 Welcome, everybody, to the Berkman Center luncheon. My name is Karim Lakhani. I'm an associate of the Berkman Center, a faculty associate of the Berkman Center, as well as a professor over at the Harvard Business School. And it's my great pleasure to introduce Jim Besson for his talk on his new book. I've known Jim for over 12 years. I met when I was a PhD student at MIT. And Jim immediately struck me as being very different than the typical academic you would meet. He actually has a lot of experience with technology. His entrepreneur himself in the world of technology. And his first company actually helped establish the desktop publishing paradigm, as well as the first commercial WYSIWYGZ. And so in many ways, we are indebted to him for helping to get one of the first office worker productivity revolutions going way back when, in the 80s. And Jim, that experience was foundational for Jim to both understand technology, understand innovation, and understand the challenges of the pathway from invention to innovation. And in many ways, this book is his take. And it's a very novel and an important take on trying to understand the ongoing waves of technological revolutions that we observe in the economy and the worrisome effects we think about in terms of jobs and inequality and growth. And so I found the book to be fascinating. It's a great read, both in terms of conceptually, the arguments that Jim is making, as well as the empirical evidence he lays to bear. And he's really, I think, from an academic, thoughtful perspective, he's really pushing both policy developers and policy makers and policymakers to think hard about historical evidence about innovation and jobs and wages, as well as the current dilemmas that we face in terms of what we think is happening in the economy and potential solutions to the problems we think are going on. So with that, I'm going to give it to Jim. He's got about a 20 to 25 minute presentation on the book. And then I'll ask a few questions, and then we'll open up to the audience for more questions. OK, Jim. Thanks, Karim. So the book is called Learning by Doing. We tend to think about technology. A lot of people confuse, I think, technology and inventions. In the bigger picture, technology involves a lot of knowledge and skills, much of it held by large numbers of people, not just inventors. And developing that new knowledge when we have a major new wave of technology is a big social problem, a big social challenge. And that's the focus. Today, we have new technology everywhere. Technology has changed the way we work, the way we shop, the way we entertain ourselves, the way we communicate with each other or not. Technology is everywhere except in most people's wallets. Since the beginning of the personal computer revolution, the median wage has been stagnant in this country after rising over a long period of time. So some people see this and think it's an evidence that we're really at a fundamental break from history, a fundamental break from the past. But before we can say that we're really breaking with history, we have to understand what that history is. And history is important. Over the last 200 years, technology has largely been responsible for over 10-fold increase in the average way. So if technology has led to stagnant wakening more recently, is that a break? Of course, technology is not the only thing, but technology has been the most important thing driving that huge increase. But in fact, in the past, we saw something similar. So these are the wages of weavers and spinners in the Lowell Textile Mills starting at the time of the Industrial Revolution. And they were stagnant too for 30 or 40 years, and basically until after the Civil War, when they started growing very rapidly. This was a time, though, even those first 30 or 40 years when there was huge productivity gains in weaving. The reasons why things were stagnant then are certainly different now. The technology was different. The society was different. The challenges were different. But my thinking is that there's enough about what happened then that it's worth looking to understand what may be happening now. There's enough similarity in the challenges we face so that it provides some insight. So an awful lot of people will argue these days that the reason we're seeing stagnant wages is that robots are stealing our jobs. But this, too, is nothing particularly new. This is the weave room of a textile mill in Fall River circa 1910. And you'll notice there's lots of machines, but very few workers, even then. The best mills at that time had about 24 looms per weaver. So it was a huge amount of automation. Weaving was perhaps the very tedious task of weaving. And it was one of the most important inventions of the technologies of the Industrial Revolution. Over the course of 100 years, effectively, 98% of the tasks that a weaver performed were automated. This shows the amount of time that a weaver needed to produce a yard of cloth relative to the hand loom back in 1810. And it had dropped to only 2% of that time by a little after the turn of the 20th century. 98% automation, robots stealing jobs, machines taking over from workers, mass unemployment, right? No, the weaving jobs actually increased. The blue line is showing the number of cotton textile workers. And it's the same if you just look at weavers. What happened? The economy and technology were just more dynamic than we think when we talk about robots or machines taking over. The automation reduced the price of cotton cloth. Lower price cloth meant much greater demand for that, for cloth. So people started consuming much more cloth. At the beginning of the 19th century, producing cloth was a very, very time-consuming process, very laborious. A lot of it was done at home. And people wore very few items of clothing, and they used cloth for very few things other than clothing. By the 20th century, cotton and other textiles were being used for all sorts of things, and people had consumed much, much more clothing. So there's a very dynamic response. We see something similar today. We know that the ATM machine has taken over work of bank tellers, right? And people will tell you that it's eliminated the jobs of bank tellers. Not so. The top line shows the number of bank tellers employed. The bottom line shows the number of ATM machines installed in the US. A huge increase in ATM machines from 95 to 2005, and not much of a dent in number of bank tellers. Why? It clearly took over work from bank tellers, but what happened was it made it cheaper to operate a bank branch. So there were many more, the banks installed many more branches. The job of the bank teller became less a job about handling cash and more a job about interacting with customers, helping market those customers, some of the higher margin products, mortgages, loans, whatever, that the bank sells. So the teller became part of more of a marketing specialist. The dynamic response, economic response, meant that those jobs did not disappear, and it meant that the skills of the teller