it hard for me to find someone to do that, and prevailing wage preferences probably mean that anyone I found would want more money than I'm willing to pay.
I'm not claiming that, in the future, assembling a week's worth of gainz for strength nerds will be a significant source of jobs (it seems like the sort of thing that'd get automated as part of the productivity increase). But I think this is a general pattern, and as stuff gets cheaper and transaction costs go down people in your "100% unemployable" bracket will be able to find piece-work jobs like the one I described.
RSF + BluntObject,
I hadn't thought of myself as such a sunny guy. This model is part of why I find Iain Banks's Culture novels as persuasive as they are.
Blunt,
You have to consider negative costs as well. Simple issues like Workman's comp make it unprofitable to hire stupid people who can hurt themselves from not paying attention and are likely to. Worse...if I have a house robot with an effective IQ of 70...I can tell it to make meat slop.
The question is not what could someone do...but what could someone do cheaper than a robot.
If IQ<x people* are unemployable at present, but they were employable in the past, is the difference purely a function of the market value of their output falling below the minimum socially acceptable payment?
I could speculate about other factors. The first to spring up is supervision. A substandard worker with effective supervision is far more effective than without. (The benefits of supervision should fall as worker ability increases.) We as a society and in our corporate structures have significantly reduced supervision compared to my impression of Taylor style working conditions.
Why? Well one likely contributor is that the value of the supervisor's time has significantly increased.
Now if we have AI available that can provide 100% dedicated supervision to each worker, the AI can do the stuff that's fairly easy for a computer (monitoring, recording, providing auditory and visual information) while the human does the stuff that's difficult for a computer (complex pattern recognition (eg. which are weeds and which are flowers), or soft touch human-human interaction (come on Mrs Smith, time for your bath))
*Where IQ is shorthand for a combination of factors including intelligence, emotional control, education, psychological stability etc.
Dr. Pat,
Of course re: IQ.
As for taylor-style Scientific Management...it's still a trade.
It may be that there remain tasks which are people-easy but computer-hard...but I'm not convinced it lasts too long. I think customer service is one such...but not something the low IQ folks are real good at.
Some thoughts:
>>Unemployable people--isn't this essentially the Third World? We have entire countries rendered unemployable, or employed at the apocryphal "dollar a day", by bad public policy, yet people still survive and increase.
>>If you can replace 150 people with 15, that sounds most excellent in the long run, and I'm all for it. The problem I see with a lot (99% etc) of claimed productivity increases is that we don't get what's advertised. Same with trade. A lot of trade is done for political reasons, or fads, and in the long run we would be better off keeping it "in house".
Moldbug proposed that useless eaters be "virtualized" -- stuck in virtual reality pods ala The Matrix. Both this and your notion of useless eaters I find highly unworldly. IQ 70 people can do many, many things that computers cannot do. I.e: windows. You got a robot that does windows? Or brush removal, in tight spaces. They can make beds and empty bed pans. They can stand guard.
It is true we don't use IQ 70 people for most things any more, but this is because we have taught them to hate and resent authority, and to use violence to get what they want. At least in the abstract, such cultural programming can change. (I realize that changing it would be a revolution or restoration and thus highly unlikely.)
But yes, that window will move. Perhaps in future robots will be better than humans at escorting the infirm to the toilet. What I do not think can be automated are human services that humans want because of our human nature. For example, it is a pleasure to be served. This is part of the value proposition offered by restaurants. It's not the food, it's having a waitress wait on you, take your order, bring you your food, and a cook to cook for you. (Cooking is another thing that pretty stupid people can handle.)
As I argued in 2008: In the case of robots, unless you want to hypothesize robots smarter than us, harder working, cheaper to make, sexier, better in bed, wittier, and overall better friends, acquaintances, and companions... then we have some absolute advantage over them. So we don't even need to bring up Ricardo and comparative advantage. It may be that humans cease to farm, build things, etc. So what? We'll concentrate on brainwork and service: managing robotic workers, designing products, programming, creating art, acting, waiting tables, etc. If this sounds a bit familiar, there's a reason: substitute "mexicans" for "robots", and "whiterpeople" for "humans", and that's basically California.
