Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
I, Robot, Am Not Taking Over any Time Soon
In “Conservatives are Too Quick to Dismiss the Rise of the Robots,” James Pethokoukis worries that whereas in the past, technology has given rise to new jobs to replace those lost to innovation, this time it may be different.
James provides us with an excellent specimen of the kind of thinking that constantly causes macroeconomists, politicians, and other self-styled high-level thinkers to make serious errors when analyzing changes to economies and human societies. I’m not picking on James, who’s an otherwise excellent analyst, but on this error, which is so common that it really needs to be discussed.
This phrase, in particular, jumped out at me:
Just think about the progress made in autonomous vehicles and the fact that the most common job in most states is that of truck driver.
His statement displays a top-down approach to analyzing the problem. Truck drivers drive vehicles. Soon, vehicles will be autonomous. So, no more truck drivers. But this kind of of analysis doen’t reflect the true complexity of truck driving.
And this isn’t just about truck drivers. To explain the problem and why it matters so much, I need to digress. The economy is a complex system. Society itself is a complex system. Workers in a society have to be productive within this complexity. These systems are very different from, say, a complicated machine. A complicated machine can be understood in a reductionist way: Take apart a motor; understand the various parts and what they do; and you can understand what the motor does. If you have full understanding of the motor, you can treat it like a black box, with inputs and outputs, and ignore the complicated workings inside. Engineers and scientists use this type of analysis to break down complicated problems and organize them into simpler ones.
Complex systems turn this upside down. A complex system looks simple on the surface, but becomes increasingly complex as you drill down. Complex systems are more than just a collection of parts – their behavior is governed not just by what each part does, but by the interactions among the parts. For example, you can’t understand a brain just by learning what neurons do. You must also — at least — understand the web of billions of neurons and the interactions among them.
The other problem with analyzing complex systems from the top down is that these systems function by means of feedback from the bottom, which causes constant iteration. If the price of steel goes up, that information changes the behavior of steel producers and consumers. That, in turn, causes its price to change again. The new price may create more consumers or more producers, or cause manufacturers to substitute other materials, which in turn causes the prices of those materials to change, and so on, ad infinitum.
You can think of such systems as a kind of self-programming computer: They constantly take in data, process it, and in response change the output. This process of feedback and constant change makes these systems very sensitive to initial conditions; as a result, seen from on high, they are opaque, and behave unpredictably Hence the most common word in a macroeconomist’s vocabulary is apt to be “unexpectedly.”
Conservatives tend to understand this, because the thinkers we tend to read and follow understood it. Adam Smith’s phrase, “the invisible hand,” suggests how well he understood the way emergent properties drive complex systems. Hayek’s opposition to “scientism” and the pretense of knowledge were an evocation of complex systems theory. In fact, Hayek is considered one of the early contributors to that field.
Statists believe that the economy and society can be treated the same way. If they can find the levers that control the economy, the smart people at the top can push and pull on them and drive the ship of state. Social scientists want to be mathematical and scientific, just like the engineers and physicists, so they go through contortions to create models decorated by a few numbers and formulas into which they can be plugged. They use these to justify applying “scientific” techniques to managing people and the interactions between them. This is what Hayek called “scientism,” not science.
When macroeconomists reduce the economy to aggregate variables like GDP, employment, capital, inflation, or the consumer price index, they’re abstracting away everything that really matters in an economy in favor of a few numbers that are amenable to mathematical modeling. These numbers may indeed be useful when trying to understand the state of an economy, but the variables can’t be tweaked by central planners in sure confidence that the outcomes will be predictable. Attempts to do so lead to unintended consequences and to the destruction of the feedback forces the system needs to remain healthy.