changed. Of course, there are other cases we can look at where jobs did disappear. So in the 1970s and for almost 100 years prior, most printed type was set on linotype machines like this. And in the 1980s, we got computerized publishing, desktop publishing, and the number of typesetters and compositors dropped precipitously. But at the same time, a lot of that work was taken over by desktop publishers and graphic designers. And in fact, there were more graphic designers jobs added than typesetter jobs lost. So you need to look at the entire picture to get a sense of jobs. And what you see is there are some areas like manufacturing where we have a net loss of jobs. And part of that is technology. Part of that is globalization, offshoring. And part of that has been happening for a lot longer than the 1980s. It's been happening. The manufacturing share has been declining in some industries since the 20s or 30s and certainly since 1950. Those are mature technologies, so they are being gradually replaced. But in the areas where the occupations where computers are being used most heavily, the total level of employment growth was faster than average. Computers weren't on net replacing jobs. They were displacing people to new jobs. So I'm going to argue that one of the key reasons technology is not helping wages these days is because of the challenges of developing new skills and knowledge. And again, it's helpful to look back at the textile mills. Lowell, Massachusetts, was the center of the textile revolution. The Merrimack Valley was sort of the Silicon Valley of the day. Here's Lowell 100 years ago. The Lowell mills hired to man the power looms, to operate the power looms, I should say. They hired mill girls from all over New England, often teenage farm girls before they were married. And they brought them into a unique environment where there were all sorts of cultural and educational opportunities. We're familiar with the way Google and some of the other Silicon Valley firms offer all sorts of amenities to attract highly talented, skilled people. There was something like that going on then. So this is Lucy Larkham, who entered the mills when she was a 12-year-old girl. She later became a college professor and is a dorm named for Wheelock College. While she was at the mills, she studied German and botany. She heard lectures by John Quincy Adams and Ralph Waldo Emerson. She wrote poetry for one of the factory girls' magazines. And her poetry caught the attention of Longfellow and John Greenleaf Whittier, who became her mentor and started her essentially a career of writing and teaching. She herself compared the mills to women's colleges of several decades later. But the key thing is, the reason the mill owners did this was to attract very talented people who could go into this strange environment, noisy, all sorts of weird machines doing things that were, you can imagine at that time, highly futuristic. And they had to acquire a whole new set of skills to use those machines effectively. So learning was key. We tend to think about the invention, but there's a huge amount of knowledge that goes into building, installing, operating, maintaining, organizing, marketing, a major new technology like this. And it takes a long time for it to develop. It takes a long time for the institutions to develop that can train people for the labor markets that can provide the right incentives for people. And it takes a long time for the technology itself as part of that process to improve and develop. And it's difficult because a lot of that knowledge can't be learned in the classroom. It has to be learned through experience on the job. We can learn by doing. So we can see that with the mill girls. Here is the productivity of a weaver in one of the Lowell mills based on the months that she was on the job. The number of yards she could, of cloths she could produce in an hour. And it rose dramatically over time. And in economic terms, that represents a significant investment. Some historians tend to think of factory workers as unskilled, and they were unskilled when they walked in the doors the first day, but they acquired skills on the job that were very significant, that were critical to the economic success of the technology. And that required a substantial investment. In fact, we can calculate the investment. And it turns out it's comparable. The investment in these mill girls was comparable to investments that a craftsman might make in skills. So they had real skills. Skills was an important part of the equation. But what about wages? I already mentioned about how wages were stagnant for a long time. What went on then was that there really wasn't a labor market at the beginning. In the early years, 1830s, 1840s, only about 18% of the weavers they hired had previous experience. In other words, they were training everybody. There was no labor market. You couldn't, you had a difficult time if you wanted to recruit somebody. And a worker who was looking at these skills had to realize when she leaves this job, there's not another employer who's gonna bit her away or another place she can get work. She can't look forward to a career, so she's got less incentive to invest in developing skills. After the Civil War, a robust labor market developed, and there's some reasons why it took so long. And then wages started growing very rapidly. Then 87% of the weavers had previous experience, and there was a very active turnover. If I worked at one mill and I didn't like what I was getting paid, I could get a job at another mill because somebody would pay me better. And so there was competition for the skills of the weavers. A number of reasons why it took so long, took decades to develop a labor market. And one was that the technology kept changing and it wasn't standardized. Different mills use different technology. They organize work differently. They provide a different sorts of training and also because of a number of unique problems, there was little mobility, little ability for employees to go from one mill to another. I won't go into the, a lot of the details are very specific to the history. But the net result was once a labor market developed and once the training institutions developed and once things were standardized, wages shot up. And so by 1910, the weavers were making three times as much as they were making in the 1830s. And so these mill girls, this is Amiske in around 1910, they were showing off their Sunday best, not their work smocks when they got photographed. What about wages today? You know, we can look at the line of type operators, the typesetters and the graphic designers and interestingly the median wage of the graphic designer last year was only a buck higher than the median wage of a typesetter in 1976. Very slow growth in wages. Yet, I think anybody who knows that industry would be hard-pressed to tell you that the graphic designers don't have a much broader range of skills than the typesetter operators had. What's going on? Well, we have similar problems with lack of standardization, lack