Kevin Maney :
Our brains suck at data. Even worse, our brains often use random data to deceive us and make us do dumb things. Cognitive bias, it’s called. This hot company out of Seattle, Tableau Software, believes it can help people use data to be less stupid. Hard to imagine a technology that could more profoundly affect our daily existence. In this era of Big Data, we have data about everything. The data, though, is typically packed into hard-to-decode Excel spreadsheets or locked in databases that only pros can query. Tableau wants to change that by giving ordinary people data visualization software that’s as easy to use as Facebook, so we can make more decisions based on data instead of on our instincts, which are mostly terrible.
In a way, this is a new twist on an age-old story of human knowledge. Little by little, data has encroached on our cognitive biases. Depending on who’s counting, we have four or five dozen kinds of cognitive biases, including Confirmation Bias (we favour information that confirms our preconceptions, or, as some might call it, “the Fox News Effect”) and the Neglect-of-Probability Bias (we hear of a single dramatic event, like a child abduction, and believe it’s much more common than it is, and then force our kids to stay inside for months, making everyone in the household miserable).
In the absence of data, we believe our cognitive biases. Centuries ago, people looked at the sky and concluded that the Earth was the center of the universe. Data about the movement of planets, most famously collected by Galileo, eventually eroded that false belief-although cognitive bias was so powerful that instead of celebrating Galileo for his insight, officials put him under house arrest. In the fight against cognitive bias, data is the underdog. Over time, we’ve collected data about more and more of our world, and invented tools like computers to make sense of it. Now data is exploding like never before. Cellphones and the Internet have a lot to do with this. We’re moving our lives online-shopping, socialising, dating, reading, making restaurant reservations, hailing cabs. That means more of our behaviour is tracked and turned into data. We’re also deploying gazillions of sensors that record data about traffic, weather, chemicals in the air, water in the ground, thongs in Brazil or anything else we might want to measure. We’re even wearing sensors so personal, they turn our heart rates and sweat production into data.
Yet it’s amazing how relatively little of this data infiltrates our cognitive biases. Take one simple example. In May, an Amtrak train derailed in Philadelphia, killing eight passengers. Since then, the media and politicians have focused on ways to make trains safer from derailments. But that’s the Neglect-of-Probability Bias at work-a dramatic incident made us fear that derailment is a common problem on Amtrak. It’s not. For Amtrak, the No. 1 cause of accidents and deaths since 2010 has been trains hitting cars or people at rail crossings. That’s where attention and solutions should be focused. The Amtrak data is not affecting our thinking because it is buried in a Department of Transportation spreadsheet that would make most people’s eyes spasm. Ordinary people can’t use it. But Tableau’s software was able to suck data out of the complicated spreadsheet and present it as interactive pictures: maps and graphs that can reveal trends and patterns with a mouse click or two. That’s the promise of data visualisation. This is not trivial technology. Tableau grew out of Ph.D. research into data visualisation at Stanford University. Nothing quite like it existed before, and a dozen years after Tableau was founded, it’s still a work in progress.
The learning curve for Tableau’s tools is steeper than, say, learning PowerPoint. Tableau’s revenues have been doubling about every 18 months, and the company is now worth more than $8 billion. And yet, as CEO Christian Chabot tells me, “we probably have reached less than 1 per cent of the people who can benefit from our products. We’ve barely made a dent so far.”
Chabot imagines a world where we check our cognitive biases with data all the time, simply and easily, much the way you look at the GPS map on your smartphone instead of guessing where you are. Think of all the decisions you now make in a data vacuum. If you have that next bourbon, what’s the relative trade-off between how much fun you’ll have tonight versus how much you’ll suffer in the morning? If you take this job, will you be happy in a year? If you drink a 5-Hour Energy, will you personally get five hours of energy?
One McKinsey study showed that when businesses worked at reducing the effect of cognitive bias in decision making, they achieved returns up to 7 percentage points higher. In sports, the potential impact seems obvious. Moneyball showed how professional baseball was steeped in intuition that turned out to be wrong. NFL coaches routinely punt on fourth down even though data conclusively proves that teams would win more if they never punted. We’ll eventually have data that can help with any kind of decision, and it should make us smarter, more logical, more grounded in reality-as a species, less stupid.
But then there’s the question of whether it will make us less human. If we compiled data about all decisions in history, it would conclusively show that cognitive bias has influenced far more of them than data. We’ve gotten this far, and created our data-rich environment, by mostly making ill-informed calls using imperfect intuition. So if we were to make a data-enhanced decision about whether to make more data-enhanced decisions, we might find that we should hold tight to our cognitive flaws.
Unless my thinking about all this is hopelessly biased.