If you’ve never read it, I highly recommend reading the classic essay I, Pencil by Leonard Read. It’s a perfect description of the way complex systems deceive people who look at them only from a very high level. If you ask someone how hard it is to build a pencil, they might think about it and say, “Oh, not hard. You need a wooden dowel, a hole drilled in it, and some lead or graphite to fill the hole. Glue it in, and you’re done.” But as Read’s pencil replied in the first person, “Simple? Yet, not a single person on the face of this earth knows how to make me.“
The essay drills down into the construction of the pencil. You need some wood. Fine. Where do you get it? Will any wood do? Or are there special characteristics? And how do you get this wood? Chopping down trees? How do you do that? With an axe or a saw? How do you make an axe or a saw? Oh, you need a steel axe head. How do you make steel? And so on, and so on. Spoiler alert: By the time you walk down just a couple of steps of production, you find efforts that require thousands of people, each with specialized knowledge the others do not share. It’s an incredibly complex endeavor, and the amount of economic and physical coordination required to make pencils is astounding.
The reality of complex systems is the reason conservatives oppose central planning. Hayek knew this, and it formed the core of his arguments against an overweening state, the supposed superiority of macroeconomic modeling, and decision-making by central authorities.
This failure to see hidden complexity is not limited to politicians and economists. Most engineering projects that run over budget, or that fail completely, do so because of a failure to take into account hidden complexity lurking in the details. Software engineering has moved away from top-down design and toward bottom-up, iterative development cycles precisely because it better matches the real world. The largest, most carefully thought-out architecture developed by people in the head office generally doesn’t survive contact with the real world, which is why that type of development isn’t done much any more.
Now back to the truck driver. Can a robot drive a truck? Maybe, on a well-documented road, and under unexceptional circumstances. From the high-level view, that answers the question. But if you ask a truck driver what he does, you might find that he also loads and unloads cargo. And if you dig into that activity, you might find that he needs to rely on years of experience to know how to do that safely and efficiently with the load properly balanced and secured. He may be required to act as an agent for the company, collecting payment and verifying that the shipment matches the manifest. The truck driver is also the early warning system for vehicle problems. He has the knowledge and judgment to be able to tell if something is wrong. A rattling sound on a road full of debris might not be a problem. The same rattle heard on a smooth road? Might be a problem.
The truck driver is the coordinator of on-road repairs. His presence protects the cargo from theft or tampering. He deals with many different end-customers, many of whom are still using old-fashioned paper manifests and invoices a computer can’t deal with. He may use his judgment to determine if a check can be accepted for delivery. Each customer’s loading dock may have hazards and unique maneuvering difficulties. Then there are the ancillary benefits of human truck drivers – they cement relationships with customers. They spot opportunities. They report traffic accidents or crime to the police. They notice damaged goods in a shipment. Sleeping in the truck protects it from theft.
These are the things off the top of my head, and I’m not a truck driver. I’ll bet if you asked Dave Carter what he does, he could go into much greater detail. And if you asked other people in the chain, they’d have their own set of complexities that are part of the entire work process called “truck driving.”
Robots don’t do complexity well. They are excellent at repetitive tasks, or tasks that can be extremely well defined, and which have a fixed set of parameters and boundary conditions. A robot on an assembly line knows exactly what it has to do, and the list of potential failures (parts out of alignment, defects in materials, etc.) are well known. Even a self-driving robot car needs to know what the road looks like — Google’s cars use pre-mapped road data — and it can’t deal with situations that are very far outside the norm.
We are making strides here, and Google’s robot cars have a surprising amount of autonomous decision-making capability when it comes to things like cars stopping in front of them suddenly and obstacles on the road. But that’s a far cry from the kind of generalized human judgement required in most occupations — which is why the robots won’t be taking over any time soon.
I can’t say the same for the central planners. We seem to be stuck with them.
Published in Culture, General, Science & Technology, Technology
I find this convincingly comforting. Thanks.
Well, robots are as good at handling complexity as the guy who writes the software is at programming the robots.
The reason autonomous cars will be a thing, sooner than later, is that driving from point A to point B is really a decision problem. There are unambiguously correct answers to every possible maneuver you perform in a car.
Oddly enough, you point out that we’re stuck with central planners, actually I wouldn’t be too sure of that. If anything their jobs are the most easilly distillable to deterministic algorithms. They require zero creativity, distill to following inflexible rules, and they make no decisions themselves. The jobs of most bureaucrats also fit into this category.