of stability, lack of labor market. So first we had the typesetter compositors and they were replaced by print designers and desktop publishers. But then along came the web and we now had to have web design skills and along came the smartphone and we had to have mobile design skills and the graphic design started fragmenting into all sorts of specialties. You have information architects, user interaction specialists. Things are constantly changing. A few years ago you had to know Flash. That's obsolete now. You need to know HTML5. Standards are constantly changing. And it's difficult for the average designer. The schools can't keep up. You can go to graphic design school and get a BA in graphic design but what you learn is maybe out of date by the time you graduate and labor markets don't necessarily recognize the new skills. I may have five years of experience but the employer doesn't know do I have the right experience? Do I have the experience on the latest tools or do I have less valuable experience? The top designers can teach themselves and they are doing so continually and they can develop reputation. So what you have is sort of a divergence. The average designer has a harder time and their pay is stagnant. The top designers see rising pay and they're doing well. And you see this problem more generally. So there's a gap between what the average worker makes in computer intensive occupations. The median wage is what the average worker makes and if you look at the 90th percentile, the wages for the 90th percentile are growing much, much faster. There's a growing gap and we're seeing rising economic inequality even within occupations. In fact, much of the growth in inequality is within about half of it is within occupations. And this represents to me a skills gap that there's a gap between what the most talented people, the skills of the most talented people and the skills that can be acquired by the average person. We can solve that gap potentially, but that's the nature of the social challenge. So it's a story, in my view, of technology is not replacing workers with machines. It's displacing them. It's putting them into new jobs where they have to learn new skills. That's difficult for many people. And so they may remain employed. Unemployment is not 30%, but they are facing a difficult challenge in earning good pay. So the key thing is new skills and knowledge and how do we develop it? And I wanna, since this is Berkman, I'll talk a little about the role of knowledge sharing. There are a number of policies that we can think about, but we can talk about knowledge sharing in a number of different ways. Things like open source and direct sharing, open standards, the role of job hopping and informal exchange of knowledge. All of these things are important in developing that broad base of new knowledge and skills. And some of those things are also very important in terms of technological areas like the internet. They're important for two reasons. One, they help knowledge spread, and two, they help establish labor markets and getting labor markets with standardized skills is key to getting higher pay for those skills. So there are some interesting parallels to the past, and I'll just go very briefly since we wanna be short and open up for discussion. So we have open source software and that is something that's very new and different, but I should emphasize this is a long history, a largely forgotten history of inventors sharing in all sorts of technologies. So William Gilmore was a British mechanic who came to the US and in 1817 developed the first power, the design of a power loom that eventually became the industry, the most popularly used design in the industry. He shared that design. He charged another mechanic $10 for the blueprints and it was freely available to all mechanics. You see similar sharing in a variety of other, particularly early stage technologies. It wasn't open source per se, but a very similar ethos where mechanics would exchange designs with each other. They were not happy necessarily if a downstream manufacturer used their designs, but there was a great deal of communication and sharing among the community of mechanics. In fact, one source talks about they viewed themselves as the international fraternity of mechanicians. Open standards are a very important part of what we deal with today. Open standards for the internet have been key in providing and ensuring its high degree of innovativeness I recall going to conferences in the early 90s where firms were touting interactive TV. And at that time the internet was seen as being sort of a poor and inconsequential cousin. Because of open standards, and I write about this in the book, the level of innovation and growth of skills and knowledge on the internet just far outstripped the interactive TV people. So despite tens of billion dollars of being invested in interactive TV, it was pretty much dead within two years and the internet took over. That sort of use of open standards was also critical in the past. We're all familiar with the typewriter and we used the QWERTY keyboard today. When the typewriter first came out there were different keyboards. This is the ideal, a typewriter with a Hammond ideal keyboard. But standards and having open standards it was critical towards developing a labor market. So this is the number of stenographer typists. Standardization, the first commercial typewriter was 1873 by around 1900. They had standardized on the QWERTY keyboard and you can see the market took off. And there was a huge growth in the number of typists and stenographers which changed the nature of the office, changed the role of women in the economy. But the key thing was getting to the standard. Another thing is the role of employee mobility. I noted about employee mobility earlier in the mills. A number of very interesting studies have attributed the relative success of Silicon Valley to Route 128 in computing to a different attitude about employee mobility. And in particular different laws about the enforcement of employee non-compete agreements in California. In California, employees are much freer to go to another company. They are much freer to start a spin-off. And so their research has shown that they are more likely to start a spin-off. They are more likely to start a new company. They are more likely to innovate in California or generally in states that do not enforce non-compete agreements. The sort of networks of skills was also critical in the Industrial Revolution. The core of mechanics that started in Providence, Rhode Island branched out across the Northeast and there was active exchange of knowledge, active job hopping. People would go and work in each other's shops for periods of time to gain the broadest base of skills. So I've drawn some parallels. I think the key takeaway here is that broadly shared knowledge is key to broadly shared wealth. And so we wanna focus on discussion and we can think about different sort of policy questions that might affect knowledge sharing and employee