Our brains suck at data. Even worse, our brains often use random data to deceive us and make us do dumb things. Cognitive bias, it’s called. This hot company out of Seattle, Tableau Software, believes it can help people use data to be less stupid. Hard to imagine a technology that could more profoundly affect our daily existence. In this era of Big Data, we have data about everything. The data, though, is typically packed into hard-to-decode Excel spreadsheets or locked in databases that only pros can query. Tableau wants to change that by giving ordinary people data visualization software that’s as easy to use as Facebook, so we can make more decisions based on data instead of on our instincts, which are mostly terrible.
In a way, this is a new twist on an age-old story of human knowledge. Little by little, data has encroached on our cognitive biases. Depending on who’s counting, we have four or five dozen kinds of cognitive biases, including Confirmation Bias (we favour information that confirms our preconceptions, or, as some might call it, “the Fox News Effect”) and the Neglect-of-Probability Bias (we hear of a single dramatic event, like a child abduction, and believe it’s much more common than it is, and then force our kids to stay inside for months, making everyone in the household miserable).
In the absence of data, we believe our cognitive biases. Centuries ago, people looked at the sky and concluded that the Earth was the center of the universe. Data about the movement of planets, most famously collected by Galileo, eventually eroded that false belief-although cognitive bias was so powerful that instead of celebrating Galileo for his insight, officials put him under house arrest. In the fight against cognitive bias, data is the underdog. Over time, we’ve collected data about more and more of our world, and invented tools like computers to make sense of it. Now data is exploding like never before. Cellphones and the Internet have a lot to do with this. We’re moving our lives online-shopping, socialising, dating, reading, making restaurant reservations, hailing cabs. That means more of our behaviour is tracked and turned into data. We’re also deploying gazillions of sensors that record data about traffic, weather, chemicals in the air, water in the ground, thongs in Brazil or anything else we might want to measure. We’re even wearing sensors so personal, they turn our heart rates and sweat production into data.
Yet it’s amazing how relatively little of this data infiltrates our cognitive biases. Take one simple example. In May, an Amtrak train derailed in Philadelphia, killing eight passengers. Since then, the media and politicians have focused on ways to make trains safer from derailments. But that’s the Neglect-of-Probability Bias at work-a dramatic incident made us fear that derailment is a common problem on Amtrak. It’s not. For Amtrak, the No. 1 cause of accidents and deaths since 2010 has been trains hitting cars or people at rail crossings. That’s where attention and solutions should be focused. The Amtrak data is not affecting our thinking because it is buried in a Department of Transportation spreadsheet that would make most people’s eyes spasm. Ordinary people can’t use it. But Tableau’s software was able to suck data out of the complicated spreadsheet and present it as interactive pictures: maps and graphs that can reveal trends and patterns with a mouse click or two. That’s the promise of data visualisation. This is not trivial technology. Tableau grew out of Ph.D. research into data visualisation at Stanford University. Nothing quite like it existed before, and a dozen years after Tableau was founded, it’s still a work in progress.
The learning curve for Tableau’s tools is steeper than, say, learning PowerPoint. Tableau’s revenues have been doubling about every 18 months, and the company is now worth more than $8 billion. And yet, as CEO Christian Chabot tells me, “we probably have reached less than 1 per cent of the people who can benefit from our products. We’ve barely made a dent so far.”
Chabot imagines a world where we check our cognitive biases with data all the time, simply and easily, much the way you look at the GPS map on your smartphone instead of guessing where you are. Think of all the decisions you now make in a data vacuum. If you have that next bourbon, what’s the relative trade-off between how much fun you’ll have tonight versus how much you’ll suffer in the morning? If you take this job, will you be happy in a year? If you drink a 5-Hour Energy, will you personally get five hours of energy?
One McKinsey study showed that when businesses worked at reducing the effect of cognitive bias in decision making, they achieved returns up to 7 percentage points higher. In sports, the potential impact seems obvious. Moneyball showed how professional baseball was steeped in intuition that turned out to be wrong. NFL coaches routinely punt on fourth down even though data conclusively proves that teams would win more if they never punted. We’ll eventually have data that can help with any kind of decision, and it should make us smarter, more logical, more grounded in reality-as a species, less stupid.
But then there’s the question of whether it will make us less human. If we compiled data about all decisions in history, it would conclusively show that cognitive bias has influenced far more of them than data. We’ve gotten this far, and created our data-rich environment, by mostly making ill-informed calls using imperfect intuition. So if we were to make a data-enhanced decision about whether to make more data-enhanced decisions, we might find that we should hold tight to our cognitive flaws.
Unless my thinking about all this is hopelessly biased.
