In our upcoming book, America 3.0: Rebooting American Prosperity in the 21st Century – Why America’s Greatest Days Are Yet to Come (available for pre-order here), Jim Bennett and Mike Lotus paint a word-picture of America in 2040, which is less a prediction and more an exercise in hopeful and forward-looking thinking for conservatives and libertarians. We include predictions regarding the impact of distributed manufacturing.
The recent article in Wired by Kevin Kelly entitled Better Than Human: Why Robots Will — And Must — Take Our Jobs makes similar points. Here are two good quotes from Kelly:
Right now we think of manufacturing as happening in China. But as manufacturing costs sink because of robots, the costs of transportation become a far greater factor than the cost of production. Nearby will be cheap. So we’ll get this network of locally franchised factories, where most things will be made within 5 miles of where they are needed.
***
It is a safe bet that the highest-earning professions in the year 2050 will depend on automations and machines that have not been invented yet. That is, we can’t see these jobs from here, because we can’t yet see the machines and technologies that will make them possible. Robots create jobs that we did not even know we wanted done.
It is important to remember that technological change destroys categories of jobs, and creates new ones that literally cannot be imagined yet.
We are going to be facing a tidal wave of creative destruction in the years immediately ahead.
Our book offers some ideas about why we are well suited to benefit from these changes, and how to navigate the rapids to get from here to there.
Stand by for very interesting times.
“A computerized brain known as the autopilot can fly a 787 jet unaided, but irrationally we place human pilots in the cockpit to babysit the autopilot “just in case.””
This overstates things considerably. For starters, the autopilot can fly and land the plane, but it has no direct communication with Air Traffic Control…ATC communication is verbal to the flight crew, who may then instruct the autopilot accordingly. In principal a datalink could be developed allowing the controller to give commands directly to the autopilot, but no such datalink is in place presently.
Imagine an autopilot attempting to conduct the “simple” task of taxiing from the gate to the runway. What if there is something in the way?…say, a baggage cart left on the taxiway, or another airplane that’s not where it is supposed to be. I’m not aware of any autopilots for commercial aircraft that have artificial-vision sensors and artificial intelligence for dealing with ground obstacles of this type.
And what about the cases where things go wrong? In case of a total engine failure, for example, who or what makes the decision made by Captain Sullenberger to land in the Hudson rather than attempt to make it to Teterboro? If the autopilot was programmed to make such decisions, and it decided on the Hudson, would it be able to evaluate the direction of the swells and decide on the optimum landing direction? What if there are a few small boats in the way?…better to hit a small boat, or turn to a less-favorable direction that may bury the plane’s nose in an oncoming wave? If the autopilot decides on Teterboro, is it going to be able to coordinate with Teterboro ATC or the mechanical equivalent thereof to ensure that the intended runway is clear?
Humans have a lot of flexibility to deal with the unexpected that is very, very difficult to build into robots.
Agreed. We focus more on distributed, especially additive, manufacturing and not so much on robots.
However he is right that manufacturing will not only be “inshored” but localized.
He is also right that we cannot know what kinds of employment will exist as technology rapidly advances.
Also, as to taxiing, it seems this would be similar to the challenges faced by driverless cars on side streets. Maybe easier, actually. But will human control be required in both settings for a while yet. But by 2040, or 2050? I guess it won’t, if Moore’s Law holds, or close to it.
Driverless cars: consider the following scenario. You’re on a road with 2-way traffic and no median, going about 50. A child runs out in front of the car. The only possible way to avoid hitting him is to swerve into the opposite lane, which means a collision with an (automated) heavy truck going to other way…there’s no drive in the automated truck, but the people in the subject car will mostly be injured or killed.
What should the automation do?
I don’t see how Moore’s law helps in this kind of situation.
Today manufacturing mass produces durable goods that people consume and own for long periods of time. I think in the near future this will change in a dramatic way. Manufacturing will become something that people do, and own. It will be used to produce boutique or unique goods and also temporary goods, likely designed for easy recycling. Today if there is something we desire to use (or play with) rarely we either rent it, borrow it, or own it and store it. But what if we could produce it quickly and cheaply and then recycle it after we were done? I think such behavior will become commonplace during the 21st century.
Robin, yes.
Did you read Neal Stephenson’s The Diamond Age? His ideas are similar to yours.
David, there would have to be programming in advance to make that decision.
Machines will never be moral agents so the moral choices will have to be embedded by their makers.
}}} Driverless cars: consider the following scenario. You’re on a road with 2-way traffic and no median, going about 50. A child runs out in front of the car. The only possible way to avoid hitting him is to swerve into the opposite lane, which means a collision with an (automated) heavy truck going to other way…there’s no drive in the automated truck, but the people in the subject car will mostly be injured or killed.
What should the automation do?
I don’t see how Moore’s law helps in this kind of situation.
David, my own position on driverless cars is that most people fail to grasp the problems, and make them overly complex.