mobility. One is things like non-compete agreements and trade secrecy law, which can restrict the employee mobility. Other things are occupational licensing. One of the U.S. workforce now has nearly 30% of the workers are subject to occupational licensing restriction compared to about 5% in 1950. There's been a huge increase and all sorts of, in many cases, mid-skill or even low-skill occupations are subject to occupational licensing. In many cases, that can impede the employee mobility or the ability to enter a new occupation. We can talk about software patents and their role in affecting startups. We can talk about things like net neutrality as possibly restricting innovation. I wanted to throw a few things out, but to put it into context, one of the troubling things about the last 10 or 20 years is that we're seeing declines in employee mobility. The workforce is substantially less mobile today than it was in 1990, and we're seeing declines in firm startups. To the point where they were in the last, after the Great Recession, there were actually more firm exits than there were startups. Things have improved since then a bit, but we're seeing historically, over the last, since 1980, a very significant drop in the role of startup firms in being formed, and because we know about the role of startup firms in fostering new technologies, that's an important issue. So thanks, and now I'll turn it over and I guess to Karim to... I'll ask starting questions, but then we'll leave it up to all of you to continue the conversation. And I guess the first thing I would say is, there's been a lot of books, there's a lot of hand-wringing right now about AI, and how AI is gonna, not just take away sort of blue collar jobs, like the Google car will come in and take away taxi drivers and so on, but also law jobs too. So there's a big investment, like just as there's a big investment in marketing technology and financial technology, there's also now lots of investments in law tech, right? A lot of startups are coming in saying, if you don't need lawyers for discovery, we can just get a smart machine to do it for us. So what's your sense in terms of the real worry in the amongst economists, amongst professions about sort of technology eating up white collar jobs? And what do you sort of see, like what would you tell this audience if they care about law and discovery and so forth, like what's gonna happen to them? And I mean, it was interesting in your book to how you sort of showed there was almost two or three generations before wages went up, right? So do we expect that to continue or do we expect that to shorten? And what kind of displacement might you see? That's more than one question. Sure, yeah. So let's talk about white collar jobs. So technology's been replacing white collar jobs for 30, 40 years, at least, right? I mean, we've had accounting systems, we've had desktop publishing. The number of jobs, so what tends to happen with, and here's the critical distinction, what tends to happen is technology comes in and automates some subset of the tasks of performing a job. If that is value, if the technology's not too mature, that will tend to, at the same time, increase demand by lowering prices, by adding new value to the products or the services provided, which in turn makes the remaining tasks, the non-automated tasks, more valuable. So there's a constant trade-off. Some tasks are getting automated, but others are becoming more valuable, and there are new tasks being created to deal with the new machines and to deal with the new technology. So you see new occupations being created. AI is only going to become, jobs is only gonna become the real issue when technology is automating all of 100% of the tasks. And I don't think we're near that. Now some people think we are, some people think there's gonna be a singularity. Sure, sure. Around the corner and... If that happens, we're all screwed, but we're on that. Right, right, right, right. So some people will also argue that what artificial intelligence can do, I don't wanna get into this because I'm not sure I necessarily believe it or I necessarily know that there is some inherently human things that we want for our service. We want human interaction. And unless the technology is really able to engage us emotionally, and maybe it will be able to do it, it can't deal with it. So I don't think the fundamental issue, there's always going to be jobs disappearing, whether they're white collar or blue collar. And there's also gonna be new skills needed and new jobs appearing. So it's a turnover. And the real issue is how do we... That's what's been happening. I mean, what I talked about is what's been happening so far. I don't see it changing radically in the near future. And the difficulty is, I don't mean to dismiss how difficult the transition is that people have to make. Acquiring those new skills is difficult. So that was the second, why does it take so damn long and what's gonna happen? So you can look at different occupations and you see different things happening. So desktop publishing, it's not... I think it's very clear. We're not going to see that sort of... The wage picture there for the average worker change until we have stable business models for publishing, which is still in huge flux right now until we see a greater stability in the technology. Not that the technology needs to stop dead in its tracks, but that it occurs within a framework so that we can have stable businesses, stable labor markets, and key sets of core skills that are understood and needed and people can acquire in school. And so the schools can catch up. You can look at other occupations. Healthcare is a sector where there's a whole lot of new technology and there are also new business models, but we're seeing the skills catch up basically. So you have a lot of mid-skill workers in healthcare. Their skills have become more valuable. The employment is going up. One of the key things that I write about is for some of the mid-skill jobs is new business models, like the ambulatory surgery center. So they have all of these outpatient clinics where you can get shoulder, they specialize in something, shoulder surgery or laser eye surgery. And in this environment, even someone like a licensed practical nurse who may only have 12 months of post-secondary training can come in, learn valuable skills on the job, and within a few years earn solid middle-class pay. Wages have been going up, jobs have been growing. And in fact, there's been a huge shift in healthcare from key care, the portion of care provided by the top-level providers, by the doctors and dentists, and the portion of prepare care provided by the mid-skill providers, anywhere from medical assistants and licensed nurse practitioners to licensed practical nurses, to nurse practitioners and physicians assistants. So this, I think, calculates something like two million new jobs created in the mid-skill area. So I don't know which model is going to proceed in dominating the economy over the next, and how long things