2017-11 Book Club Transcript

Hello and welcome to the November edition of my book club. This month we read Average Is Over by Tyler Cowen the renowned GMU economist, blogger at the incredibly popular blog, Marginal Revolution, as well as an all round polymath, autodidact, incredibly well read and deeply thought individual.

This book’s main idea is that average — the middle achieving outcomes that people can expect in life, are going down and they’re going to continue to go down in the future — the most obvious implication of that trend is in work.

The future is going to see a lot less jobs that earn a middle income and a lot more situations where you have a few very high income earners and a lot of people earning a lot less income. Interestingly enough Tyler is arguing here that this isn’t because simply the rich are getting richer and that it’s an exaggeration of existing trends towards inequality, but rather he blames it on three trends he thinks are going to change the fundamental nature of work and how we do it.

The first is automation. Everyone has probably heard by now, it seems like every day there’s a new opinion piece, about how the robots are going to take our jobs and it’s going to lead to a terminator situation or a utopia depending on who you listen to. Now Tyler Cowen is a lot more reserved when he makes this claim. He’s not really arguing that the robots are going to take over but rather that increased automation is already taking place and it’s going to make a difference in how we work and what the expectations are going to be.

The next is outsourcing and globalization. The idea that China and India have recently become huge players in the economic market but also many countries in Africa and Latin America that are still growing and going to be also playing a big role in the future. Finally there’s an interesting idea which doesn’t get as much press but I think it’s an important part of this story. Tyler includes it in his book which I’m going to call “clustering” and that is a phenomenon in terms of groups or regions or companies showing again that average is over.

Some cities are really prosperous while others fail and you have some companies that are successful and everyone earns a lot of money and they have a lot of profit. I think these three trends are all important to look at if we want to understand what is your career going to look like in the future? How should you be planning your own personal development so that you can take advantage of these trends rather than have them take advantage of you?

Let’s discuss automation.

This is a popular topic, it seems like every day you see a new opinion piece about robots taking all of our jobs, we need to stop the robots, we need to accelerate the coming of the robots… really I think a lot of these pieces exaggerate the facts. Personally I think that although the advancements in deep learning and neural nets and these kind of advanced machine learning algorithms are impressive, I don’t think we’re on an imminent path that there’s going to be some AI that can do all the jobs that humans can do.

It’s probably going to be something like in a century or more — although that estimate has a lot of ambiguity to it — however, what we really need to think about in this generation is, how do we adapt in the meantime? What do we do with the fact that automation in tech has already happened?

It’s already the case that I don’t use a travel agent to book my plane tickets. I go on kayak.com or flighthub.com and I find the cheapest flight through a very complicated algorithm that searches all the airlines and finds the lowest prices. I already don’t go to a library and ask them to find me reference materials. I use Google Scholar or even a plain Google search to locate the things I need.

This is an interesting conundrum we have. We have some other technologies that haven’t quite arrived yet but are on the cusp so we can imagine things like driverless cars and that’s going to transform quite a few industries. Advancements in deep learning and machine learning are going to make tasks that previously required an element of human use more obsolete. So how do we deal with this mixed situation where human beings do some of the work and machines do some of the work, and then humans plus machines, together, do some of the work? How do we plan around that? I think Tyler Cowen has the right prescription here. He really maps out a transition path of four stages that he imagines that we’re going to go through. This isn’t something we go through collectively meaning that we’re entirely at one stage and then we’re entirely at another stage.

Rather, individual industries, individual job positions are going to go through these four stages and it may take a short amount of time, it may have already happened for some (I mentioned the travel agent analogy) for a large section of the population, travel agents are already obsolete. But there’s also probably jobs right now that are almost exclusively done by human beings. Let’s say a nurse or a doctor; there’s a little bit of machine assistance but these are by and large still being done by a person rather than a diagnostic robot.