If you do want to hypothesize uber-robots superior to us in all things, you've moved us past the singularity, and we're beyond the realm of prediction. Hopefully we can upload ourselves into them.
Leonard,
I think you're counting gross benefit, not net benefit over robot.
My experience with IQ 70 folks is that I just don't want their help. They're SO likely to do it wrong that it's probably better for me to just do it myself. Or skip it. The net cost is lower than the net benefit.
I'm predicting, not guaranteeing. Perhaps the human touch is a comparative advantage, but there's some amount of extra IQ that outvotes the human touch.
[IQ70 folks are] SO likely to do it wrong that it's probably better for me to just do it myself.
But this is a different issue than the one we started with, that these people USED to be able to do something, but have been obsoleted by cheaper machinery and rising performance costs.
Now you're saying they could NEVER do anything useful.
Unfortunately I've never worked with anyone below the tradesman level so I've little to go on here. I'm extrapolating from using child labor ie. getting my nephew to do household chores such as mowing the lawn. As a 6 year old I think I can extrapolate from him.
1. Mowing the lawn required CONSTANT direct supervision. It meant I couldn't do anything else, I wasn't gaining anything. (He was gaining experience and knowledge, that was the point.)
2.Cleaning the floor required little more than going over afterwards and pointing out what he had missed (like under the chairs, and under the dogs)
3. Picking up the toys and separating them into his and his sister's storage boxes was done quite well. Better than I could have managed (I don't know who owns which toy car, though I could probably guess at the pink dolls) and better than a machine too.
4. Taking a beer to granddad was so simple for a six year old, and done far better than any speculative robot, or even a highly trained employee could have managed (when judged on how much granddad enjoyed the service.)
My experience with IQ 70 folks is that I just don't want their help. They're SO likely to do it wrong that it's probably better for me to just do it myself. Or skip it. The net cost is lower than the net benefit.
You haven't been able to train them yourself, have you? This used to be routine -- you'd employ servants, and teach them to do things the way you want things done. These days, the closest you get to a personal servant is a corporationally-intermediated servant such as a "Merry Maids" housekeeper. Or you can hire an illegal under the table; but even then she is not working only for you.
The lack of personal service is largely, I think, a result of our affluence. However, it is also a result bad government policy -- it is a huge PITA to employ a servant within the law. It is also a result of the liberal zeitgeist, where having a servant is considered evil, and being a servant is considered at best an unfortunate temporary situation and at worst horrifying human rights violation.
You've got a bit of an incorrect formulation, a false dichotomy. The choice is not, for example,
A)10 million workers & 100k tons of material to make X cars for $Y
vs
B)5 million workers & 105k tons of material for X cars at $(0.8*Y) and 5 million people unemployed.
The choice is more appropriately between A above & C) 6 million workers & 120 million tons of material to produce 1.2*X worth of cars at ~$(0.9*Y) (or less if techniques can be moved up the supply chain) & 4 million workers moved out of auto-manufacture, a suddenly less pressing need, and into solving some other issue or providing something else that is needed or wanted.
Those pushed into providing something that society now desires more than their prior profession provided may not be happy with the change, but you correctly spotlight the alternative is to use as much labor as possible for any task.
I don't buy that a majority of the population will become unemployable, they just get moved down to solving the next most pressing need or want. The reward might be lower, but cost of living just dropped too, so that doesn't matter as much. Of course if there's a mandated floor on income levels, people can become LEGALLY unemployable.
Pevins,
I think your answer is the correct "at a point in time" formulation...and I appreciate what you're calling out.
However, the over time answer tends to be:
Time A) = your A)
Time B) = 5 million workers and 80k tons of material for X cars @ $0.8y
And most of this comes from Automation.
The zero marginal product hypothesis that I'm pushing is a question of whether at some point, buying a robot is universally cheaper (once insurance and other fixed costs of hiring persons, are factored in) than hiring some classes of worker.
Post a Comment