So yeah, the robots are coming, for the bureaucrats, and I can’t wait to work on the software that replaces the federal government employees with a bunch of microservices.
In some respects, autonomous trucks might be a good thing; they are not limited to the terms of OSHA, and can automatically calculate routes to avoid traffic. I wouldn’t be so much concerned about the future of the trucks themselves, though. Drones are beginning to drop off packages same-day. Couldn’t the same happen in the future with drones?
I think you’re missing the point: Central decision-making is not simple or deterministic – it’s hideously complex – far too complex to ever work efficiently. Which is why it doesn’t work now, even with all the big brains from the Ivy Leagues attempting it.
And you’re making the same fallacy that James made when you say that autonomous cars will be here soon because driving from point A to B is simply a decision problem. It may look like that at the highest levels, but once you dive into the problem you’re going to find out that it’s far more complex.
Show me an autonomous car today that can, in real time while driving, determine that the thing on the road ahead is a potato chip bag that can be driven over vs a rock that cannot. Or one that can intelligently drive through an unplanned construction detour onto a temporary road that isn’t in its map, and which is being directed by humans pointing the way to go.
Before you see autonomous cars on the road you will see cars that have an autonomous ‘mode’, but which still require a human behind the wheel to handle situations that the computer can’t handle. That’s starting to happen with lane avoidance and automatic braking and such, but it’s a long way from that to a car that can just drive itself on any road.
I’ve tried the new lane-keeping features in modern cars. They have an indicator that tells you when the system has enough information to work, and when it doesn’t. And that indicator goes OFF a lot.
Absolutely not true, unless you mean to say, “Robots will be able to handle complexity when we figure out how to program a robot to handle complexity”, which is a tautology.
We are at the stage in artificial intelligence where we are starting to be able to build robots that can identify things in images, match faces, understand simple sentences, and walk upright while staying in balance. Things that a human baby knows very soon after birth.
However, these are rather mechanistic attributes. Pattern-matching, computing balance, etc. Things a computer should be able to do well. Now ask a robot to make you a tasty sandwich in an ambiguous situation like a private home where stuff gets moved around, milk winds up on the wrong shelf in the fridge, some ingredients have to substituted, and all the other decisions we make for such a simple task, and which can do it in a generalized way in any home. You can have the best programmers in the world helping you. Let me know how that goes.
We can also make a labor vs. capital argument. The market will tend to evolve toward the lowest cost combinations of those two.
A robot that can completely replace a human is an extreme combination of 0% labor and 100% capital. You only get there if 1.) humans are unwilling to do the job for the price, or 2.) there is an artificial wage constraint, such as the minimum wage + Obamacare mandates + employers’ portion of Social Security + unemployment insurance + workman’s comp + union dues + sick pay + vacation + cost of lawsuits.
By the way, the auto manufacturing industry already has robots. I suspect the artificial wage constraint is what happened there.
I agree almost all the way but the coming wave of technology will be different in a very important way. It will automate the middle.
Past developments tended to displace people at the bottom while opening up opportunities above. With some effort you could move up into something better.
To use the truck driving example, truck drivers won’t be completely displaced. But they will be split. Some will move up into a kind of logistics operator handling multiple trucks from a distance. Others who don’t have those skills will be bumped down to something less.
The top 40 percent of jobs and bottom 40 percent probably can’t ever be roboted out because of the people element. But much of the middle 20 can. About half of those people will probably join the upper 40 and about half the lower.
For that 10 percent going lower it’s gonna be a rough ride.
S.D. Curran says:
I don’t believe there are any drone-based delivery services at all. Drones won’t be dropping off packages for a long time, if ever. I believe that was an Amazon publicity stunt. And while navigating a drone may be in some ways simpler than driving on a road, there are still numerous complexities when trying to make the scheme actually work for the public and at a scale that makes a difference.
As a simple example: Are we going to ground all package delivery when the wind blows or it rains? Quadcopter drones don’t handle bad weather well. If not, you have to maintain a traditional delivery infrastructure anyway, so what’s the point? Also, do you know how much noise a drone makes? One large enough to carry a heavy package? And the safety issue when a drone blows an engine over a built-up area?