The first of the four stages is “man only” where there’s a human being only who is working on this. The next is the “human + machine” where the human is doing the bulk of the work but is using the software or using tools to facilitate the work. The next is a “machine + human” combo where the machine is doing most of the work and the human being is more there to monitor and make sure it doesn’t go wrong, to fix common errors, etc. Finally, we have the machine only where now, there’s isn’t really much room for the human being to improve upon the machine’s results. The machine itself gives the best results possible.

And so Tyler Cowen in explaining these trends brings up the example of chess. In particular he talks about something called freestyle chess. That is where unlike most chess tournaments where you have your brain and basically nothing else to work out chess problems and compete against your opponent, in freestyle chess, you’re encouraged to use chess software, use chess programs, to compete with your fellow competitors. In this case it’s not the person who is the best at chess who wins but the person who is best at using chess software. Although chess might seem like a rather odd example to use when we’re talking about the economy, because it’s a small and insignificant faction of most country’s economies, chess is a useful example because it’s a fairly clean situation where we can see these four stages of transition.

So, until very recently chess was man only until the situation was that the best chess players in the world could always be a machine. Over a short period of time, the machines got better and better. This wasn’t really so much of an innovation in terms of the machines becoming smarter in the sense that they have more sophisticated algorithms (although that did play a small role) but rather it was mostly a matter of computing powerfully. Suddenly they got fast enough that now they’re actually better. The first step was this pre-deep blue era where we have man only where chess is the only thing he can do and indeed you have lots of pundits, lots of people saying that chess is the pinnacle of human existence and human creativity.

Now it turns out that this wasn’t the case. You can have an algorithm that is far away from the general AI winning at chess but this wasn’t so obvious at the time. Next you have this man plus machine. This is an area where excellent players were still making their own decisions but they were often using computers to guide their analysis so if they wanted to explore if they might use a chess engine which could enumerate through the possibilities.

Next you see the machine plus human being teams which are basically where we have freestyle chess and it gets a little bit of added help from the human judgement. The human can pick different chess engines and understand how they work, where they’re going to be better or worse and use that to combine them together.

Now, this book was written a little while ago but I feel like he is already signaling (maybe those familiar with chess can determine whether this has happened) but we might be reaching a machine only era of chess. Where the correct chess moves are so difficult to unravel that it’s almost impossible for a human being to actually improve upon them.

So it’s no longer the case where you’d use a chess engine but you’d have enough intuitive judgement to know where it’s going to falter. Rather it’s a situation where the chess engine is going to know the right move and you shouldn’t question it.

This is something we can imagine in our situation with automobiles although we’ve yet to see the full transition. Right now cars are essentially man only. Human beings alone are driving the cars. However, there’s a lot of cars on the market right now which have a little bit of man plus machine. You can see some models of Tesla, some other companies too, have these driverless car features where you let the car do a little bit of auto-pilot while you’re driving. Some of these cars are still in prototype phase but we can see that kind of man plus machine hybrid where the human being is still required to handle difficult situations (eg where the sensors don’t work very well) now it’s hard to say how long this phase will last.

It may be that the companies, because of fears of regulatory worries or safety, might want to keep working and wait until they really reach the era where they’re very good and very safe before they release them to the public. But we can also imagine a situation where your cruise control can follow GPS but you still have be behind the wheel and you have to be able to take over if needed.

Next we can think of a situation where it’s mostly machine and a little bit of human control. Imagine driverless cars where the car drives itself but it’s a bit more fragile because it has certain situations where it will break down and it lets the human know because of sensors aren’t working, or a glitch or a software malfunction, and so on.

Finally you can imagine, and this is what many people are taking about, is the machine only phase where the driverless cars are so good that it’s actually irresponsible for a human being to be operating them. They can break faster, detect dangers quicker, avoid collisions with reflexes that human beings simply can’t match. It may be the case that all new cars are required to be this version of car or that old cars have to be retrofitted. This is the machine only phase.

What I think is often missed in this discussion which Tyler Cowen goes to great pains to clarify is that there is these transitionary phases. There’s phases in the middle where you have both humans and machines and they are working together in complicated ways that is sometimes missed by this grand picture of “the computers are going to take all of our jobs”.

