Future market restructuring needs to be taken seriously

[FallenApple]

Well, some research shows that a large portion of jobs will not exist in the future. The estimate for the proportion of jobs at risk is 47%. Pdf linked at bottom.

We examine how susceptible jobs are to computerisation. To assess this, we begin by implementing a novel methodology to estimate the probability of computerisation for 702 detailed occupations, using a Gaussian process classifier. Based on these estimates, we examine expected impacts of future computerisation on US labour market outcomes, with the primary objective of analysing the number of jobs at risk and the relationship between an occupation’s probability of computerisation, wages and educational attainment. According to our estimates, about 47 percent of total US employment is at risk. We further provide evidence that wages and educational attainment exhibit a strong negative relationship with an occupation’s probability of computerisation.

This really leads to the question? What type of jobs will remain? Just jobs that require higher abstractions? Any opinions?

Clearly, unskilled jobs are the first to go. Along with any task that can be currently computed, i.e good motor skills, calculations, typing even. So repetitive tasks are out.But will this open new jobs that are more creative in nature? Perhaps more jobs for mathematicians and physicists? What about the field of software engineering/data scientist? No matter how intelligent the AI becomes, it still cannot have a Sherlock Holmes style of deduction, which means analyzing data for scientific purposes still should be around.

Skills that I think will remain are: proving theorems, writing novels, deducing cause and effect, etc. So basically: Creativity and Pattern Recognition.

 

[Orodruin]

The point is that humanity is in a situation where a major part of the population is not needed in order for society to sustain itself. This results in a situation where we as a society need to figure out how this should be handled.

 

[Choppy]

I can't help but wonder if there is a way to see this as more of an opportunity than as a change toward a dystopian future that will see half of the workforce unemployed.

A downturn in the need for "repetitive" type jobs may result in an upturn in other types of jobs that just aren't there because no one has the time to do that. Consider service industry jobs. And I don't mean just waitresses and baristas. There was a time in western society when many people were employed as butlers, maids, gardeners, etc. In the near future with the aging population there's going to be an increase in demand for geriatric care, for example, and I'm not sure that's something that robots are going to be able to do effectively. Even if they could effectively feed, toilet and bathe people, there's a simple caring element that needs to be there.

I remember a character in the movie Demolition Man - "Associate Bob." To the best of my recollection (and it's been over 20 years since I've seen the movie) he was basically a professional friend. In the real world this could manifest into things like increases in counselling, social convening, historical interpretation, debating, personal training, skill training, personal protection, etc.

A lot of this of course depends on an effectively managed trickle-down. With machines doing more work, the potential is that those who own the machines are going to make the most money and there won't be much for the rest of the population to share. You can't hire a butler or a tennis coach if you don't have an income in the first place.

 

[FallenApple]

The point is that humanity is in a situation where a major part of the population is not needed in order for society to sustain itself. This results in a situation where we as a society need to figure out how this should be handled.

I can see how this might ultimately turn out to be a good thing. But I'm worried about the transition period. It could take decades, and during those times, people will be unemployed.

 

[Evo]

I can see how this might ultimately turn out to be a good thing. But I'm worried about the transition period. It could take decades, and during those times, people will be unemployed.

How is this different from any other time in past history? Please give examples, from say, the industrial revolution, just for one.

You do realize that every time we progress technologically, we put people out of work. What about the book business? What has the internet done to that business? Now anyone can self publish online for virtually nothing.

 

[russ_watters]

Well, some research shows that a large portion of jobs will not exist in the future. The estimate for the proportion of jobs at risk is 47%. Pdf linked at bottom.

This number is not meaningful unless compared with other times in history.

 

[russ_watters]

The point is that humanity is in a situation where a major part of the population is not needed in order for society to sustain itself.

That has never been true before and I see no reason being presented why this time will be different. People keep saying this time will be different without ever presenting a reason.

I think this next disruption will usher in a new era of prosperity and standard of living growth - just like all the others did.

