We Need To Stop Worshiping Algorithms

In 1954 the economist Paul Samuelson received a postcard from his friend Jimmie Savage asking, “ever hear of this guy?” The ”guy” in question was Louis Bachelier, an obscure mathematician who wrote a dissertation in 1900 that anticipated Einstein’s famous paper on Brownian motion published five years later.

The operative phrase in Bachelier’s paper, “the mathematical expectation of the speculator is zero,” was as powerful as it was unassuming. It implied that markets could be tamed using statistical techniques developed more than a century earlier and would set us down the path that led to the 2008 financial crisis.

For decades we’ve been trying to come up with algorithms to help us engineer our way out of uncertainty and they always fail for the same reason: the world is a messy place. Trusting our destiny to mathematical formulas does not eliminate human error, it merely gives preference to judgements encoded in systems beforehand over choices made by people in real time.

The False Promise Of Financial Engineering

By the 1960s a revolution in mathematical finance, based on Bachelier’s paper and promoted by Samuelson, began to gain momentum. A constellation of new discoveries such as efficient portfolios, the capital asset pricing model (CAPM) and, later, the Black-Scholes model for options pricing created a standard model for thinking about economics and finance.

As things gathered steam, Samuelson’s colleague at MIT, Paul Cootner, compiled the most promising papers in a 500-page tome, The Random Character of Stock Market Prices, which became an instant classic. The book would become a basic reference for the new industries of financial engineering and risk management that were just beginning to emerge at the time.

However, early signs of trouble were being ignored. Included in Cootner’s book was a paper by Benoit Mandelbrot that warned that there was something seriously wrong afoot. He showed, with very clear reasoning and analysis, that actual market data displayed far more volatility than was being predicted. In essence, he was pointing out that Samuelson and his friends were vastly underestimating risk in the financial system.

Leading up to the Great Recession, other warning signs would emerge, such as the collapse of LTCM hedge fund in 1998 and of Enron three years later, but the idea that mathematical formulas could engineer risk out of the system endured. The dreams turned to nightmares in 2008, when the entire house of cards collapsed into the worst financial crisis since the 1930s.

The Road To Shareholder Value

By 1970, Samuelson’s revolution in economics was well underway, but companies were still run much as they were for decades. Professional managers ran companies according to their best judgment about what was best for their shareholders, customers, employees and the communities that they operated in, which left room for variance in performance.

That began to change when Milton Friedman, published an Op-Ed in The New York Times, which argued that managers had only one responsibility: to maximize shareholder value. Much like Bachelier’s paper, Friedman’s assertion implied a simple rule-of-thumb with only one variable to optimize for, rather than personal judgement, should govern.

This was great news for people managing businesses, who no longer had to face the same complex tradeoffs when making decisions. All they had to worry about was whether the stock price went up. Rather than having to choose between investing in factories and equipment to produce more product, or R&D to invent new things, they could simply buy back more stock.

The results are now in and they are abysmal. Productivity growth has been depressed since the 1970s. While corporate profits have grown as a percentage of GDP, household incomes have decoupled from economic growth and  stagnated. Markets are less free and less competitive. Even social mobility in the US, the ability for ordinary people to achieve the American dream, has been significantly diminished.

The Chimera Of “Consumer Welfare”

The Gilded Age in America that took place at the end of the 19th century was a period of rapid industrialization and the amassing of great wealth. As railroads began to stretch across the continent, the fortunes of the Rockefellers, Vanderbilts, Carnegies and Morgans were built. The power of these men began to rival governments.

It was also an era of great financial instability. The Panic of 1873 and the Panic of 1893 devastated a populace already at the mercy of the often avaricious tycoons who dominated the marketplace. The Sherman Antitrust Act of 1890 and the Clayton Antitrust Act of 1914 were designed to rebalance the scales and bring competition back to the market.

For the most part they were successful. The breakup of AT&T in the 1980s paved the way for immense innovation in telecommunications. Antitrust action against IBM paved the way for the era of the PC and regulatory action against Microsoft helped promote competition in the Internet. American markets were the most competitive in the world.

Still, competition is an imprecise term. Robert Bork and other conservative legal thinkers wanted a simple, more precise standard, based on consumer welfare. In their view, for regulators to bring action against a company, they had to show that the firm’s actions raise the prices of goods or services.

Here again, human judgment was replaced with an algorithmic approac that led to worse outcomes. Over 75% of industries have seen a rise in industry concentration levels since the late 1990s, which has helped to bring about a decline in business dynamism and record income inequality.

The Chimera Of Objectivity

Humans can be irrational and maddening. Decades of research have shown that, when given the exact same set of facts, even experts will make very different assessments. Some people will be more strict, others more lenient. Some of us are naturally optimistic, others are cynics. A family squabble in the morning can affect the choices we make all day.

So it’s not unreasonable to want to improve quality and reduce variance in our decision making by taking a more algorithmic approach by offering clear sets of instructions that hold sway no matter who applies them. They promise to make things more reliable, reduce uncertainty and, hopefully, improve effectiveness.

Yet as Yassmin Abdel-Magied and I explained in Harvard Business Review, algorithms don’t eliminate human biases, they merely encode them. Humans design the algorithms, collect the data that form the basis for decisions and interpret the results. The notion that algorithms are purely objective is a chimera.

The problem with algorithms is that they encourage us to check out, to fool ourselves into thinking we’ve taken human error out of the system and stop paying attention. They allow us to escape accountability, at least for a while, as we pass the buck to systems that spit out answers which affect real people.

Over the past 20 or thirty years, we’ve allowed this experiment to play out and the results have been tragic. It’s time we try something else.

Greg Satell is a transformation & change expert, international keynote speaker, and bestselling author of Cascades: How to Create a Movement that Drives Transformational Change. His previous effort, Mapping Innovation, was selected as one of the best business books of 2017. You can learn more about Greg on his website, GregSatell.com and follow him on Twitter @DigitalTonto

Like this article? Sign up to receive weekly insights from Greg!

Image by Avi Chomotovski from Pixabay