What does this mean? For many industries and many types of jobs, these middle two phases where you see majority human being or majority machines but there’s still a team between these two elements, is the type of work is going to change.

For many of these things this might last even over a century or more where we have this gap where there’s something that machines can do, something that the humans can do, and they work better together but there’s no dominant approach where completely human being or completely machine is the better approach.

Some ways, this is already the world we live in. This is a world where we’re all using software all the time to do our jobs. Many, many opportunities are coming because the person needs to understand how to use the software. But I think that with a lot of innovations in machine learning and pattern recognition that is only going to continue and we’re going to have more and more sophisticated software programs that are going to require more and more sophisticated people to employ them.

Now one of the mistakes that Tyler Cowen notes that many people make is that they make the erroneous assumption that because these innovations are coming from engineering, science, maths, technology, and so on, there’s many people who think well maybe it’s the case that these are going to be all STEM jobs so the only thing we can do is teach everyone to become a programmer or an engineer and what the author suggests here is that that’s not the case.

A lot of the professions that are going to come up are not going to be technical jobs. Instead it’s going to be using technology to do some task that is not necessarily technical. What this means that it’s important to understand the technology and be sophisticated in its use, but not necessarily that you have be the one making the technology.

So a really good example is the work that I do on this blog, writing and speaking on this podcast right now. This entire profession that I’m doing is new and enabled by technology but what I’m actually doing is much more in the liberal arts direction as an actual profession. I’m not doing very much programming however I do have to understand how the software works that runs my blog and the search engines that will index it and send me traffic and lots and lots of technical things but the ultimate task of my job is not technical in nature.

Tyler Cowen thinks that a lot of the jobs that he thinks that people are underestimating how important they’re going to be are things that involve people. These are the things that currently the machines have a really difficult time doing. Things like marketing, one on one personal interactions, coaching, teaching, these are all professions that are likely to preserve in some role but they’re going to be hybridized with machines.

We can see that with people who are doing online marketing or online advertising that it is a new field of advertising where it’s not enough to create a catchy slogan and have a nice design you also have to understand how the internet works and how your ads are going to show. There’s a real hybridization of tech skills and your core skill which is marketing.

You could also see this in professions like teaching. So right now a lot of teaching is still in the human only phase but with the rise of massively open online courses and platforms like Khan Academy it’s possible to imagine that there are some functions of teaching that will be replaced by technology.

So you can imagine that, instead of having the teacher deliver the lecture you have the best teacher in the world delivering the lecture in some pre-recorded studio in ideal conditions with the perfect public speaking voice. And then what the teacher is doing is conveying the material and the content but rather what they are doing is helping the student so they’re going around and making sure they understand and facilitating the learning process.

These are just two professions that are often said, you know, teaching is definitely something that is claimed to be, you know it’s going to go out of business once the machines take over but I think what we’re really seeing is that it’s going to change the nature of what it means to be a teacher. Same with marketing.

Certainly for me, what it means to be a writer and journalist has changed dramatically and I think this is something we can expect to continue in the future. If we understand this road map, these four transitionary periods, and showing how doing some task that may or may not be technical with a machine compliment is going to be the standard for work in the future. This can give us ideas about how to plan our own career.

Before I go into the steps I think you should take to apply these ideas to yourself, I think we should move on to another trend. This is outsourcing. Basically a lot of this accounts for China and India becoming fairly low income countries to now middle or upper middle income countries.

In many ways China and India’s rise is a huge change in how the economy works, the centre of gravity of the economy if you will, has shifted away from the United States. This has some downsides for people in Westernized countries. I think some people fear that outsourcing or free trade is going to destroy North American or Western European jobs but we also have to think of this as a positive.

Personally I think that outsourcing is mostly a positive story. Although the automation story is something that has more debate but I think that the obvious plus side is that this has been good for the Chinese and the Indians and the people who have been lifted up from poverty in this situation.

There’s something spiteful in a lot of punditry that says that billions of people coming out of poverty is anything but a good thing just because it happens to negatively impact a few people who were, to be honest, were relatively rich to be begin with.