will take, but I think that's sort of the, you can see both things happening, and it's a question of how long it's gonna take the right thing to dominate. Yeah, and I think part of the worry is that, you know, that the elites weren't affected before, right? So the middle-girls you saw weren't Harvard grads, but in fact, they went there to get the education, but I think what's happening is that, part of why we see this, for example, is that we could imagine that the market searcher for law firms will change radically, given this technology, and if law grads' initial work was to go read volumes of documents and then be the heuristic classifier of knowledge, and that can be done much more efficiently through AI, then that's gonna be the issue, and I think we see the same thing in marketing, we see the same thing in a range of occupations where the elites are now, their jobs are being impacted, and hence the worry, but anyway, so we won't have an answer for that yet, but let's open it up for questions and thoughts and comments. Oh, please introduce yourselves so that everybody knows who you are. Brian Kay in MIT, Initiative on the Digital Economy. Jim, could you say something about, well, let me start with the observation that weaving is more like manufacturing than it is like those other occupations, so why don't we see the same kind of standardization and wage-raising effect in manufacturing that you point out in weaving? Is it because it's a two heterogeneous category that there's too much dispersion, and can you tie that into cluster effects? So going by your theory of standardization promoting active markets, promoting higher wages, do you get higher tides raising boats in Silicon Valley than we got in Route 128, and is this an explanation of why certain industries, like textile manufacturing, remained concentrated in one area? Automobiles would be another example. How do you think about the geographic and clustering dynamics? It's a two questions. Right. I thought they might tell you, but I wasn't sure. So, okay, well, weaving is manufacturing, so I'm not sure what, maybe I should update the weaving story because I let it end in 1910. So employment grew and wages grew up through the 1920s. With the depression, things leveled off. I was focusing on cotton textiles, and so there's some complications that you had. But basically you had a stable population employment of textile workers over different types of cloth. Up until 1970s, slow decline, I mean, the technology continued to improve labor productivity about 3% a year. So you had this constant, constant, constant improvement in the technology, saving the amount of, reducing the amount of labor needed to produce a yard of cloth. And it continued, and it, what affected the textile industry in the U.S. was more than technology, though, around 2000, globalization, off-shoring. Well, it moved to the south. It moved to the south in the late, starting in the late 19th century. But even there, much of it was in the south. It was still, in 1970, a weaver was earning something close to the median wage. It was a middle-class occupation. There's not gonna be the same kind of skills base there, right? Where's it be? So, I make a distinction, and I think this is a critical thing and it relates to the geography as well, is the problems of developing skills are difficult for early-stage technologies. Things are not standardized. It's hard to teach in school. They're changing rapidly. Once things become standardized, it becomes much easier to relocate. Once something can be taught in a classroom, so you see this. For instance, the periodic table, I write about this in the book, the periodic table and techniques of analytical chemistry in the 1860s, 1870s, standardized a large degree of chemical knowledge and made it much easier to train chemists. Training chemists had been something of an apprenticeship prior to that. It became something that classroom training worked so that you see schools across the world starting to train chemists and you see the practice of chemistry, which was originally clustered around small numbers of labs by lead scientists, run by lead scientists, growing all over the world. So, you see, there's a constant pattern where new technologies are often critically developed in geographical areas because the exchange of knowledge from person to person, from job hopping through other means of informal exchange, is critical in an early-stage technology because so much of the knowledge is informal or tacit. Once things become more standardized, semiconductor technology can migrate from Silicon Valley to Taiwan. It becomes much easier. It's still difficult to set up a plant in Taiwan and they went through hell setting up the first plants and getting them running efficiently, but it was a less difficult problem than the original problem of developing the technology. So, that's the interplay between geography and sort of the maturation of technology. So, as a technology matures, a number of things happen. It becomes more standardized, but also the demand effects become less. So, in the early years, there was all this pent-up demand. If I could lower the price of cloth, I could greatly improve the consumption of cloth. Today, if you lower the price of cotton cloth 5%, you're not gonna be generating all that much additional consumption and you're not gonna be as... There's gonna be a net loss of jobs if you have a 5% savings and labor cost. My name is Francesca Schwartzman. I'm a researcher at the Fletcher School at Tuft University and I was wondering, so the way you describe the displacement of workers, there is a lack for a certain period of time where the workers have to shift and learn new things. And within these time, you have to enable the workers to still pay their rent and keep up their standard of living. So, who do you see should be responsible for that shift in the workforce? Do you think governments should be taking care of that? Or do you think that there is a massive shift in corporate social responsibility and big companies take care of that? And how would you incentivize the companies to actually do this and take care of that shift? Well, everybody's responsible. That's the easy answer. So, companies have a difficulty at the same time in terms of they're not able to hire the people with the skills they want. Every year, there are these surveys of corporate managers and 35, 40% regularly report they have difficulty hiring people with the right skills. So, what do they do about it? So, some companies are investing in training programs working with local community colleges setting up work study programs so that people can learn some skills in the classroom and some skills on the job. There, some trade, some industries are setting up certification programs so that skills learned on the job through experience can be certified so that they become standardized in a way so that other companies can hire them. But that's providing that sort of retraining or vocational training, perhaps an apprenticeship program. Companies in the