So many companies play defense. Cut costs, watch the competition, follow best practices. Great entrepreneurs like Jeff Bezos and Elon Musk play offense. They see the world with fresh eyes, taking off the blinders that keep companies using technology to make slight improvements to existing products and practices, rather than imagining the world as it could be, given the new capabilities that technology has given us.
I can think of no clearer example of the failure of imagination by traditional companies than the idea that robots and AI are going to put humans out of work.
My new book WTF? talks about what the great technology platforms have to tell us about the future of business and the economy. How is work changing?What does technology now make possible that was previously impossible?What work needs doing?How do we make the world prosperous for all?Why aren’t we doing it? And what are some of the key skills we need to master.
One of the key ideas in the book is that our maps of the world are steering us wrong. In 1625, Henry Briggs published a map stolen from the Spanish by the Dutch, which showed California as an island. This map guided exploration for the next hundred years.
Henry Briggs, 1625: “California sometymes supposed to be a part of ye westerne continent, but since by a Spanish Charte taken by ye Hollanders it is found to be a goodly land…” In 1705, a Jesuit priest, Eusebio Kino, led an overland expedition across the top of the Sea of Cortez, and argued that what came to be called Baja California was in fact an island. My question is why it took an overland expedition, rather than just sailing up the Sea of Cortez!
One of the bad maps that we have in economics and business is that growth goes on forever. In Silicon Valley, there’s often talk of “exponential growth,” but it’s pretty clear that even Facebook is showing only linear growth. But even linear growth runs up against limits. Facebook is at 2 billion users, a third of the world’s population.
I think the idea from evolutionary biology, about fitness landscapes with peaks and values, is actually a better way to think about growth. In my career, I’ve watched a number of migrations to new peaks, and I’d like to share with you some observations about what happened, and why. And then we’ll talk about some lessons for digitalization of the overall economy.
When a new wave of technology hits, a new company almost always becomes dominant. The dominant company of one technology wave sometimes manages to survive, but it loses its privileged position as the technology marketplace migrates to a new peak. The path to the top of each new peak requires new competencies – a new fitness function – and the old competency actually holds back the previously dominant company.
One of the things that I’ve learned is that the surest way to drive entrepreneurs to seek the fitness peak of a new technology and a new business model is for dominant players to take too much of the value for themselves. And just as in biology, it’s easier to get to the new peak from the valley. I watched this happen with Microsoft in the 1990s. The company had used its dominance over the operating system to lock out competitors. But the innovators just went elsewhere, where there was an opportunity for open innovation, and invented the future on the way up a new fitness peak. Tim Berners-Lee introduced the World Wide Web in 1990, and Linus Torvalds introduced Linux in 1991. Between the two of them, the paradigm changed. Software was now a commodity. Big data was the new source of competitive advantage, with Google at the latest peak in the fitness landscape.
Net lesson: You lose when you try to capture too much of the value for yourself. And you lose again if you hang on to the old rules of business when faced with the resulting change in the fitness landscape.
Google demonstrated that the best way to bridge from one fitness peak for another is to give away value to others. We can see this with the story of both Android and Chrome. They didn’t try to capture all the value for themselves, but gave it away to device manufacturers and network providers.
So we see both in coming up from the valley at the bottom, or by trying to extend from the top that generosity is the best strategy.
OK. Next topic. Another big idea in the book is that we are thinking wrong about AI. AI is not separate from us. We are all living and working inside a machine. It isn’t just this worker in a Google data center.
Our modern systems are massive hybrid AIs. These Ais are not external to us. We are part of them. We are inside them. They shape what we think and how we act.
When you look at a company like Google, you see that humans are working alongside automation in very new ways. Even in a company as driven by computer technology as Google, there are humans who keep things running. There are other humans who write code and AI models, and manage and train the algorithms of search, advertising, and the Google Brain. There are other humans – all of us - who contribute new knowledge and seek it out, reinforcing neural pathways by what we link to, and what we pass on.
This has implications that we are only now starting to become aware of.