 

[Orodruin]

That has never been true before and I see no reason being presented why this time will be different.

I am not saying that people will go unemployed. I think you are misreading my post.

 

[russ_watters]

I am not saying that people will go unemployed. I think you are misreading my post.

Evidently I am. I can't reconcile it with what you are saying now. Can you clarify?

 

[Orodruin]

Evidently I am. I can't reconcile it with what you are saying now. Can you clarify?

We are already at a point where the majority is not needed for sustenance of society. I am working as a scientist and getting paid for it although I do not contribute directly to keeping us alive. We also have reduced working hours compared to pre-industrial society - also signifying a lower need for using the population as labour to produce what society needs.

I think that, as a society, we should be happy about this and make sure to use it to provide sufficient sustenance and health care for everyone. Of course, you can add qualified jobs in other additional sectors (such as research), but since we have the capability to make sure that people have descent lives - why would you not do this?

 

[russ_watters]

We are already at a point where the majority is not needed for sustenance of society. I am working as a scientist and getting paid for it although I do not contribute directly to keeping us alive. We also have reduced working hours compared to pre-industrial society - also signifying a lower need for using the population as labour to produce what society needs.

I think that, as a society, we should be happy about this and make sure to use it to provide sufficient sustenance and health care for everyone. Of course, you can add qualified jobs in other additional sectors (such as research), but since we have the capability to make sure that people have descent lives - why would you not do this?

Ok, fair enough. I was under the impression that the premise of the thread - like similar ones before it - was that this is an impending crisis (people unemployed, leading to increased poverty), not an impending opportunity. Perhaps @FallenApple can clarify.

 

[StatGuy2000]

Ok, fair enough. I was under the impression that the premise of the thread - like similar ones before it - was that this is an impending crisis (people unemployed, leading to increased poverty), not an impending opportunity. Perhaps @FallenApple can clarify.

Russ, I believe you have a myopic view of history if technological revolutions necessarily lead directly to a new era of prosperity and standard of living growth. At a certain timescale, yes, I do believe that the technological changes that we are experiencing will lead to that state (as has occurred in the past), but if you look at a cursory look at history in both various European countries as well as the US and Canada, the first Industrial Revolution led to masses of former farmers migrating to the cities (or emigrating to the US and other countries), and many of these people often lived in extreme poverty. Over time, these former farmers did ultimately improve their living standards over several generations, but one could argue that this arose primarily through the direct actions of governments (again, at least initially).

 

[FallenApple]

I can't help but wonder if there is a way to see this as more of an opportunity than as a change toward a dystopian future that will see half of the workforce unemployed.

A downturn in the need for "repetitive" type jobs may result in an upturn in other types of jobs that just aren't there because no one has the time to do that. Consider service industry jobs. And I don't mean just waitresses and baristas. There was a time in western society when many people were employed as butlers, maids, gardeners, etc. In the near future with the aging population there's going to be an increase in demand for geriatric care, for example, and I'm not sure that's something that robots are going to be able to do effectively. Even if they could effectively feed, toilet and bathe people, there's a simple caring element that needs to be there.

I remember a character in the movie Demolition Man - "Associate Bob." To the best of my recollection (and it's been over 20 years since I've seen the movie) he was basically a professional friend. In the real world this could manifest into things like increases in counselling, social convening, historical interpretation, debating, personal training, skill training, personal protection, etc.

A lot of this of course depends on an effectively managed trickle-down. With machines doing more work, the potential is that those who own the machines are going to make the most money and there won't be much for the rest of the population to share. You can't hire a butler or a tennis coach if you don't have an income in the first place.

Well, for jobs like bartenders or waiters/other service jobs, much of it serves a social function. I don't think someone sitting at a bar would want to talk to a mechanical arm. But many jobs would be eliminated outright, like data entry. Hopefully it will result in new jobs that are better. But the new jobs will either have a large social function, or a creative/analytical function. So I think these are the skills that people need to strive for, and quite urgently too.

How is this different from any other time in past history? Please give examples, from say, the industrial revolution, just for one.