It’s always tempting to say that we’ll just ‘figure out a way’. But there are some futuristic technologies that have issues far beyond what we can deal with today. Notice there aren’t any flying cars over our cities? They’ve been predicted since the 1930’s. They never happened – and they never will. Because they don’t make sense once you dive into the issue deeply and understand the root problems.
Liberals make the same error when it comes to alternative energy. They see a farm being powered by a windmill and just assume that we could put windmills everywhere and problem solved. They see solar power as just a few government research dollars away from solving all our problems. How hard can it be, really? The answer: Incredibly freaking hard.
Such is the way of the ivory-tower intellectual. :)
Probable Cause said:
My main job is in factory automation, and I’m working on a project for an automated car assembly plant right now. I’m very familiar with automation and its limitations.
A typical car robot is a movable arm at a station with some kind of tool on the end – a welding gun, a camera sensor maybe, or a drill. Its job is to wait until the part moves to its station on the assembly line. It then looks up the specs on the part so it knows its exact dimensions, and the activity to be done – “drill 12 holes exactly 1″ apart starting from 1″ from the top”, or “Pick up part B with a suction cup arm, position it on the jig, and spot-weld it in place.”
Some of these robots are very sophisticated – they can make quality determinations with vision systems (i.e. spotting a blemish in the paint with a camera), they can handle different kinds of parts (assuming each kind has been programmed in), etc. But they are essentially mindless automatons going through the same motions over and over again.
Automation wasn’t as much a response to the cost of labor as it was a method for improving quality. In manufacturing, you want repeatability of process – for something to be done exactly the same way, every time. That way, you can instrument the process, examine the defects, adjust the process to get rid of them, and improve quality. But when humans are in the loop it’s much harder to control for quality because humans make random errors, intentional errors, errors that change based on how long they’ve been on shift or who they are on shift with, yada yada. Some of them will lie about what they did to avoid punishment or embarrassment, and mislead the control engineers. And so on. Sorting all those issues out is hard, and controlling for them even harder.
One of the reasons Japan started eating America’s lunch with car quality was because the Japanese embraced automation while the American auto workers fought against it and delayed its implementation. So the Japanese cars were much higher in quality, and that forced American companies to follow suit.
Casey said:
It’s not that simple. Certainly the first wave of industrialization which replaced raw muscle power with machines affected manual laborers the most.
But computerized automation has not been like that. Look at the industries that have been leveled by automation so far: Switchboard operators. Draftsmen. Typesetters. Magazine publishers. Recording engineers. Cartographers. Accountants. Machinists. Bank employees. These are not jobs at the bottom – they are often jobs that in the past required highly trained, well-paid people, but where the job itself was well defined and repetitive.
For an early example, the word ‘calculator’ used to mean a person who had a job simply computing numbers. An astronomer might make observations and then hand them over to a ‘calculator’ to compute the orbit of the object. Statistical analysis was incredibly tedious, but required highly trained people to run the numbers. Those jobs no longer exist because even simple computers could do the job faster and better.
Or maybe I’m making your point for you, if you’re extending your point back to the start of the computer age.
There will definitely continue to be job losses due to computerization and automation. But there’s no evidence to believe that the future will be fundamentally different than the past in this regard. The mechanization of agriculture destroyed tens of millions of jobs and caused society to radically restructure.
However, you can also look at it as a force that freed massive amounts of human capital from low-value work and made it available for other things. And the presence of that huge pool of labor and the increase in wealth from mechanization solved that problem in short order.
We don’t know what the future will look like. But we do know that humans are a very valuable resource. We’ll find uses for them.
I should add that there is a big caveat to all this: If enough people buy into the notion that displaced jobs are ‘not coming back’ and use that reasoning to push through massive welfare programs and or a ‘guaranteed living wage’ for displaced workers, then you WILL see massive permanent dislocations of labor.