So that brings me to this idea of Uber as a networked platform. The the digital world is no longer separate from the physical world. Uber drivers and passengers are all part of a vast digital machine.
There’s one other of these hybrid proto-Ais to consider, and that’s our financial markets. And that’s where we should be worrying about Skynet, that fabled AI gone wrong, hostile to humans. Like Google and Facebook and Twitter, our financial market is a composite organism made up of its human microbiome, which shapes its behavior, combined with machines driven by encoded objectives.
A business model is the way that all the parts of a business work together to create marketplace advantage and customer value.
So if you look at the business model of Uber or Lyft, you see that there is:
A magical app that lets drivers and passengers find each other in real time A networked marketplace of drivers and passengers Augmented workers able to join the market as and when they wish All managed by algorithm
All of these pieces come together in 21st century platforms.
Managed by algorithm is one of the key ideas. What does that mean?
I like to show people this side by side picture of workers in a Victorian sweatshop with a group of programmers at Pivotal. Google chief economist Hal Varian once said “My grandfather wouldn’t recognize what I do as work.” So he says! I say “The more things change, the more they stay the same!”
But I’m only joking. Once again, there’s a bad map here. If you look at those programmers with a 20th century mindset, you think they are cranking out software the way that those workers were cranking out clothing or manufactured goods. But with a 21st century eye, you realize that the actual workers at Google or Facebook or Amazon are programs. Those programmers are their managers. Each day they examine the performance of their workers and give them feedback to improve their performance. Eric Ries calls this the “Build. Measure. Learn” cycle.
Now here’s the thing. All these modern algorithmic software systems have “objective functions,” something they are relentlessly optimizing. Uber and Lyft optimize for passenger pickup time. Both of them are trying to create a matching marketplace in which passengers will find drivers within three minutes. Google optimizes for relevance in search results and ads, using hundreds of different algorithmic systems and AI to deliver results that people will be satisfied with. Facebook deploys its algorithms to find content that its users will find engaging, that they will spend time with and want to share with their friends.
We don’t have to wait for a future AI though to see how objective functions can go awry. I believe that Facebook’s struggle with fake news is a great example of the runaway objective function. Facebook told their systems to optimize for engagement – to show people more of what they liked, commented on, and shared. Their idea was that this would lead to more human connection. It turned out instead to increase hyperpartisanship and to drive people apart, and now they are trying to stop it.
Facebook’s engineers are a bit of the same situation as Mickey Mouse in Walt Disney’s retelling of Goethe’s story The Sorcerer’s Apprentice. Mickey borrows his master’s spellbook, and compels the broom to help him fetch water. Unfortunately, he doesn’t know how to stop the broom, and before long
He is desperately trying to find a way to stop the power he has unleashed. This is what Mark Zuckerberg and team look like right now. That’s a runaway objective function at work.
Elon Musk’s fears about AI are based on this same idea of the “runaway objective function.” Every AI and every machine learning system has something it is trying to optimize for. And it can be single-minded in pursuit of that objective. Elon explained this well in a Vanity Fair interview: “Even robots with a seemingly benign task could indifferently harm us. ‘Let’s say you create a self-improving A.I. to pick strawberries,’ Musk said, ‘and it gets better and better at picking strawberries and picks more and more and it is self-improving, so all it really wants to do is pick strawberries. So then it would have all the world be strawberry fields. Strawberry fields forever.’ No room for human beings.”
(Nick Bostrom first articulated the idea of the runaway optimization of an objective function in the context of AI with the thought experiment of a self-improving AI that had been given the goal of maximizing paperclip production.)
And what is the objective function of our financial markets? When, in 1970, Milton Friedman said that the social responsibility of business is to increase its profits, and when, a few years later, Michael Jensen began to preach the gospel of shareholder value maximization and the need to align executive compensation with rising stock prices, they didn’t mean to create the devastation they wreaked on the economy, but it’s time to recognize it.