You do realize that every time we progress technologically, we put people out of work. What about the book business? What has the internet done to that business? Now anyone can self publish online for virtually nothing.

How is it the same as the industrial revolution? In the past, repetitive low skill jobs are replaced by new repetitive low skill jobs. Shoe makers were able to move to the factory assembly line, which is arguably even more low skilled! At least with personalized shoemaking, there's elements of arts/ craftsmanship.

Here for the situation of increasing improvement in machine learning, the same will not happen; people will be forced to climb or fall, which makes the situation rather urgent.

Jobs get replaced and new ones get built, as always. The issue is will new jobs appear at the same rate as the old jobs that were lost? If not, then how quickly? Then, what types of jobs will surface? That is the more interesting question.

 

[Evo]

The issue is will new jobs appear at the same rate as the old jobs that were lost? If not, then how quickly? Then, what types of jobs will surface? That is the more interesting question.

And you're expecting someone here to know the answer?

 

[FallenApple]

And you're expecting someone here to know the answer?

It's just an interesting discussion on what might happen or what is likely to happen. I just want to see what people think of the situation. It's actually quite relevant for people's career path.

 

[FallenApple]

Here is a figure from the report.(page 28)

 

[russ_watters]

Russ, I believe you have a myopic view of history if technological revolutions necessarily lead directly to a new era of prosperity and standard of living growth. At a certain timescale, yes, I do believe that the technological changes that we are experiencing will lead to that state (as has occurred in the past), but if you look at a cursory look at history in both various European countries as well as the US and Canada, the first Industrial Revolution led to masses of former farmers migrating to the cities (or emigrating to the US and other countries), and many of these people often lived in extreme poverty. Over time, these former farmers did ultimately improve their living standards over several generations, but one could argue that this arose primarily through the direct actions of governments (again, at least initially).

It sounds like you are trying to disagree with me but ultimately not.

Yes, change is often painful, even change for the better. You didn't really respond to the question posed in my post, so I don't know which angle you are taking, but it doesn't seem to me that either angle (massive, systemic unemployment or not) would necessarily produce a result as "troubled" as the early industrial revolution. Everything about our society is more mature and able to deal with such issues (albeit naive and overconfident...).

The main reason robots don't scare me is that we've already had two robot revolutions(Three? Five? I guess I've lost count, but that's the point: they're ALL robot revolutions!) and the next one will be slower because this time the robots will be orders of magnitude more expensive, so they'll be much more difficult (slower) to implement.

 

[StatGuy2000]

It sounds like you are trying to disagree with me but ultimately not.

Yes, change is often painful, even change for the better. You didn't really respond to the question posed in my post, so I don't know which angle you are taking, but it doesn't seem to me that either angle (massive, systemic unemployment or not) would necessarily produce a result as "troubled" as the early industrial revolution. Everything about our society is more mature and able to deal with such issues (albeit naive and overconfident...).

The main reason robots don't scare me is that we've already had two robot revolutions(Three? Five? I guess I've lost count, but that's the point: they're ALL robot revolutions!) and the next one will be slower because this time the robots will be orders of magnitude more expensive, so they'll be much more difficult (slower) to implement.

I was specifically responding to your quote in post # 7:

"I think this next disruption will usher in a new era of prosperity and standard of living growth - just like all the others did."

My initial reaction upon reading your quote above (which you provided without further clarification or caveat) was that you were coming across to me as naive, with an unreasonably narrow view of history. Your subsequent, more nuanced reply sounds to me to be more realistic, at least with respect to what we can plausibly expect this next disruption will entail -- a period of disruption leading to initial turmoil in the labour market, followed by potential greater prosperity and higher standards of living (although I should note that I do have concerns of whether prosperity that could potentially be reaped from greater automation will be more equitably shared across populations).

 

[russ_watters]

I was specifically responding to your quote in post # 7:

"I think this next disruption will usher in a new era of prosperity and standard of living growth - just like all the others did."