It’s kind of like the US Post Office and the holy grail of handwriting recognition. You don’t NEED that. You just need a freaking BAR CODE that the computer can read, and offer a DISCOUNT on any mail that has one. These things can be printed on ANY home computer, on a machine sitting in each post office, and could, basically, make it so that 99% of all mail gets directly, individually touched by AT MOST two people – the guy who takes it out of the outgoing mailbox, and the guy who puts it into the incoming mailbox. WhoTF needs to recognize handwritten addresses? The people too lazy to bother to print out a barcode… and they’ll pay 10-20 cents more a letter for the privilege.
I am sure a lot of this is already done with SPAM, but there’s no reason it can’t be done with freaking day-to-day mail.
==========
Similarly, with cars. The cars don’t need to recognize and understand what is around them — they mostly need to grasp that they don’t want to HIT ANYTHING. And with computer reaction times, this can mostly work, because you have two things working for you — one is that computers react far faster than most people, and they can share information much better than humans can — a car that is about to collide with some object — and has already determined that the collision cannot be avoided by the physics of the situation — can broadcast a signal to any car approaching, which can immediately begin emergency stop procedures. There might be ONE collision but the chances of a series — as happens all the time and is one of the major factors in highway deaths due to the speeds involved — can be avoided.
I believe the casual street-level options are currently a bit tougher to define and may not lend themselves to simple control, but highway situations, you could totally automate those RIGHT NOW. Not only do highways have limited chance of random objects drifting into the path of traffic (i.e., kids, dogs, and so forth), but they also have far greater line-of sight and line-of-radar open to them to act accordingly with far faster response times than humans can.
More critically, they can negotiate with one another to minimize or eliminate damage/collision — if one car needs to swerve four feet into the adjacent lane to miss an object, then it can do so, and tell the car next to it in that lane that it needs the space… and then THAT car can dodge two feet onto the sidebed to avoid a direct collision. Humans can’t even HOPE to attain that kind of coordination to avoid a crash. And at the same time, those two cars are broadcasting the presence of an object in the road to all the cars nearby so they slow down and can increase their reaction time to any unpredictable behavior.
I believe that the highway scenario can generally be vastly improved over the current human-driven situation… but it’ll take some time before people really trust the computers to drive more than themselves… which is the main issue.
And I’d also comment that this can also vastly increase highway efficiency — not only can cars actually draft one another — no reason a collection of cars can’t create a “train” of 10-20 autos all increasing the gas mileage of all but the first one by travelling five feet behind one another — but cars can also anticipate the exact minimax need to slow the car to keep a consistent speed going as much as possible when slowdowns are required. And lastly, at least ONE reason people want to speed is because driving takes their attention and time away from other things — so some people may be willing to just take a slower, more fuel-efficient trip if they can be playing Half-Life the whole way from Atlantic City to Gramma’s house in Poughkeepsie.
Highways are and should be the first place we put computer-controlled cars. And yeah, there needs to be a lot of pre-testing done by QA types to invent scenarios and make sure the software responds correctly and even has a response to various scenarios.
I mean, when it comes down to it, the first rule of driving is actually pretty simple: Don’t Hit Anything. ;-)
It is already happening. We all feared at one point that even large things like cars and plane would be imported from China, we all imagined in horror the ships and containers with hundreds of thousands of chinese made cars landing in our markets. But it won´t happen at all. Not only it is not going to happe but there are reversing trendsÑ The costs of transportation has made it impossible for China to compete in the production of large domestic appliances, like refrigerators, washer and drier machines, even for microwave ovens, that are sold in North American and European markets. Whirpool, Mabe and other big producers are now establishing in Mexico or in the US, combining manufacturing of some parts across the border to gain advantage of the NAFTA agreement while other manufacturers have decidedly to stay in the US and compete from home. The same is happening for large plasma and lcd television sets, no longer is China the leading producer and exporter in the world, they are now being produced in the US or in Canada or Mexico, always closer to home, because of transportation costs and time. China has lost its competitive advantage for manufacturing of many products already.
In addition rising costs of transportation, to technological applications such as automation and robotization impacting production costs, you also have customization, regional and local specialization. I read of a woman somewhere in Seattle who, having lost her job in a textile factory producing baby clothes that closed doors and moved to China, she decided to open her own shop at home, as opposed to her former employer, she was able to select better materials and to customize them, with the baby´s and mother´s name, dates, colors, etc. She was able to cater to her own local community, offering a specialty product at a higher price, while the company operating from China could only offer cheap products at a WalMart center.
Sounds like a very good book to read.
“But as manufacturing costs sink because of robots”
Manufacturing costs have been sinking because of technology..including robots..and better process management..for a long, long time. Is there really evidence that today’s technologies are a significant deflection from the long-term trend line?
Consider:
*A Spinning Jenny (1764)…which was not even a power machine…could do as much work as 6 or more women at traditional spinning wheels. Textile machinery of the early industrial revolution offer much greater efficiency improvements.