(Milton Friedman penned an op-ed in the New York Times arguing that the social responsibility of business was to increase its profits. Anything else was, in effect, taking money from its shareholders. Then in 1976, William Meckling and Michael Jensen wrote a paper outlining the reasoning behind aligning the interests of management with shareholders, which was eventually accomplished with executive pay via stock options. So called “shareholder value” thinking was soon taught in business schools, and that’s when the great divergence between productivity and wages began.
My late friend Andrew Singer gave me a wise piece of advice many, many years ago, which remains as true in the days of AI as it was in the early days of Macintosh programming, when he said it to me. “The art of debugging is figuring out what you really told your program to do rather than what you thought you told it to do.” Just as Facebook is having to grapple with this question, we need to grapple with it as well with regard to our entire financialized economy, which is driving us towards inequality.
In this regard, a recent study by the BCG Henderson Institute used machine learning to look at financial reports from tens of thousands of companies, and based on the language they used, put them into two buckets, characterized by Aristotle’s two branches of economics: oikonomia, the management of the household, and chrematistike, the pursuit of wealth. They discovered that those who pursued oikonomia – thinking of their stakeholders more broadly – actually outperformed those practicing chrematistike.
This is also the conclusion of Oxford developmental economist Kate Raworth, author of the book Doughnut Economics, who talks about the job of economics not being how to keep growth going up and to the right, but instead about how to keep the world in “the doughnut,” the narrow band between human undershoot, in which much of humanity is left out of the bounty that is possible, and economic overshoot, where humanity takes an unsustainable proportion of the world’s resources, leading to instability.
She also roots this idea in Aristotle’s notion that there are two ways to think about economics: oikonomia, the management of the household, and chrematistike, the pursuit of wealth. Image: Kate Raworth and Christian Guthier/The Lancet Planetary Health
Take a look at the divergence of productivity and real median family income! Why do we see that, despite the continuing growth of productivity, family incomes have stagnated, and as Raj Chetty’s research has shown, most children in developed countries can no longer expect to do better economically than their parents. Inequality has skyrocketed. Chrematistike has triumphed over oikonomia, creating societal deficits in many of the areas that Kate Raworth calls out – both in undershoot and overshoot.
Source http://stateofworkingamerica.org/charts/productivity-and-real-median-family-income-growth-1947-2009/ via https://en.wikipedia.org/wiki/Income_inequality_in_the_United_States
In this regard, I’m in love with the title of this book, which seems to beautifully sum up oikonomia. Alvin Roth got his Nobel prize in economics for exploring how to design better marketplaces. He worked on kidney transplant marketplaces, and he showed that if you can increase trust, you can create a “thicker marketplace,” in which it is easier to match up those who have something and those who need it.
Hal Varian’s former protege Jonathan Hall, now chief economist at Uber, pointed me to this book, and said it was really shaping how he approached his job at Uber. That gave me a language for thinking and talking about platforms that I hadn’t had before.
One of the fundamental ideas in free market economics is that independent actors, each seeking their own self interest, find areas of mutual exchange and advantage. The idea is that this exchange is the surest way to motivate the best allocation of society’s resources towards society’s needs. Critiques of this approach point out that not everything has a price, that there are market failures, that there are asymmetries of power, and so on, and that perfectly competitive markets only exist in the minds of economists. And of course, the market is not at all free. There are all kinds of hidden rules that we take for granted that shape markets. We have tended to understand that governments are in effect the shapers of their economies, in ways large and small.
But set against this, in my mind, is the notion that large swaths of the modern economy are dominated by platforms that, in effect, are massive privately controlled markets. They are managed by the platform owner, which uses complex algorithms to decide “who gets what and why.”
It seems to me that the great opportunity of the 21st century is to use our newfound cognitive tools to build better marketplaces
Google, Amazon, Facebook … the promise of telemedicine.
The great opportunity of the 21st century is to use our newfound cognitive tools to build better marketplaces
Think about the way that Google search uses hundreds of factors to determine what results come up in response to billions of daily searches – and that price, traditionally thought to be the coordinating hand of markets – play no role. There is a priced sidecar market based on advertising, but even there, in a pay-per-click world, it is not price by price as modified by the predicted willingness of the user to click, that determines which ads come up first. So in effect, Google’s algorithms are performing the coordinating role. What’s particularly interesting is that the Google Brain, Google’s deep learning system, is now the third most important factor in determining search results. And its choices are made for reasons even its creators don’t always understand.