My initial reaction upon reading your quote above (which you provided without further clarification or caveat) was that you were coming across to me as naive, with an unreasonably narrow view of history. Your subsequent, more nuanced reply sounds to me to be more realistic, at least with respect to what we can plausibly expect this next disruption will entail -- a period of disruption leading to initial turmoil in the labour market, followed by potential greater prosperity and higher standards of living (although I should note that I do have concerns of whether prosperity that could potentially be reaped from greater automation will be more equitably shared across populations).

Hmm...I thought this was a discussion, but evidently I'm talking to myself... but getting reviewed. Looks like I'm up from one star to three, so that's something!

 

[FallenApple]

Ok, fair enough. I was under the impression that the premise of the thread - like similar ones before it - was that this is an impending crisis (people unemployed, leading to increased poverty), not an impending opportunity. Perhaps @FallenApple can clarify.

It is not so much an impending doom thread so much as a new jobs thread. I'm most interested in what would happen once the dust settles.

However, one thing that is of concern is, what new jobs would those of lower skill move to? Other jobs with low skills? Those won't exist anymore.

 

[russ_watters]

It is not so much an impending doom thread so much as a new jobs thread. I'm most interested in what would happen once the dust settles.

Thanks, that clarifies.

However, one thing that is of concern is, what new jobs would those of lower skill move to? Other jobs with low skills? Those won't exist anymore.

I'm going to hedge and play devil's advocate a bit and say that I'm not sure it is that simple. In the past, CNC machining eliminated many mid-level skill jobs and spreadsheets and computer programs eliminated high skill jobs. Just because a job is low skill for people doesn't necessarily make it easier than a high skill job for a computer/robot to do. It may or may not be. In addition, high skill jobs tend to be highly specialized, so while a high-skill person may not fall back to minimum wage, they stand to drop a lot further than a low-skill person would if their job is eliminated.

And last, robots are expensive so it may be more difficult to displace low skill jobs with robots for cost reasons.

I also think there is a fundamental concept in economics at work that people fight hard against but is nevertheless real: Low skill jobs are not worthy of higher pay precisely because they are low skill. I know I'm crossing threads a bit with discussion on minimum wage, but my point is this: because low-skill jobs are easy and low-skill labor is cheap, it will *always* be possible for employers to find *something* for low-skill people to do, and pay them accordingly. I think how many jobs disappear for good is more a function of how we structure our economic system than it is about how technology competes with people.

 

[mfb]

The main reason robots don't scare me is that we've already had two robot revolutions(Three? Five? I guess I've lost count, but that's the point: they're ALL robot revolutions!) and the next one will be slower because this time the robots will be orders of magnitude more expensive, so they'll be much more difficult (slower) to implement.

What is expensive about distributing software?

As an example, self-driving cars and trucks won't be much more expensive than other vehicles but they save the driver - they will be worth the additional investment within months. Retrofitting the hardware in existing cars/trucks might cost a bit more, but compared to the yearly salary for drivers that is still cheap, and the vehicle can now drive 24/7. They will come rapidly as soon as the software and the legal frameworks are ready.
Every job that requires relatively unskilled work without mechanical components (this includes everything done on computers without special skills needed) can be automatized with just new software. Simple mechanical tasks can be done with quite cheap robots as well. All the "Office and Administrative Support" jobs, most of "Sales and Related" and a lot of "Service" in the study in post 1 are in this group.

The industrial revolution was slow because you had to build factories, and you have to build factories building tools to build factories first and so on. The AI revolution mainly needs computers as hardware - and we have them already.

No matter how intelligent the AI becomes, it still cannot have a Sherlock Holmes style of deduction, which means analyzing data for scientific purposes still should be around.

It is possible to do that - but it will take more time than many other tasks.
If AI can do science, it will be able to do nearly everything.

 

[256bits]

What is expensive about distributing software?