US are experimenting with apprenticeship programs. All of that becomes one part of the puzzle. In terms of government support or social support so there's some interesting research now coming out both in the past and in the present. So, it turns out, Avner Greif is a very interesting economic historian at Stanford and so he's done studies about the level of social welfare support and what that had to do with innovation in the industrial revolution in England and in Europe and so it turns out that the counties or areas that had a social safety net, a significant social safety net were actually more innovative. They were able to adopt new technologies more rapidly and he attributes this to basically you had less of a fear of revolution and a rebellion. People were, by providing the support people were able to focus on working with a new technology, were less resistant to it. I don't know, I don't know the whole answer but I, yes. I thank you for the fantastic talk. I'm Shuang, I'm an anthropologist at Harvard. I'm recently doing a research of Uber in China so it's really interesting to me. So I'm wondering if you can talk more about the increasing economic inequality you pointed out in your presentation. I very much agree with you that it's not really about replacing labor but more about displacing. For instance, in the Uber case it's not about nobody's driving taxis right now but how Uber drivers are of a very different demography of traditional taxi drivers. Over 50% of them are college graduates. So this is one question and also the follow up question is how do you think the suggestions you made in the end is really going to change this increasing economic inequality because as we can see the knowledge is almost the most stagnant social hierarchy in our society. So our education system going to change that or is the technology innovation going to intensify this economic inequality even more? I'm not sure I fully got your question but let me ramble around and see if I stumble on it. So I think on the second question, the question is in terms of knowledge needed to succeed. I mean I think in the book what was fascinating was that it's not so much about the formal knowledge that the universities teach, right? But the learning by doing, that's the title of the book, the learning by doing which is like, this is all sort of sticky knowledge that you acquire by participating in tasks and that is highly uncertain until it gels. So maybe you wanna, I think that's what I got. Yeah, so yes there is a hierarchy in terms of education and perhaps in some ways, I write about this a bit in the book that inequality is actually getting worse in some ways. I mean you're seeing if you look at government funding it tends to go to four year research universities and while funding going to community colleges has basically been stagnant. So there's great inequality and people write about how there's educational inequality. So I think that's an issue but as Kareem was pointing out a lot of what we have to think about with technology isn't education per se. So there's this interesting interplay between education and technical knowledge. In the early mills they recruited literate. You had to be literate. You had to be able to read, to work in the mills. Now it's not because you needed to read to operate the loom. It's because if they hired literate mill girls it turns out that the literate ones were more productive. They were better at learning in some general sense so they could learn this new technology and acquire a different set of skills. So the new technology comes along initially there's a greater demand for educated workers even if their education isn't necessarily required for performing the job. As technologies have matured and this was true in textiles and there's some research showing it's true more generally the education requirements often go down. And as the institution's developed to learn on the job to provide vocational training, technical training to people who have less formal schooling we're providing a way to develop knowledge more broadly. And I think that's sort of the nut of the issue. We tend to focus when we talk about skills and education. When we talk about skills and knowledge these days we tend to focus on education but a central part of my argument is it's much broader than just education. Yeah, and if I may add, I've been engaging in this Twitter debate if you can have a debate on Twitter about this question. So, you know, certainly in Silicon Valley there's a view that formal schooling is not useful for the skills needed to succeed in the data technology revolution that we're now living in. And I think it's a fascinating, we face this at the business school like should our MBA students learn how to code, right? We don't hire computer scientists. Am I gonna teach them Fortran and Pascal or C++? Probably not. But those skills are needed, right? So do you go outside and do you hire somebody else to come in and teach that? There's an explosion in sort of the Code Academy online learning types of things as well that are trying to address this gap between learning how to think about the basics of programming languages that a CS major would get versus actually being able to program and create an application. But I think the interesting thing is that these skills again what seems to me to be slightly different this time around is that the change required is so widespread. So just as law graduates will need to have a program so will the business graduates and so will biologists and so will chemists and so on. So there's almost a general purpose technology aspect to this and that the formal, our institutions aren't yet capable enough to deal with providing those skills in a formal way. So then we have to go outside the institution or rely on people to just learn themselves to be able to pull that off. Yeah, and that's, you hit on a key point which is things are more difficult now because so many areas of technology and of industry are being affected at once whereas you think about weaving as part of manufacturing which back then was only 20% of the economy and here we're talking about information technology affecting all industries. All industries, maybe 70% of industries intensively. Yeah, your name's her and. My name's Benjamin Melanson, I am a web developer and it's an exciting industry for a lot of us in it because in large part because you have people who may have had education but it's in other things. Yeah, the universe is still caught up even with web development which changes all the time but compared to other areas of information technology it's at least been around a little while. I guess I'll just go for the real big question, it's we have the productive capabilities in the world and have for quite a while to provide for what everybody needs. We've not been distributing it well and we've done even worse giving people sort of the even greater resources and control over their lives that would and allow people to really invest in learning new skills and invest in setting