Should We Fail Fast, Hard, and Often? Or Think Carefully About Investments?
Many innovation experts argue that failure is an important part of the innovation process and thus innovators should be failing fast, hard, and often. They cite Thomas Edison, Marie Curie, and other great inventors and scientists who failed many times before they succeeded. Edison purportedly said: “I have not failed. I’ve just found 10,000 ways that won’t work.” He continues: “Many of life’s failures are people who did not realize how close they were to success when they gave up.”
But do we remember Edison for these failures, or do we remember him for his successes? After all, he invented electricity, light bulbs, phonograph, film, and movie cameras? If he had not succeeded at these inventions, it is highly unlikely that we would know his failures nor his comments about failure. History is filled with people who failed in their attempts to innovate and they are not remembered because, well, they failed.
One corollary to the argument that we should fail fast, hard, and often, is that we cannot select winning ideas without investing in losing ideas. In other words, the reason for Edison’s success was not because he had good ideas, it was because he tried harder than others. Other corollaries include opportunities are everywhere and experimentation is more important than careful economic analysis.
The reason why these arguments are misleading is the subject of my seventh article in my series on technology, innovation and starutps. In their celebration of failure and bad ideas, innovation experts are misleading potential innovators, particularly, young people. They are encouraging young people to innovate without thinking carefully about their ideas, careful thinking that is needed to reduce the chances of failure. And many of these bad ideas are leading to big failures, for instance, today’s startups are doing far worse than those of the past, trying to commercialize the same idea over and over with apparently little learning occurring between these failures.
What body of knowledge can guide us in our search for good innovations? It is called science and engineering. Science provide us with explanations for physical phenomena and thus can help us separate the good ideas from the bad ideas. For instance, Thomas Edison relied on existing engineering and scientific knowledge to do much of his work, benefiting greatly from Michael Faraday’s advances in electricity and magnetism to develop his system of electricity generation. Marie Curie’s work on radioactivity built from William Roentgen’s work on X-rays and his conclusion that x-rays came from inside the materials, a phenomenon that Marie Curie later explained winning her a first Nobel Prize (with her husband) and later a second one by herself.
This type of engineering and scientific knowledge can still guide us in our searches for good innovations. For example, consider the controversial issues of plastics and sustainability. Chemical engineers know that it takes less energy to produce plastic than glass bottles, and plastic than paper bags. Therefore, to replace plastic bottles and bags with glass and paper ones, we must recycle glass bottles and paper bags many times more than plastic bottles to reduce energy consumption. Similarly, to replace plastic with paper in packaging may reduce the amount of plastic waste, but it will increase the number of trees that are cut down, and the energy associated with processing wood and paper. This engineering and scientific knowledge is unfortunately ignored in a mad dash to eliminate plastic waste, a mad dash that will likely cause the world to use more energy to produce bottles and bags, and to cut down more trees that it is doing now. Why do we need to fail multiple teams and still not understand basic engineering and scientific principles? The mantras of fail fast, hard, and often, and we cannot select winning ideas without investing in losing ideas has created an excuse for not thinking carefully not only at startups and VCs, but also in engineering and business schools.
Similarly, engineers know that hydrogen for vehicles require many more processing steps (electrolysis, compression, fuel cells) than do the batteries for electric vehicles and thus have higher electricity and capital costs than do EVs. Hydrogen must be separated from water in electrolysis (using electricity), then compressed, stored, and used in a fuel cell. In contrast, batteries can be directly charged with electricity. These types of details are ignored by many however, and thus we are currently in a hydrogen boom with more than 30+ startups developing hydrogen solutions. (One benefit of compressed hydrogen is the much higher energy storage density than batteries, making it useful for industrial vehicles and utility storage (a second benefit is high resistance to sub-zero temperatures).
The danger of thinking we must fail fast, hard, and often or we must invest in bad ideas to select good ideas can also be seen in today’s startups. In their celebration of failure, innovation experts have encouraged young people to create a wide variety of startups, the biggest of which involve ride sharing, food delivery, e-commerce, fintech, and cloud computing. These startups are also pushed forward by university entrepreneurship programs that have grown from about 16 in 1970 to 400 in 1995 and more than 2,000 by 2014.