The acceptance of the concept, the changes necessary within the company, feasibility outlooks, infrastructure changes, ...
Something such as driverless trucks won't happen in a day, but like most things the penetration takes time.
Manpower work-hour shifting from driver personnel to dock personnel means that the cost savings won't be all that significant IMO, at least until the whole transportation system has adapted.
Who knows where it will go - amalgamations of fleets, larger automated hub centres, dedicated road systems for automated container transport systems,...
In the meantime, drivers, or some individual from the transport company would need to be available to sign off on the load from the dock, safety check of the vehicle, dangerous goods check with placards if need be, fuel fillup, hookup to the trailer, destination selection, etc. The first to select driverless vehicles would be TL transport, or full load from one location to the next ( ie a distribution centre ). LTL, or less than a full load pickup ( here a skid, there a skid ) would be needing a driver for some time to come, as the places these guys go were definitely designed on the back of a cigarette package 55 years ago ( hey, even some of the newer places seem to have the question of "how do we get stuff in here?" as an afterthought - a well designed loading/unloading dock takes a lot of space and a lot of money - utopia it sure ain't ).

So until everyone gets as rich as that fancy place in the Middle east with a D...with cash to spare, penetration is going to be difficult and drawn out, and we will suffer the congestion and sharing the road with every third vehicle being a truck and driver.

 

[russ_watters]

What is expensive about distributing software?

The licensing agreement?

But seriously, I think you and I are on the same page about my point that they are all robot revolutions, because that question implies that what is coming in the future is mostly just smaller, evolutionary improvements in hardware and software and isn't fundamentally different from what has come before...but just might be faster. The OP of this thread is pretty well written (open) and the clarification helped, but I think based on other similar threads, more people are concerned about a sudden, cataclysmic change enabled by general purpose robots that could instantly render thousands of job descriptions obsolete (and the OP did specify he's worried about the pace). Think like the movie iRobot, where suddenly, everyone gets a new humaoid robot that can do basically anything we can. That is what I think will be fabulously expensive and it isn't something I consider realistic from either a technical or financial angle. The idea of McDonalds replacing its order-takers with iPads just doesn't scare me. It just isn't a big deal (and plenty of places have already done it). That's an example of an easy change and it is not happening quickly - at least not as quick as I think it should/could.

So again, the reason I don't have the same fears as others is that I think the next 50 years holds gradual and continuous advancements in software/robotics/automation that will over a long span of time amount to a major change in our working/living landscape...just like the last 50 years have.

For the specifics:

I don't share your optimism for the prospects of driverless cars/trucks because I don't think it will be that easy/cheap to retrofit existing vehicles and if it is easy and cheap, then it's too cheap and companies will simplly charge more money for it due to supply and demand.

For the "Office and Administrative Support" jobs, that's already happened, so I don't see any reason to fear it happening again. One small example is just how fast email replaced letter writing. I'm not sure if companies used email much before the WWW went big in 1995, but by the time I got out of the navy and got my first real job in 2002, letter writing was fully dead and replaced with email.

A similar example from my industry (building design/engineering) is that 30 years ago, probably 3/4 of the jobs were "designers" or "draftsmen" and all of the draftsmen and almost all of the designers are gone. Overall, that equates to probably 2/3 of the jobs in my industry disappeared in 30 years...and that's not even including the loss of administrative support over the same time period.

There are a lot of individual examples, of course. Will the future pace be faster and more disruptive? Maybe, I but I doubt it. I think people discount problems that happened but didn't turn out to be problems and then don't apply that moving forward (Y2K is forgotten because it didn't cause disruptions, but it really was a big problem). But these are just guesses -- as are everyone else's.

What annoys me about these studies though is that they are ultimately meaningless because they don't compare the predicted future to the known past. Maybe it is the fault of people who write news articles about these studies and not the people who write the studies (because studies by nature tend to be narrowly focused), but the study linked in the OP is so narrowly focused that it is isn't helpful for discussing the topic of the thread. Personally, I think it's mostly that reporters don't like context because context makes the terrible boring and terrible gets more clicks... though the author of the study does spend 15 pages on pointless historical anecdotes, without ever discussing the historical context of the title statistic of the paper. So maybe it's the study's fault in this case.