their kids up for doing some really exciting things and it's because you mostly don't have much of a safety net. So I just, I mean, I don't know quite where exactly what I want to ask with that, it's just that this isn't new but it may become more blindingly obvious to people that the problem isn't the technology, the problem is sort of the distribution of control over resources because you can all learn to code but not gonna make sense. It's not like, I almost feel like the wages at the top of industries aren't high enough. Like if it were obvious that if you're at the top of the industry, if you just go out and learn for yourself and become on top of your game that you're gonna have a huge, you're gonna be set for life, like people would find their way there. And so it's sort of, it's like the old thing, like people say, oh no one will do this job, but no one will do a job at the price that people are willing to pay for it. So I guess it's sort of, do we have to look at what prices the market is willing to bear for certain times of work and why that is? And I guess that goes back to what resources people have if there would be huge industries for the training of people and there are, but they'd be even bigger and better if the people who needed the training had the resources to pay for training. And just the whole point that we have a huge economy, a huge amount of economic activity, but what is that working towards? And right now with extreme inequality and the inequality we've generally always had, you're ultimately working towards making richer people richer. And just how that plays out through all of these, yeah, just where technology goes, where the technology moves to, and if there's anything from the frame of reference of this discussion that would inform sort of how we try to structure our, our own advancement as workers in any industry. So, I mean, I picked one thing there which is, so my argument is we faced this extreme growing inequality in the early 19th, first half of the 19th century and developing a widespread, a large number of people who had skills that could command good pay was the key towards making things more equal and more equitable. And so I am guessing that a similar challenge faces us today, which leads me to disagree with you on the point that it's not just what the 90th percentile, the 90th percentile might be a sign or a symbol, but knowing that people can go out to Silicon Valley and become billionaires, and sometimes in just a few years, doesn't necessarily help the average person who wants to be able to acquire a skill that can command a good wage. And so it's, there, to some extent, two different, two different problems. And you see, one of the reasons you see such extreme differences is because we don't have a way of we're not doing well at lifting up the middle. Jim, you had some specific policy recommendations on this. So maybe just go through that, just to sort of, not in detail, but there's like four or five categories of policy that you think could potentially help towards this, both reducing this inequality, but also potentially shrinking the time needed for this adjust, for this adjust. Okay, that's good. Let me see if I can hit on the main thing. So I guess I've mentioned a couple ideas, but to the extent that things are becoming standardized, having vocational training, having, you know, supporting the community colleges, work study programs, apprenticeship programs, those sorts of policies, I think, are very helpful, both whether they're done by corporate or government. The employee mobility is very important. So we've taken some steps in the wrong direction, regarding employee mobility in terms of much more extensive use of employee non-compete agreements, broadening of trade secrecy protections, so that at least in some states, it makes it difficult for technical people to switch jobs, standardization and certifications, and that tends to happen mostly, I think, in an industry level, but to the extent that government can, hey, well, government does play an important role. Government procurement plays an important role. So in the past, government procurement has played a critical role in advancing early-stage technologies in things like computers and semiconductors, wireless communications, but also in the 19th century, mechanical skills, what's called interchangeable parts. One of the ways it did that was by making sure that open standards were used, that there was a knowledge exchange between people, that there were diverse players so that you had a lot of startups involved. Increasingly, government procurement has tended to favor politically influential, large companies. You think about defense procurement, particularly under Rumsfeld, when there were also security concerns. There was a clampdown on graduates, foreign graduate students, and the research projects were directed towards large defense contractors rather than startups, and you're seeing a move away from supporting startups. In other areas, you see the decline in employee mobility, you see the decline in startups. I think there are a number of complex factors going on with both of those, but there's some things that, some research is pointing to changes. So there is research showing that non-compete agreements have an effect on employee mobility and innovation in startups. There is research showing that software patent litigation has an effect on startups and venture capital funding. Those are a few things. Hi, how's it show up at Berkman Center? What you've talked about is very specific, well, sounds very specific to the United States, or at least to kind of develop to Western economies. To what extent do you think it applies to the developing world? A good question, and I probably should say I don't know. So, developing world has, well, I've talked about the developing world, and I've talked about sort of the middle income world. A lot of the problems faced in economic development and among the poorest countries are problems of transferring existing technology and getting up to speed on that. Now that's involved some of the same issues about skills, but in some ways it's easier because that technology is more standardized. So, I mean, we see that pattern happening. It's striking how some of the same techniques, the same organizational tools, the same approaches in terms of training labor that were used in Lowell, Massachusetts were later used in the US South and were later used in Japan and have more recently been used in China and getting their textile industries started up. To the extent of having mill girls in dormitories. Right, if you look at Shenzhen right now, very much looked like in China, very much looked like Lowell, Massachusetts. Yeah, yeah, yeah. And I mean, you see the same, I mean, to some extent there's conscious imitation. You know, to some extent, which I think is probably generally a good thing. But, you know, those examples show that it can work. It's not the only issue. I mean, the problem with developing nations is that they have a lot of problems