There’s certainly some selfishness and myopia in viewing these trends. I also think that even if we are just analyzing it — now I’m Canadian, I live in North America — but this is a benefit for the world because China and India becoming much larger even if it’s been eroding some middle class jobs (industrial jobs and manufacturing) and a lot of simple tasks in accounting and customer service have moved offshore. But what has also happened in the explosion of these economies is that we’re seeing a much larger market.

So, really, this also a benefit and a cost. Like automation, this is where we’re seeing this average is over because the people who were doing the jobs that were protected by the fact that they had to be done in North America and they were under that kind of regime where they couldn’t be outsourced, these were many times the people who were losing out.

However the people who are more robustly protected against this kind of thing are the people who can market larger products. So in many ways, Silicone Valley and the tech world has been one of the benefactors of this trend because these are globalized products (even China still buys iPhones and Apple products) and it has allowed a lot of economic growth domestically.

What we’re really seeing, to repeat the title of the book, is that Average Is Over. There’s a few people perhaps concentrated in these types of industries that can benefit from a large globalized market place and they’re less hurt by the fact that there’s greater competition. On the other hand you also have a story which I haven’t focused on as much which is people at the lower income spread; they are also less effected by this because the people who are in customer service jobs at McDonald’s or people who are doing retail jobs, these are less likely to be outsourced because they are local and require face to face contact. They are also less likely to be automated because they are the same kinds of jobs that are harder to do with a machine.

There’s one more trend I want to discuss just very briefly not because I think it’s the most interesting but it’s something I haven’t heard discussed a lot. Often you’ll hear stories about outsourcing or automation because they tie into easily visualizable narratives about robots and foreigners, but “clustering” I think, is also an interesting story.

This is the idea that it’s not simply between individuals that we’re seeing this “average is over” but through network effects, we’re seeing little clusters where there’s going to be a clump of people or institutions that do really well and then those that do poorly. One way you can think about this is in regions. At the city or state level, there’s going to be some cities which are incredibly prosperous and there’s going to be many regions that are almost economically barren. They don’t produce that much and they struggle economically. I think this is a lot of the political story of what’s happening in North America right now.

You can see the change in the electoral dynamics that elected Donald Trump for instance. You see blue states which are these concentrated urban markets which have benefitted from globalization and you have poor more economically disadvantaged interior and rural regions which have suffered so again, this is “average is over” and unfortunately if we are to believe Tyler Cowen, this is only going to get worse.

So that’s one element of this clustering is that certain cities like San Francisco and New York are just dominating and then other areas, not so much. And we also see this in firms themselves so within different companies. I don’t have the statistic at hand but I read an economic analysis that was suggesting a large portion in the income inequality (not wealth inequality) was due to differences between firms rather than within a firm itself.

So within a firm itself the picture of inequality is often the one that gets the most press. This is one where you have greedy CEO’s earning 500 times their base level employees. I don’t deny that that’s the case. But what was interesting about this report is that perhaps an even bigger problem is the “between the firms” picture. So the best, highest earning firms, all of their employees do better and the people in the lower firms, all do a lot worse, perhaps with some exceptions for the top employees.

So what this means is that this might create this superstar firm effect where you want to work at Google or Apple and you don’t want to work for Sears or some other struggling company and this destiny to go in the right workplace environment or work for the right company is going to be increasingly important as well.

To recap, there’s three trends that all spell this “average is over” and the average is the middle income, middle class and their outcomes are all depressing. The lower class and the upper class are somewhat more resilient to this. I focus more on the upper class because in many cases the middle class is going to shift to the lower class as a result of these changes.

There’s three trends: automation, the fact that machines and technology are more prevalent and that is creating displacement in terms of jobs being replaced but more importantly it’s changing the type of work we’re doing. It’s not enough to be able to do a menial task, you need to be able to assist a computer to do such a task. This requires a higher level of skill and perhaps a college level education or higher.