What I want to see isn't "47% of jobs are at risk". That's meaningless on its own (what is "likely"? When?), much less without context! I want to see:

"25% of jobs will likely (better than 50% chance) be replaced by automation in the next 20 years. If correct, this stands to be the worst of any time period studied in the last 50 years. The next worst were 22% of jobs in 1980 were lost by 2000, 18% of jobs in 1960 and 12% of jobs in 1990."
[made-up numbers of course]

[Edit] A quick Google tells me there are 3.5 million truck drivers in the US and the US lost 5.6 million manufacturing jobs primarily due to automation from 2000 to 2010. So we've seen worse than what driverless trucks can bring - and recently.

This doesn't go far enough back for what I really want, but another Google tells me we lost 900,000 admin support jobs in only 2 years, from 2009-2011.

 

[Stephen Tashi]

The point is that humanity is in a situation where a major part of the population is not needed in order for society to sustain itself.

If I think of society as a single semi-rational organism, I can agree with that observation. However, if we consider humanity to be more than the humanity of the developed nations, is it true?

Differences in geography result in differences in natural resources, but I think differences in politics and culture are the major reasons why nations differ so greatly in material prosperity. What it means for a society to "sustain itself" is a cultural question. Thinking of small groups of individuals, do they need actors, television commentators, rap musicians etc. to sustain themselves? - probably not. However, an entire society creates a demand for such things.

 

[mfb]

The acceptance of the concept, the changes necessary within the company, feasibility outlooks, infrastructure changes, ...
Something such as driverless trucks won't happen in a day, but like most things the penetration takes time.

It won't happen within a day, but it will happen within a few years. Significantly shorter than a typical working life. The approach "we don't hire new drivers and the old drivers will retire eventually" won't work.

Sure, you keep the loading/unloading business, but how much time does that need compared to driving?

I don't share your optimism for the prospects of driverless cars/trucks because I don't think it will be that easy/cheap to retrofit existing vehicles and if it is easy and cheap, then it's too cheap and companies will simplly charge more money for it due to supply and demand.

Unless you suggest the industry will do price rigging, I don't see why it would be expensive. The hardware costs a few hundred dollars, add generously a few thousands for the installation and the software. A vehicle that can drive 24/7 (apart from loading/unloading) can probably replace at least 3 drivers. Even if self-driving would cost $100,000, it would be a great return of investment.

Driverless trucks and taxis are just one example, many more will come at the same time.
Precise numbers? Predictions are hard, especially about the future. But it looks like it will happen more rapidly than in the past. And it will happen more complete - with most low-skill jobs being replaced.

 

[russ_watters]

Unless you suggest the industry will do price rigging, I don't see why it would be expensive. The hardware costs a few hundred dollars, add generously a few thousands for the installation and the software. A vehicle that can drive 24/7 (apart from loading/unloading) can probably replace at least 3 drivers. Even if self-driving would cost $100,000, it would be a great return of investment.

No, I mean supply and demand. The need to retrofit 3 million trucks is a large demand, requiring a large industry that doesn't exist yet to make it happen. Large demand plus small supply equals high price.

The history of technological adoption shows it always works the same way (DVD, solar panels, electric cars, pcs, whatever): someone invents something new and great and starts selling it. It doesn't matter how cheap it is to make, they are the only game in town, have limited capacity, and have to recoup their borrowed R&D cost to avoid bankruptcy. So they sell it expensive. Meanwhile, there is no support infrastructure, so there is limited benefit to the new technology, high adoption cost beyond the product itself and high risk. So people are slow to adopt it. Gradually these things work themselves out and the price goes down while the adoption rate goes up. But it takes 5 years at least, just for the ramp-up to relevancy. Then another 5 at least for a meaningful adoption rate.