Amazon has over a billion SKUs, and again, the products that come up first in search are influenced by the quality of information about the product, reputational factors such as user reviews and ratings, as much as they are by price. This is of course true in physical world markets as well, but without the algorithms, it would never be possible to sift through such a massive catalog.
And a lot of my advocacy with Uber and Lyft has been around whether they are properly balancing the benefits to passengers, to drivers, and to society (considering externalities like traffic congestion), and whether, like Google, they should be moving away from algorithms that optimize for passenger pickup time and low prices to take other factors more heavily into account.
Might some future AI and big data system do a better job than our traditional markets of the job of allocating society’s resources and distributing the gains? It’s pretty clear that the market, as currently operating, is not doing an optimal job.
The time is now to bring into being the world John Maynard Keynes envisioned in “Economic possibilities for our grandchildren.” a visionary essay he wrote at the outset of the Great Depression. “how to use [our] freedom from pressing economic cares, how to occupy the leisure, which science and compound interest will have won for [us], to live wisely and agreeably and well.” Keynes was wrong, as visionaries often are. He was too early. But he was right.
Keynes was wrong, as visionaries often are. He was too early. But he was right. The time is now to bring into being the world he envisioned in “Economic possibilities for our grandchildren.” “how to use [our] freedom from pressing economic cares, how to occupy the leisure, which science and compound interest will have won for [us], to live wisely and agreeably and well.”
The point is that The fundamental economic questionis no longer how to incentivize productionbut how to incentivize fair distributionof the fruits of increased productivity https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/where-is-technology-taking-the-economy
There’s plenty to go around. It’s just not going around.
I believe that it is the result of our economic objective function gone awry. Our politicians and our businesses bought into an economic theory that said that if we optimized relentlessly for shareholder value, it would be good for the economy as a whole. It turned out not to be true. So just as the Facebook engineers are trying to re-engineer their algorithms, we need to re-engineer the economic algorithms that underlie and shape our markets, giving us outcomes that are not those that we really want!
The last topic I want to address is the idea that the robots are going to take away all human jobs.
We’ve seen calls for Universal Basic Income, with the assumption that there will be nothing left for humans to do once corporations outsource all the work to machines.
Now, as I’ve just noted, I’m a fan of making sure that everyone has enough to live a good life, but I’m really appalled by the idea that we need UBI because there will be nothing left for humans to do!
This is a great example of TS Eliot’s marvelous line from Murder in the Cathedral: This last temptation is the greatest treason: to do the right deed for the wrong reason.
Here’s what actually happened when Amazon added 45,000 robots to their warehouses, they added more than 250,000 human workers. The human workers are part of a complex ballet of human and machine, programmers and warehouse workers and delivery drivers, websites and robots, all coordinated by algorithms to work with uncanny speed and precision, delivering many products within a few hours in the luckiest zip codes. Why was this? Amazon didn’t just use the robots to do the same thing more cheaply. They packed more products into the warehouses, and used the partnership of humans and machines to get them out more quickly, so that in some zipcodes, you can get products the same day.
Jeff Bezos calls this the flywheel. Lower costs lead to lower prices, which lead to more customers, which draws more sellers, offering a greater selection, which leads to better customer experience and more economic activity in a virtuous cycle. This has been true as long as market economies have been around. But you have to work at speeding up the flywheel, like Amazon does.
This isn’t just true in the narrow world of e-commerce.
There’s no work left for humans? Seriously. WTF?
There’s so much work to be done!
Dealing with climate changeRebuilding our infrastructureFeeding the worldEnding diseaseResettling refugeesCaring for each otherEducating the next generationEnjoying the fruits of shared prosperity
Natasha Iskander of NYU calls all of these things “biophilic work,” the work of improving and remediating life. This will always be with us.
