The Bumpy Road
Book Review:
Tim Harford, "Adapt: Why Success Always Starts with Failure", Hachette Digital, 2011.
In my review of The Undercover Economist and The Undercover Economist Strikes Back, I noted that there is also a third Tim Harford book on Conceptually's bookshelf. Indeed, that book is Adapt -- a book about bottom-up trial-and-error approaches to solving complex problems. For an economist who believes in the power of markets, this seems like a logical literary foray. (Recall that, for economists L. von Mises and F. Hayek, the market's price system functions as a decentralized "economic calculator".) However, Adapt is about much more than economics; it touches on a diverse range of issues including terrorism, climate change, innovation, evolution, nuclear accidents, and even art.
In the first of eight chapters, the author talks about how the complexity of the modern world can make it hard to solve problems, even for leaders and experts. Even the largest companies rarely survive more than eight decades -- failure appears to be commonplace in wealthy economies. But overall, the market is so successful exactly because it allows for an evolutionary process wherein many variations are tried out and only the "fittest" are selected. Ironically, people prefer CEOs and politicians not to throw around random ideas or "flip-flop"; yet trial-and-error has a better track record of solving tough problems than expert leadership. The downfall of the Soviet planned economy was its lack of experimentation. People like Peter Palchinsky, who criticized Soviet engineering and suggested reforms, were executed. Fortunately, Palchinsky left us with the following principles:
- Seek out new ideas and try new things [variation];
- Try new things on a scale where failure is survivable [survivability];
- Seek out feedback and learn from your mistakes [selection].
Chapter 2 dives more deeply into how organizations learn, using the U.S. War in Iraq as a case study. Back in 2005-2006, the situation looked dire as marines massacred civilians in Haditha. The secretary of defense at the time, Donald Rumsfeld, had trouble listening to advice from his commanders -- he refused to adapt. But no leader can make the right decision every time, so a hierarchical chain of command can be dangerous, especially when there is pressure to agree. Things started going slightly better for the U.S. military around 2008, after Rumsfeld was replaced by Robert Gates, who appointed people like General David Petraeus. Petraeus rewrote the army's counterinsurgency doctrine to incorporate more diverse opinions, and took advice from Colonel H.R. McMaster, who trained his soldiers to be more respectful towards Iraqi civilians and delegated authority to junior officers on the ground. The lesson was that success comes not from centralized "information dominance" but from local "knowledge of the particular circumstances of time and place".
In addition to resisting conformity, other ways to create new ideas include investing in a variety of bold experiments. Not every project will pay off, so it is sensible to develop several ideas in parallel. Unfortunately, as Harford notes in Chapter 3, the number of patents produced per researcher has been falling over the last few decades. Perhaps patents are not the right solution; they don't seem to encourage large-scale long-term projects such as solar power or HIV vaccines. Government grants aren't a panacea either -- agencies like the National Institutes of Health are reluctant to fund risky projects, even though such research might be more influential. What could work are prizes (e.g. the X Prize Foundation), since they are open to anyone (and don't cost anything until the problem is solved). However, Harford suggests that prizes be combined with grants and patents, because we need a pluralistic approach to stimulating innovation.
Variation is only half the story: there also needs to be a way of selecting what works. The fourth chapter of Adapt examines the challenge of global poverty, and how people like Muhammad Yunus (of Grameen Bank fame) have used trial and error to help the poor. Unlike traditional development aid given by governments and international donors, Yunus takes the "worm's-eye view" by seeing obstacles at close range. Esther Duflo and Michael Kremer among others have advocated for randomized trials in foreign aid. Such experiments can yield interesting results: for example, to boost academic achievement in Kenya, deworming treatments worked better than distributing textbooks or flip charts. But what about questions that are FUQed -- "fundamentally unidentified questions"? Can experiments tell us about complex issues like poverty or corruption? Yes! The economist Benjamin Olken found that anti-corruption audits work better at reducing corruption than village meetings (in Indonesian road projects), while a team led by Macartan Humphreys found that community-driven reconstruction projects work at promoting social cooperation (in war-torn Liberia). You just need the right identification strategy. It is worth mentioning that you also need a way to ensure that the good approaches are actually used, like robust feedback loops (e.g. monitoring). Besides randomized trials, the concept of charter cities (zones in a country with different economic rules) is meant to offer both variation and selection, as they attract bold entrepreneurs while letting citizens "vote with their feet".
What about the great global challenge of climate change? In the next chapter, Tim Harford dispels the notion that climate change is an easy problem. It isn't a matter of sheer willpower -- we first need to figure out the carbon footprint of our behaviors. For instance, a naive environmentalist might suppose that unplugging his phone charger, taking the bus, eating locally-produced beef, and installing a rooftop windmill would be eco-friendly... However, beef has a high environmental impact (better to go vegan) and local products aren't always better (e.g. it takes more fossil fuel to produce lamb and tomatoes in the UK than it does to import them from New Zealand and Spain, respectively). Buses only have lower emissions per passenger mile than cars if they carry enough passengers (which isn't always). A small rooftop windmill generates only 1/5th the power needed for a 50-watt light bulb. And unplugging the phone charger saves only 6 grams of carbon dioxide a day (compared to the 75 g footprint of a packet of potato snacks). Clearly, good intentions aren't enough. Even top-down regulations can backfire, as Harford points out: mandating that buildings or transportation use renewable energy will most likely result in land being used for biomass rather than growing food, and trucks congesting cities to deliver the biomass. Perhaps the most effective strategy is a simple carbon tax or carbon permit scheme -- it uses the power of economic evolution to its advantage.