The result of this hype is record high startup formation and venture capital funding but record low startup performance. The five-year period between 2015 and 2019 had the highest venture capital funding of any five-year period for the U.S. and the world. But these startups are doing far worse financially than those founded 20 to 50 years ago , covered in my previous articles for this series. They are less profitable at IPO time than in the past despite increases in the medium time to IPO and the most successful ones (Unicorns) are doing far worse than the most successful ones of the past. The most successful ones of today are taking longer to become profitable or achieve top 100 market capitalization status than those of 20 to 50 years ago, and few have share price increases greater than the Nasdaq.
Behind these poor performing startups is the celebration of failure, bad ideas, experimentation, and the hype of startups and new technologies. Many of today’s biggest loss makers have commercialized innovations that do not increase productivity and the result is that thus startups are unprofitable in every industry except payments. For instance, ride sharing and food delivery startups use the same vehicles, drivers, and roads as did previous services and thus do not have large productivity advantages over the previous services. Moreover, both have and continue to increase congestion, a conclusion found in almost every city the analysis has been done. With dozens of these startups in existence all around the world, no one seems to learn anything from failure.
Other technologies commercialized by startups are also not much better than previous ones. Online sales of juicers, mattresses, and exercise bikes may have been revolutionary 20 years ago, but they are sold in the same way that Amazon currently sells almost everything, and the products are also no different from previous ones. New business software enables more cloud-based work, a big advantage during the lockdowns, but not a huge one during normal times. Fintech startups use algorithms to find low-risk borrowers or insurance purchasers, but the advantages of these algorithms are still small or non-existent. Online education may deliver content differently, but it is the same content and most critiques of education claim that the content is the issue not the delivery method.
These startups, their VCs, and the business schools that trained them should have thought much more carefully about these ideas than they did. Rather than think that it is good to fail fast, hard, and often, or we cannot select winning ideas without investing in losing ideas, they should have been thinking how they can reduce the chance of failure and increase the chance of success. An easy way to do this is to think about the technology, its constraints and design tradeoffs, and drivers of improvements. For ride sharing, transportation experts have known for decades that congestion was the issue to be solved, not replacing the dispatcher with a computer. Experts have also known that food delivery is much different from book delivery because food must be delivered in specific time slots and its temperature and consistency must be carefully managed. For every technology or business, there are constraints and tradeoffs that reflect an underlying economic system that should be understood and analyzed before a design is selected and a business model is devised.
Unfortunately, today’s business schools are not skilled or trained to teach such economics. Their journals emphasize theories not context, unlike the past when for instance the case study system was widely used in business schools. But as publishing papers have become more important, context has been thrown out the window. After all, a professor publishing many papers gets promoted, not the one who teaches well, and thus a successful professor has no time to read newspaper articles, interview managers, or talk with their colleagues about specific industries. It is easy to see why such professors like to believe we must fail fast, hard, and often or we must invest in bad ideas to select good ideas; it justifies their research and poor teaching, completely absolving them of the falling performance of startups and slowing productivity overall. Why bother with learning the context when your theory says context is not important?
More fundamentally, why even bother to acknowledge there is problem with today’s startups? For instance, Harvard Business School’s course on innovation and entrepreneurship, offered in April 2020, did not even acknowledge the huge losses and other problems being experienced by today’s startups. Nor did the course acknowledge the productivity slowdown and instead claimed that “innovation is a burgeoning part of the economy.”
There is another way to teach innovation to students, however, a way that emphasizes the economics of electric and hydrogen vehicles, of plastic vs plastic, of buses vs taxis and ride sharing services, and the overall context of many other technologies and services. I taught a course on the economics of new technologies for seven years at the National University of Singapore, a course that covered these technologies along with many others and analyzed the problems of today’s startups, particularly ride sharing. But for proponents of fail fast, hard, and often, we cannot select winning ideas without investing in losing ideas, opportunities are everywhere and experimentation is more important than careful economic analysis, such courses are a sacrilege, a course that challenges everything they know and do.
 A recent linkedin post made this argument, obtaining almost 1000 reactions
 The Rise and Fall of American Growth, Robert Gordon, Princeton University Press, 2016.