to solve all at once. And so, developing that knowledge and that skill is a difficult problem and it's part of the solution. I can give another illustration, which is by 1910, the same textile equipment that was used in US and England was being shipped around the world. It was in India, it was in Japan, it was in Russia. But, the productivity of English and US workers was six times the productivity of workers on the same equipment. In some cases, even with English mill managers. And it's because of the difficulty of transferring the skills and acquiring those skills. That said, there becomes sort of a second level problem. Once nations have gotten to, you know, the people talk about the, what is it, the middle income gap, you know, where a nation can acquire the new technologies, develop the basis skills, but then has difficulty moving on to sort of the frontier of cutting edge technologies. In the book, I write about this question of why doesn't Japan have a vibrant software industry? Japan had, you know, it was the second to the US in terms of its computer industry, but its computer industry is now beginning to suffer because it's got a very weak software industry. And the difficulties the software industry faces, to some extent stemmed from the trade-offs that Japan had to make in terms of getting up to, getting to become top level in computing. So you have very dominant computer companies. You have very little in the way of independent software companies in Japan. You don't have a great deal of employee mobility, and so it becomes very difficult for startups to hire people in Japan in software. So there's a whole other set of problems. So the general, the specific issues I talk about and the policy issues are really US-centric. I think sort of the general problem of how do we develop broad-based technical knowledge is common and it has its own particular features in a developing environment. Time for one more question. Yes, please. I'm a law professor at Suffolk. So I was so intrigued by the Lowell Mills beginning because growing up here, the Lowell Mills have a whole, I mean, they're a myth onto themselves and there's a story about feminism, there's a story about women in work at a time when that was, I mean, you can tell lots of different stories about the Lowell Mills. And so that got me thinking about the role of the company in communities. That is the role of the Lowell Mills in this community was diverse. And it seems to me that there's another story, or I'm asking, is there another story one could tell about the difference in wages or the stagnancy in wages now as compared to previous shifts that has to do with the change of the role of the company in our communities. That is what we think the, what role we think the company does or should play in our society. And so that's more of like a sociological question, but it's also the question of the value that we think the company brings. And if our dialogue has shifted to a more sort of efficiency economic, sort of law and economic model about what we think these different actors in our society play, the role of the company maybe has changed and its job is not to train, its job is not to ennoble or anything like that. And so I'm just wondering if there is, there may be other ways of explaining why the wages have been stagnant in terms of what we think companies are supposed to do. What a successful company is, in fact. Good question. First off, an important caveat I should say is I don't think the issues I'm talking about or I don't think technology is the entire question of stagnant wages. There are a lot of factors involved. I may fall into, so this is a good question. First I wanna be careful that we don't romanticize the mill owners in Lowell, because they were some tough bastards at times too. So they squashed strikes starting from 1836. They reduced wages. They were a combined, I mean they had very definite interest and wages were stagnant because they wanted them to be stagnant. No different from the Silicon Valley wage fixing lawsuits. I mean the Apple, Google everybody was contributing to. So the Silicon Valley companies may want to attract, may offer amenities to attract talented people but at the same time they're looking out for their bottom line as well. You can look, well, so there may be an issue in terms of short-term outlook. So there is a question, so people like Peter Cappelli raise this issue. Employers are whining about the lack of skills. Why don't they just invest more in training? And if you look at in terms of what companies spend on formal training, it's declined over the last 20 years. That may not be a good statistic and formal training is I think maybe not, this is the kind of training that matters so much with new technology. A lot of it's learned on the job. Much more significant investments learned on the job. But yeah, I think there probably is a real element of short-term outlook by company. You know, not wanting to feel that they need to make long-term investments in skills development. You see that in some industries and you see different patterns in other industries. Yeah, and I think if I can just ask anything else. I mean I think there's certainly these concerns about financialization and how our capital markets are causing the short-term thing. But you know, my sense again on the early comment that the current digital revolution that we're sort of seeing so many industries get impacted. 70% of the industries are becoming really software industries. If you sort of think about an airline merger and airline is as much a software company as it is as it flies planes to your hospital that I think the demand for skills and the demand for the workers is going off the roof, right? And so right now we see the medical school competing with Facebook and with Google for the same data scientist. And so the question will become is how will the work, and that's like, this is one example, right? Pfizer does the same thing. And the question becomes how will the, will market forces, you know, at least help alleviate that by either raising more wages or making more investments in those sorts of things. But I think the tension is never gonna go away, right? The tension that, you know, the jobs of the world will want to suppress wages, right? And we'll want to limit curtail mobility, right? And what push for non-competes versus what's good for the general social welfare. And I think that tension is not gonna go away. I think we're gonna be in this dynamic environment. Going forward. So we're almost out of time. I just want to again urge you to read the book. I learned a ton, both conceptually, how to think about what's going on in the economy. But then there's great history and current examples in terms of how we think about innovation, how we think about the relationship between knowledge and skills, and the long journey from invention to implementation that Jim lays out so well in the book. And just want to thank Jim for the work he did in the book and for being here. Thank you. Thank you. Thank you.