The "meaningful adoption rate" is getting faster, but the "ramp up to relevancy" is not likely to keep getting faster:

https://hbr.org/2013/11/the-pace-of-technology-adoption-is-speeding-up

If, for example, they follow the tablet rate, it would take 5 years to reach 10% and another 5 to reach 40%.

 

[mfb]

All your examples are hardware-based, where you have to manufacture something new to sell it to millions of new customers. The internet infrastructure had to be built up as well. And you still see some acceleration trend in it. What self-driving cars need are mainly cameras and computers. The camera and computer industries exist already. And you have a smaller number of customers ordering a much larger volume. Car producers and taxi/truck companies.

Let's take examples that didn't need hardware (at least for the customers). Google as search engine went from "completely unknown" to the dominant search engine in just two years. The Internet explorer replaced Netscape within three years, and Chrome needed just three years from a completely irrelevant browser to the most used browser. Facebook needed three years from an unknown small website to the dominant social media platform. All these examples would appear as nearly vertical lines in your plot. And all these examples don't even have a strong monetary incentive for the customers.

 

[Mondayman]

The idea of McDonalds replacing its order-takers with iPads just doesn't scare me.

This doesn't scare me either, considering how low skill the order-takers are at my local McDonalds.

 

[Orodruin]

Google as search engine ... Facebook ... And all these examples don't even have a strong monetary incentive for the customers.

The "customers" are the advertisement companies. I would say they have a pretty strong incentive. The users' private information is the product.

 

[russ_watters]

All your examples are hardware-based, where you have to manufacture something new to sell it to millions of new customers. The internet infrastructure had to be built up as well. And you still see some acceleration trend in it. What self-driving cars need are mainly cameras and computers. The camera and computer industries exist already.

But not *in* the trucks! Backing up:

I think the article/graph I posted is highly applicable, though the author didn't delve into why the speed-up is happening with tech devices. It's happening for two related reasons:
1. These new devices are not really new. They are evolutionary steps and more importantly convergences of other products and technology. The iPad is just a big iPhone and the iPhone is a convergence of PDA technology developed in the early 1990s, with cell phones...and PDAs were stripped-down, hand held Mac computers.
2. Because they are convergences (plus service based with the phone), they are cheap.

That is what autonomous cars represent and you are correct that the hardware is relatively off-the-shelf, which will make it follow the trend of fast adoption of other such devices. But you're essentially pretending the hardware already exists in the users posession, just waiting for the software to be installed. And it just isn't true. Retrofits are going to be hand-installed, which makes them expensive and time consuming. It's much harder than, for example, upgrading from Windows 3.1 to Windows 95, and very few people did even that: they mostly phased-it in as they bought new computers and got rid of the old ones. But maybe now you are suggesting that?:

And you have a smaller number of customers ordering a much larger volume. Car producers and taxi/truck companies.

You mean you expect car/truck manufacturers implementing them in new vehicles to be the primary adoption "vehicle"? Not retrofits? I agree that's' likely (because retrofitting an old car with the mechanical devices needed to drive it is a big and custom task). But that makes the implementation timeline even longer because the *average* age of a commercial truck on the road today is 17 years! Yes, they can ramp up and replace faster, but then that just makes it even more expensive, losing the value of the truck you are discarding (which often has a longer term loan attached to it).

Let's take examples that didn't need hardware (at least for the customers). Google as search engine went from "completely unknown" to the dominant search engine in just two years. The Internet explorer replaced Netscape within three years, and Chrome needed just three years from a completely irrelevant browser to the most used browser. Facebook needed three years from an unknown small website to the dominant social media platform. All these examples would appear as nearly vertical lines in your plot. And all these examples don't even have a strong monetary incentive for the customers.

I agree with @Orodruin and I don't even know if Musk and Zuckerberg knew what they were really doing from a business standpoint when they created these things. A great many of the computer/internet pioneers abhor capitalism and business interests (which is a fascinating human study, but a different topic); They are virus spreaders and it is only recently that the business plan got attached to it and then named: It's viral marketing and market research. Due to the late date, Zuckerberg probably should have known, if he didn't. When he created facebook, he was a paid, freelance software developer. He'd created viral social fads before, and Google was already established, so I suspect he recognized the real business potential.