Next you have outsourcing which is really in a global term, the opposition of “average is over” it’s lowering of income inequality as many large poor countries are becoming middle income countries. But for the Westernized and developed countries this presents a problem because many of the easy to outsource tasks are going over there.

Finally this clustering effect where even within countries and industries you’re having a small group and clusters of people capturing a lot of the benefits because of networking effects and they all work together which gives you synergy and connectivity and that’s why you have superstar companies like Google and Amazon which have just been growing wildly because they can benefit from these in house resources.

So what should you do about this? I think there’s a risk of describing the future and endorsing it. I don’t want to do that. I think that many of these trends are negative for a lot of people, maybe even you, but I think that they are definitely a change for a lot of people and in many cases change is a bad thing because change creates more difficulties. At the same time I don’t want to be overly pessimistic. All of these things are coming from a place of driving the economy to a large size. So automation, well yes, it’s going to take jobs, but it also means those are jobs that people don’t have to do any more.

If you were considering an economy of one person in your household, having a dishwasher is better than washing the dishes yourself even though to a certain extent the dishwasher has “unemployed” you from washing dishes. Obviously this displacement or this replacement of jobs is a shift towards a more capital intensive environment where you don’t need as many people to perform it has some negative outcomes particularly for the people who get out of the game where they can now earn a normal living but it has the side effect that it’s increasing the economy as a whole so what this means for human beings has to do with what kind of choices we’re going to make as a society.

The next outsourcing I think is mostly a positive story and the few people who see it as a large negative I think are overly focused on their narrow selfish concerns of where they are and they’re forgetting the fact that China and India becoming richer is good for Chinese and Indian people and it probably has good impacts on average North Americans.

Clustering is somewhat worrisome because it does possibly stem from monopoly powers so you have big companies that are controlling large swarths of the economy and it’s really difficult to compete against them. That’s something that’s worrisome. However, I think that most people despite some of the journalists attacking Silicone Valley, most people have a positive impression of this sector where the clustering is happening most.

So what should you actually do about this? I think there’s a couple steps that one can take although of course these are large macro-economic trends so it’s difficult to say with certainty what is the correct move. But the first thing you should do is develop skills in yourself that are going to resist automation or resist outsourcing. And this is something that I find interesting because many people when they think of automation them immediately go to STEM fields. They think, “oh I should be a programmer or engineers” and this isn’t to say programmers or engineers aren’t great professions. In some areas like San Francisco, there’s such a shortage of them that they can earn incredibly high salaries. But it’s also not necessarily true that these are immune to outsourcing or automation. It may be the case that in the future most of the engineering jobs are in China or that the programming work in done in India so we can’t really be banking just because it’s visually similar to automation that we foresee coming in the future that these are jobs that are going to win.

I think the author gives a strong argument for why it’s going to be the jobs that involve people skills and that leveraging a machine a plus people skills are going to be very powerful. So for instance marketing, coaching, being the kind of person who is using the data and the information but is communicating it, conveying it, packaging it, and working with other people.

I also believe that one of the ways you can avoid this automation trap is simply by developing deeper skills. So what we’re seeing the machines take over is in the case of something that would be very fast for a human to learn. Or they are not very sophisticated in terms of complexity whereas the more you go up the skill hierarchy the more ambiguous the skill application becomes, the more complex it is, the hard it is to outsource.

One example I heard was that accounting was a profession that was going to go out of business because in the future no one is going to get an accountant, you’ll just have a machine that does your accounting for you. This represents a bit of a misunderstanding of what accountants actually do, because a book keeper, this is a profession that has been largely over taken by many simple softwares so that we’re now in the machine plus human category for book keeping where there’s complicated accounting software that does everything and then the human being just has to double check that they’re doing the right thing and maybe handle a few smaller tasks. You don’t need armies of people doing book keeping in an actual business ledger.

On the other hand, accountants are very often the people who are making strategic decisions on the basis of money. They are going to be deciding how is money flowing in a company, what’s good, what’s bad, how to diagnose these issues and these are very context sensitive, they all require communicating with people. They all require a deep level of intuition.