Imagine if politicians had responded to the mechanization of agriculture by giving huge income subsidies to ex-farmers, so long as they remained living on their defunct farms. That would have caused a stagnation and a permanent underclass of unemployed farm laborers, frozen in place by their growing dependency on subsidy. And then the same politicians would have used the presence of these permanently unemployed workers as proof that their predictions were right and their subsidies necessary – ignoring the fact that it was the subsidies that caused the problem by removing the incentive to take the hard steps necessary to thrive.
This is the ultimate risk of buying into the notion that the robots are coming for all our jobs – it will justify government intervention in propping up workers who lose their jobs, which will remove the incentives people need to find a way to adapt to the new reality.
Also, the internet killed off thousands of corporate phone support jobs. Remember before the internet, and you had a problem with your printer or your thermostat? You had to call the Acme Corporation between the hours of 9 and 4 (eastern time), and wait on hold for the next available help desk person in order to ask your question. Now you Google it, any time day or night, and you immediately get the pdf of the owner’s manual for that exact make & model, frequently asked questions & answers, and unauthorized guidance from people having the same problem.
We’re talking about different things. I’m talking about automating the bureaucrats who enforce rules already made. The people who make the rules are a different problem, but also not the bulk of federal employees.
Think of TurboTax software. Yes the tax code is hideously complex, but it can also be solved by a computer with enough data. No one really needs to do anything except push enough data to complete taxes.
Also, I think you’re wrong about the driving. The holistic act of driving boils down to a myriad of discrete tasks, each of which are decision problems. Yes, it’s complex, but each task is not, and we can figure out how to make that work even if it’s harder than we think it is right now.
Two things, 1) I am extending back and I think it all adds up to a big sudden boom in the near future. And 2) this is temporary in that future generations will think nothing of the new reality. But this next 20 years will be disruptive. And I don’t like hearing conservatives trying to sell the big picture to a displaced truck driver trying to feed his kids.
Dan, great post. Great comments.
I’m smarter for having read this. Thanks.
Probable Cause:
You’re absolutely right about the tech support jobs. ‘Librarian’ is another job that seems doomed, among many others.
But here’s why it always seems like ‘this time is different’: It’s easy to speculate on the jobs that will be lost, but impossible to know the jobs that will be created. So it always seems like a one-way street.
It was easy to see that switchboard operator and draftsman were doomed professions long before they actually vanished, but no one knew that they would be replaced by desktop publishers, web designers, javascript programmers, bloggers, youtube performers, and the myriad other occupations that have been created in the online world.
What will be the jobs of tomorrow? Who knows? If I knew where the growth industries would be I’d be a rich guy. Did you know there are people making a living making virtual products for sale in virtual worlds? There are virtual dress designers, virtual architects, and they make virtual goods that they sell for real money for use in virtual worlds.
We are on the threshold of revolutions in materials (graphene, nanotetch), in virtual reality, AI, and a few other key enabling technologies each of which could spawn a revolution in how we work and play. Maybe 10 years from now you’ll be able to make a living doing virtual tasks in a virtual world. Maybe nanotech clothing will create whole new industries for people who can marry clothing design with technology. Maybe ‘drone pilot’ will be a huge growth industry. We really don’t know.
The fact that the job losses are more visible than the jobs that will be created is at the root of a lot of this anxiety – and that’s been true for 200 years.
Excellent post, Dan. I work with steel fab factories from small job shops to major corporations, and I hear a lot about how automation is wiping out these jobs. That’s different from the reality of what I see. There are some robots doing welding and such, but they are the exception and not the norm.
This is a really interesting article, a lot to chew on. I’m one of those that wonder how we’ll handle the problem of inactivity for a large percentage of society as other areas approach the efficiencies of agriculture, so I’m glad you addressed that specifically. I hope you’re right that vast new areas open requiring humans.
“So yeah, the robots are coming, for the bureaucrats…” This, I hope, is also right. It’s one area in which I’ll welcome our new robot overlords, as Instapundit is wont to say. I’ll take a cold, metallic heart over no heart at all.
I’ve said it before and I’ll say it again. And again if necessary. We already have an item that does 95% of what people note as the advantage to robotic trucks. They are called trains.
Wake me when you have a rig that can navigate between rows of other trucks in a crowded yard and can back itself into a dock while opening its rear doors.