Chapter 6 deals with another major issue: financial crises. There are parallels to be drawn with oil rig explosions and nuclear meltdowns, since there is little room for trial-and-error, and these systems are complex and tightly coupled (i.e. there are many ways for things to go wrong, and when they do, things can quickly spiral out of control à la domino effect). The psychology and engineering of safety tell us that adding more layers of safety measures isn't always better -- what we need is a clear and intuitive presentation of information, plus a way to loosen the coupling of the system. For example, banks should be allowed to go bankrupt without toppling other banks. Harford offers a couple of ways of decoupling: require banks to hold a bigger capital cushion; use "contingent convertible" bonds (which turn creditors into shareholders in the event of a crash); insist that banks file contingency plans; and split so-called casino banking from utility banking. We should also empower employees to spot latent error and fraud (apparently they do so better than financial regulators). The aim should be to get more reliable indicators of what's going on, and make failure survivable.
The last two chapters are about adapting in the contexts of business and personal life. As Harford points out in Chapter 7, an individual can succeed without adapting by being at the right place at the right time. But in principle, one can apply the same Palchinsky principles in everyday life: try new things, make sure failure doesn't ruin you, and learn from your failures. In organizations, letting small teams make decisions and promoting strict transparency are new management practices that help companies adapt quickly to local circumstances. The idea of "peer monitoring" is for employees to win the trust of their colleagues and double-check each other's work. Google gives its engineers slack to spend 20% of their time on any project that seems worthwhile -- this has helped generate several successful products (e.g. AdSense) but also many failures. Other companies (ranging from supermarkets to publishers and pharmaceutical companies) conduct experiments all the time, sometimes emulating Lockheed's "Skunk Works" (a division that is somewhat isolated from the rest of the corporation, sheltering them from vested interests). Unfortunately, disruptive innovations are often smothered because senior executives fear losing status. When a company dies, however, it's arguably not a huge deal, because (as explained in Chapter 1) failure is behind the success of the market system.
Finally, as an individual you can learn and grow by experimenting and failing (privately, if possible). The hardest part is probably to admit failure, since cognitive dissonance makes us prone to denial. Additionally, we find it hard to make peace with our losses (see: the sunk cost fallacy) and we reinterpret them as successes (or bundle losses with gains). To overcome these obstacles, we need a personal "whistleblower" (a good friend who will support you but also give accurate criticism). We need both pluralism and discipline: a combination of wild leaps to find new peaks, and small steps to master what is working. And, as Harford concludes, we need the courage to bear the cost of failure, because a successful experiment can transform our lives for the better "in a way that a failed experiment will not transform them for the worse".
***
Adapt offers a suite of practical tips on learning from failure: empower troops (or workers) on the front lines to make their own decisions, counter conformity, experiment with several ideas in parallel, offer prizes to incentivize innovation, deploy randomized controlled trials, find ways to loosen the coupling in complex systems, reward whistle-blowers, take advantage of community monitoring or peer monitoring, create safe spaces for failure (skunk works), and make peace with your losses. Upon reflection I noticed a few similarities with other books I've read: in Algorithms to Live By, B. Christian and T. Griffiths write about the trade-off between exploration and exploitation; and in Decisive, the Heath brothers advise us to widen our options, to reality-test our assumptions and to prepare to be wrong. These ideas seem to fit with the Palchinsky principles of variation, survivability and selection, and with Harford's call to combine pluralism and discipline.
Of course, Tim Harford also offers more specific advice on global development (try charter cities and randomized experiments), climate change (carbon pricing), and financial crises (capital cushions, CoCo bonds, contingency planning and narrow banking). As an economist who has worked for the World Bank Group, these are Harford's areas of specialty. For more of his writing on economics, see The Undercover Economist and its sequel.
As with his other books, the author's writing in Adapt is clear and accessible. And as with those other books, Harford defends free-market economics; but here, the reason for the market's success (decentralized trial-and-error where one firm's failure benefits the system as a whole) is transferred to other areas of life, showing us how economics is more broadly relevant. Additionally, Harford demonstrates how economists can learn from other fields such as safety engineering (e.g. about tightly-coupled systems). Thus, Adapt is an interdisciplinary work, reminiscent of the closing chapter of The Undercover Economist Strikes Back, wherein Harford writes that economists should "remember Bill Phillips's engineer-like mindset and diverse interests".
The elephant in the room is that the premise of Adapt isn't very original. In fact, the notion that success is often preceded by a string of failures is somewhat clichéd. Nonetheless, Adapt is worth reading for its in-depth exploration of the topic replete with case studies, data, and argumentation. While the lack of novelty and some minor issues (e.g. no visual illustrations, subheadings or chapter summaries) held me back from giving the book 5/5 stars, it was still a fun read.
***
"Be brave enough to suck at something new."
-- Found on Reddit
Comments
Post a Comment