*Blanchard’s copying lathe (originally used to make rifle stocks) allowed the making of complex shapes without skilled labor; indeed without much labor at all. This was circa 1850.
*Numerically-controlled machine tools, which greatly improved the economics of small-lot production, were developed in the early 1950s and were pretty common by the late 1970s. (Interestingly, one early approach to machine-tool automation was a record/playback system very similar to the training mode that Kevin Kelly describes for the robot called “Baxter.”
*Mainframe computers were introduced to offices in the 1950s and were quite common by the late 1960s: they eliminated vast clerical forces involved in things like life-insurance policy billing.
Etc.
There are unquestionably many innovations in intelligent machines which lie ahead, but I’m not convinced that the productivity impact will be greater than previous innovations such as those described above. I also think that much of the potential productivity improvement, in manufacturing and elsewhere in the economy, lies in the better organization of work (including making better use of existing technology) rather than being dependent on new technologies per se.
An interesting website devoted to small-scale manufacturing is Kathleen Fasanella’s Fashion Incubator. Many of the articles will be over the head of those who (like me!) don’t know the trade, but others are relevant to manufacturing in general.
Interesting times are coming.
One of the interesting effects of driverless vehicles may be the reduction in the number of large trucks on the road, and increase in the total number of smaller trucks powered by natural gas. At the manufacturer shipments could be packed in the vehicle for direct delivery to a single customer. They could then be sent over road or rail, driving 24/7 without rest stops. Look for lots of unemployed teamsters and strife when the introduction begins. Remember the etymology of sabotage.
The customized, localized, additive, distributed (CLAD) manufacturing revolution will come, but it will take decades. 2040-2050 seems a realistic time frame. What are the consumer products that will first be manufactured in this way? Fashion is my candidate and David Foster’s link is interesting. And shoes seem like a strong candidate as our feet are different sizes but we must always buy a pair of the same size.
There’s a lot more manufacturing in America today than is often realized….the stats show China and the US as about equal, with each at about 19% of world output.
One factor that has contributed to offshoring of operations that really should have stayed onshore is sloppy analysis by executives, consultants, and finance people who failed to adequately consider the impact of TIME on profitability. If you have to order your stuff in large batches and it takes a month at sea and Chinese and American inland transportation, then it creates problems with inventory carrying costs, lost business due to stockouts, unrealistic expectations for accuracy of demand forecasting, difficulties in responding to fashion or technological trends, difficulties in responding to quality problems, etc etc. One would have hoped that after decades of graduate education in management more people would have been able to at least make a stab at this kind of analysis, but evidently not.
Our friends at Evolving Excellence often write about these factors. Note Kevin’s post today about the 14-knot floating banks.
“Look for lots of unemployed teamsters and strife when the introduction begins.”
We speculate that self-driving passenger vehicles will be first for this very reason.
“The customized, localized, additive, distributed (CLAD) manufacturing revolution will come … .”
Mrs. Davis, what is the source of this term? Did you originate it?
Our friend Google seems to blithely unaware of it.
I guess so. It seems to encapsulate all the critical factors at work. The one that wasn’t discussed much here was customization but I think it is a critical advantage.
The advantages of driverless trucks are so immense that it is hard to imagine it will take long after driverless cars are introduced for them to make inroads. The incidence of accidents where a driver crashes into an automated passenger vehicle that was unable to evade the driven vehicle will create sufficient public pressure. No matter when it happens, there will be strife.
“It is important to remember that technological change destroys categories of jobs, and creates new ones that literally cannot be imagined yet.
We are going to be facing a tidal wave of creative destruction in the years immediately ahead.”
We are already in the tidal wave, but it has ntohing to do with robotics or other factory automation technologies.
Consider: Product leaves factory in China with a cost of $8. It is trucked to the port, shipped via container ship to US port, trucked or railed to distribution center, trucked to big box retailer, then finally sold for $24. Customer’s choices are limited to handful of product options stocked by big box retailer. 2/3 of the price the consumer pays is for non-vaue adding costs and profits for logistics and retailers that do little or nothing to add value for the end customer.
Alternative: Product leaves US factory with a cost of $18, shipped via UPS directly to consumer who bought the product on line and paid $23. Consumer pays lower price, gets superior US value, selects from virtually unlimited range of choices on line … better deal all the way around.
The technology that is creating the demise of ocean freight, logistics companies and retailing is the Internet and the ease and security of on line buying. Walmart et al are struggling to see how they will be relevant in the future, and their executives lie awake at night worrying not about their brick and mortar retail competition, but the likes of Amazon. Video stores, book stores, music stores have all been killed by the Internet – exposed for adding cost but not value.
With some 18% of all retail purchases coming on line and rapidly increasing, the millions of people, and trillions of dollars worth of land and buildings invested in the old retail model it will not go down without a fight, but it will go down. That handwriting is already on the wall.
US Federal, state, county and local laws and regulations are an enormous bar to hiring and employment. I have taken an oath, along with 52 friends, never to provide a job in the US because of complicated IRS withholding and reporting laws.