So it’s certainly possible that in the far future accountants will be replaced by machines but my own personal belief that this is a profession that will be resistant to it at least for a while. Now why did I bring up accounting? Because you can see accounting as sort of being on a spectrum of skill within that profession. So book keeping on the sort of lower end of accounting skill and then you have various types of accountants to high level advisory roles and the most deep, most insight-based type of accounting.

So what you really want to be doing in your profession is pushing up this scale, because the further you push up this scale, the deeper your skill and competence is, the more you’re going to resist erosion from this middle vacuum of white collar easy to do jobs that will be taken over by machines or outsourced to other countries.

The second piece of advice I would offer is learn to work with the algorithms instead of against them so you want to position your career so you are able to benefits from these trends by being something that you can add value to the technological landscape. This is something that is difficult to talk about in general — it’s going to matter and manifest a lot more in the specifics — but I think that deep understanding of how technology works and how to apply it is going to be incredibly valuable.

I am personally of the belief that deep learning and machine learning is a bit over-hyped right now because everyone is saying that these few algorithms are going to take over everything and I personally think that they’re probably going to handle some specific cases but there’s a lot of cases we can’t still solve with these particular algorithms. However, let’s say that we’re just thinking about it from that lense, I think understand machine learning is very important because if it happens to impinge on your work then you’re going to want to have a deep understanding of how are these algorithms actually doing what they’re doing so that you can figure out where are they likely to go wrong and where they’re likely to get a good and bad result so you can use your human intuition to operate them efficiently.

A good example of this is learning how to use Google properly. I know lots of people that really struggle to use Google but if you’re very good at it, you know what kinds of keywords are going to bring up the results that you want. Where Google will understand what you mean and where it might not. You’ll also understand how to use some of the different syntactical constructions such as double quoting around things, you can search for more exacting phrases or using the minus sign to remove certain results that are crowding it out, but using Google is a basic skill but I think it’s an example of this broader trend that to do your job properly is going to have a technical competence so you’re able to augment what machines and what automated systems are able to do, and apply it to yourself.

There’s a certain sense that this applies to outsourcing as well. I think it’s a lot harder for the average employee to take advantage of outsourcing but you definitely see this in books like The Four Hour Work Week by Tim Ferriss where he advocates these highly outsourced businesses where you only have to work four hours a week and you’re still earning a good income because you have other people in lower income countries doing the work for you for whatever is their decent amount of salary. I think this outsourcing approach is something that tends to apply more to business owners but definitely knowing how to delegate work and allocate work more efficiently in a global sense is also something that potentially has advantages although I think they’re more limited in applicability to the average person than learning to work with technology.

Finally, and this is not going to come as a surprise to those of you who listen to me all the time, but my last piece of advice would be to master the learning process itself. If anything this trend means that the pressure to learn, the pressure to adapt and change environments quickly is going to go up and up and up. Already in my lifetime I’ve seen transitions where things used to be completely human and now they’re completely machine and I haven’t been alive for that long so there’s a good chance that in your span of your career you’re going to face that possibility or that transition in multiple areas of your work. And so if you are good at the learning process, if you know how to quickly pick up new skills by organizing projects and your time to quickly advance yourself, that’s going to put you ahead. And the people who are rigid and inflexible, who think I’m not good at learning, these are going to be the people who get left behind when there’s a gap where you could still do the machine plus human being and then it transitions to a point where if you don’t know a machine or how to use the software, you’re completely out of luck.

So this book does have a pessimistic tone and I think there’s definitely some challenges to be faced in the future but there’s also big opportunities. I highly recommend this book because it goes into a lot more details and I think it will extend these arguments a lot more than I did but I think this summary gives you a good picture about what this book is about.

That’s it for this month, next month I’m going to be doing Godel, Escher, Bach by Douglas Hofsteadter. This is a very interesting book and it is really I think a very soft and interesting interlude into what is consciousness and how patterns express themselves in particular in music, in literature, and in art.

So thank you for listening and I will see you next month.