Dan, I REALLY enjoyed this and your follow-on comments. Very thoughtful, and insightful.
I think lines can be connected between the complexities of human work and the issues of any complex system, such as the models that are used for a wide range of things – but most famously for Climate Modeling. I was quite impressed with this report. It essentially says:
In other words, modeling these systems may ACTUALLY be impossible. And I think that driving down a road that is under construction and has ambiguous signage, obstructions, cones, potholes, etc. is a perfect example of how decision-making cannot be made using any software model, no matter how complex. Human decision-making in those circumstances are clearly effective and yet probably incompatible with any software description that can be run repeatedly.
Something that cannot be predicted is quite hard to write code to govern. I think.
It is an interesting question. What will everybody do when robots take all our current jobs?
Cat videos and shoes.
Humans have an infinite capacity to consume cat videos and shoes.
Once someone develops the Artificial Intelligence (AI) that is capable of determining if a cat does something cute or funny and/or design an attractive running shoe humans are screwed.
Driving really isn’t that complex. Particularly if all cars are automated. Accidents will plummet. Cars will be able to know well in advance whether a particular route may be faster. Traffic will be almost non-existent. The whole system would be way more efficient.
A slightly different take here. I expect a law we all know is at the bottom of it all: work expands to fill the time available for it.
The thing about central planners is that they’re clever and unscrupulous. Milton Friedman suggested a computer (and a 1982 IBM PC would be overkill) could run monetary policy. Note that today the Federal Reserve is still going about its business and Milton Friedman is dead. QED.
Casey, Speaking as the victim of “sudden unintended acceleration,” a computer problem which, diabolically, does not show up on computer diagnostics, the idea of more computer control is not a comfort. Then there is the issue of hacking into car computers. I think I’d rather deal with bad drivers than hacker mayhem.
Why do I have this feeling that anonymous or Great Ghost of Godel will soon show up to tell us how robots will be able to supersede humans in 10-20 years?
Very good article. The failure to understand complexity is at the heart of our dysfunction. Macroeconomists, the political class, many corporate managements, almost all intellectuals and most importantly, the regulatory state in all of its glory act as if the world were mechanical and it’s future knowable. We will adapt to robotics if allowed. We will be allowed if we abolish most of the regulatory state, replacing it with the rule of law. We will fail to adapt, or our adaptation will be unnecessarily painful and long if we try to control the process under the illusion that we can.
John, I don’t see it as a quantitative (how much computing power) kind of issue. I see it as a qualitative issue having to do with things that are unknowable because there is no certain future outcome that can be teased out by a computer, no matter how powerful.
I guess I would say it is related to the question of how deterministic the future really can be. My answer is that complex systems are actually unmodellable – as per the reference I brought up above. Better computers have not led to better near or long-term weather prediction because the weather is far more complex and non-linear than any model can capture.
It is like my argument in an earlier thread about the dangers of the notion of perfection. Society (and a free market) works in large part because we have thousands or millions or billions of free agents, each deciding for themselves what they think is Good. Because there IS no Right Answer for what is good or perfect. Nor should there be. If there was, then in the computer-controlled society of the future, then a top-down controlled Communism WOULD work after all.
Don’t cross the streams, John.
There is a perfect that we all must work toward. If not then we are just computers ourselves processing the world around us. And John is right. If that’s what humans are then computers will eventually do that better.
But computers will have no concept of moral perfection or perfect beauty. Nothing will be better or worse than anything else. It will all be simply process.
Isn’t the main hurdle for emerging technology lawyers and insurance? When an autonomous vehicle gets into an accident with an human vehicle or worse, a human pedestrian, who gets the blame? It the computer going to be assumed to be infallible? Is the government going to give automated vehicles some sort of immunity? Some delivery drone is going to injury someone someday. Then what? New safety regulations and new bureaucratic rules that will crush the emerging technology. Some class action lawsuit will allege some harm, like power lines supposedly causing cancer. No major infrastructure can be built without years of environmental impact studies.
In this era of big government and Big Legal, nothing new of note is going to be created above the level of gadgets like phones and tablets.