We are a social species. Doing things for each other is in our nature. As long as there’s a reasonably fair distribution of the fruits of productivity, we will make an economy for each other. As Clay Christensen noted in his “Law of Conservation of Attractive Profits,” whenever one thing becomes a commodity, something else becomes valuable.
But meanwhile, we have vast problems to solve, the result of the extractive economy of the past.
Fortunately, technology can help. So let’s go to work solving new, hard problems. As my friend Nick Hanauer put it “Technology is the solution to human problems. We won’t run out of work till we run out of problems.” Are we done yet? Are we done yet? I lthink often of Larry Summers’ laconic refutation of the Efficient Markets hypothesis: “There are idiots. Look around.” My refutation to the end of work doomsayers is similar: “There is lots of work to be done. Look around!”
Why can we only see AI and other WTF? Technologies of the 21st century as engines of disruption and destruction, rather than as engines of creativity and prosperity? Why aren’t we talking about universal basic income as our birthright – the result of human ingenuity?
Let the machines do as much of the work as they can. Let humans get on with the real work of the 21st century.
Thank you very much.
There’s plenty to go around.
The Real Work of the
21st Century Tim O’Reilly @timoreilly oreilly.com wtfeconomy.com Harvard Business School March 28, 2018
How is work changing? What
does technology now make possible that was previously impossible? What work needs doing? How do we make the world prosperous for all? Why aren’t we doing it? wtfeconomy.com
We have to let go
of the maps that are steering us wrong In 1625, we thought California was an island
Technology has a fitness landscape
In my career, I’ve watched a number of migrations to new peaks, and I’d like to share with you some observations about what happened, and why. And then we’ll talk about some lessons for companies like Google, but also for the overall economy. Personal Computer Big Data and AI Smartphones Apple
A Business Model Map of
Uber A magical app that lets drivers and passengers find each other in real time A networked marketplace of drivers and passengers Augmented workers able to join the market as and when they wish Managed by algorithm
Algorithmic systems have an “objective
function” Google: Relevance Facebook: Engagement Uber and Lyft: Passenger pick up time Scheduling software used by McDonald’s, The Gap, or Walmart: Reduce employee costs and benefits
The runaway objective function “Even
robots with a seemingly benign task could indifferently harm us. ‘Let’s say you create a self-improving A.I. to pick strawberries,’ Musk said, ‘and it gets better and better at picking strawberries and picks more and more and it is self-improving, so all it really wants to do is pick strawberries. So then it would have all the world be strawberry fields. Strawberry fields forever.’ No room for human beings.” Elon Musk, quoted in Vanity Fair https://www.vanityfair.com/news/2017/03/elon-musk- billion-dollar-crusade-to-stop-ai-space-x
“Economic Possibilities for Our Grandchildren”
The world of his grandchildren—the world of those of us living today— would, “for the first time . . . be faced with [mankind’s] real, his permanent problem—how to use his freedom from pressing economic cares, how to occupy the leisure, which science and compound interest will have won for him, to live wisely and agreeably and well.” John Maynard Keynes John Maynard Keynes
Dealing with climate change Rebuilding
our infrastructure Feeding the world Ending disease Resettling refugees Caring for each other Educating the next generation Enjoying the fruits of shared prosperity
This is what technology wants
“Prosperity in human societies is best understood as the accumulation of solutions to human problems. We won’t run out of work until we run out of problems.” Nick Hanauer
Let the machines do as
much of the work as they can. Let humans get on with the real work of the 21st century.
Tim O’Reilly Founder &
CEO, O’Reilly Media Partner, O’Reilly AlphaTech Ventures Board member, Code for America Co-founder, Maker Media @timoreilly • O’Reilly AI Conference • Strata: The Business of Data • JupyterCon • O’Reilly Open Source Summit • Maker Faire • Foo Camp • … • 40,000+ ebooks • Tens of thousands of hours of video training • Live training • Millions of customers • A platform for knowledge exchange • Commercial internet • Open source software • Web 2.0 • Maker movement • Government as a platform • AI and The Next Economy