If facebook users were actually customers and facebook had cost its early users a lot of money (the penalty for early adoption), it would not have succeeded. It has a very low value to its users and its original "customers" would have been poor college students. Facebook, the website, is not a "product", it is a social fad and a platform for marketing and data collection, which are the real products.

The software for autonomous cars also isn't easy to make and so the development is already fabulously expensive, proprietary, and taking forever (albeit subsidized by other businesses). But there is a way around the difficult software problem: more extensive infrastructure until the software catches up. And that's already happening. The most successful self-driving vehicle company isn't Tesla, it's Amazon, with their forklift-driver-less warehouses. These are smaller, closed, controlled "tracks" for the vehicles. Next-up: shipyards. After that, cities. Then, closed routes for trucking. In that sense, we're already on a growth curve for driverless "trucks", but it is slow and evolutionary, as the technology develops and matures.

Regardless, we all see this coming (because it is already in progress), even if we disagree on how/how fast it will develop. Hopefully, truckers and kids deciding what careers to pursue see it coming and plan accordingly*. Did Blockbuster see Netflix coming? That's a truly new and unique - yet simple - business model that fell out of the sky, fully developed, and blew-up a mature industry in a very short time. Netflix went active in 1998 and Blockbuster went bankrupt - gone - in 2010. Driverless trucks are not Netflix. It's already happening in a pre-mature way, and happening much slower, which means it will be much less disruptive because we all see it coming and can plan for it.

*The great prognosticator Matt Groening, who predicted President Trump, produced Simpsons Episode 220: "Maximum Homerdrive", about driverless trucks, in 1999.

 

[Stephen Tashi]

A forseeable political clash involving autonomous vehicles will be city, state, and federal highway departments wanting to take control of autonomous vehicles in order to efficiently route traffic. Eventually cars and trucks will be required to provide such access.

 

[PhotonSSBM]

Nobody seems to have given much discussion to what should be done if low skill jobs see a dramatic downturn. Russ and others have said some things, but nothing fleshed out as to what to do with the unemployed in the worst case scenario. I'd be curious to hear what others think about a situation where people will need a welfare state to provide basic needs to tens of millions more people than before. Medicaid, food stamps, housing will all have to see a huge increase in funding if the worst happens. That seems to be the popular mentality, but I wonder if anyone here feels there's a different/better solution.

 

[Choppy]

Nobody seems to have given much discussion to what should be done if low skill jobs see a dramatic downturn. Russ and others have said some things, but nothing fleshed out as to what to do with the unemployed in the worst case scenario. I'd be curious to hear what others think about a situation where people will need a welfare state to provide basic needs to tens of millions more people than before. Medicaid, food stamps, housing will all have to see a huge increase in funding if the worst happens. That seems to be the popular mentality, but I wonder if anyone here feels there's a different/better solution.

In principle there is a very simple solution: tax new revenues generated by the automation of industry and use that to provide a guaranteed minimal income.

In practice, there will be a lot of very practical problems associated with such a highly complex task, starting with whether it should be attempted in the first place. There is a strong argument for doing nothing. People on their own are industrious - or at least they can be - and may not need this problem to be solved for them at all. Sure jobs will disappear in particular sectors, but jobs will open up in others.

 

[PhotonSSBM]

In principle there is a very simple solution: tax new revenues generated by the automation of industry and use that to provide a guaranteed minimal income.

In practice, there will be a lot of very practical problems associated with such a highly complex task, starting with whether it should be attempted in the first place. There is a strong argument for doing nothing. People on their own are industrious - or at least they can be - and may not need this problem to be solved for them at all. Sure jobs will disappear in particular sectors, but jobs will open up in others.

This is my opinion as well. I am very interested in people like Russ' views who oppose most things relating to welfare and how, if the situation truly arose as predicted, they would help (or not help) the people without jobs.