In this episode, we are joined by Scott E. Page, a Professor of Complex Systems, Politcal Sciences, and Economics at the University of Michigan, Ann Arbor. Scott is an external faculty member at the Santa Fe Institute and is an author and speaker who has worked with Google, Bloomberg, Blackrock and NASA. Today, he discusses his book, The Diversity Bonus.
With me today is Scott Page who is Professor of Complex Systems, Political Science and Economics at the University of Michigan. He’s an external faculty member at the Santa Fe Institute, an author and speaker at Davos and with companies like Google, Bloomberg, Blackrock and NASA. Welcome to the show, Scott.
Thanks, it’s great to be here.
So, Scott, you’ve just published a lovely book called The Diversity Bonus. What is the diversity bonus and why should we be interested in this?
The Diversity Bonus is a book that talks about when you’re working on a complex task predicting problem solving, that sort of thing, that there actually is a bonus and an added value that comes from people thinking about the problem differently, bringing different tools, that sort of thing, so there literally is a mathematical bonus and it’s about how individuals and firms can realize that bonus.
And why does this matter today in 2017?
I think the reason it matters is that the set of problems we face if you look at any organization now, so much of their value comes from finding better solutions to really complex problems or even identifying what the problem is. So, you know people talk a lot about how we get so much data but the thing is that data then just means it enables us, if we can use it correctly, to make better decisions, but it requires going to look at it in lots of different ways, you’ve got to bring a lot of diverse perspectives to it in order to achieve these bonuses.
And I think in the book you refer to the nutritional example, maybe you can just say a bit more about that because I guess that brings the idea to life quite nicely, doesn’t it?
Yeah, I mean one way to think of this if you think about, you know, there’s an obesity epidemic in the United States and you could think, ‘OK, what does it mean, how would we solve that? What are the dimensions in which this plays out?’ right? So, there are genetic contributions, there are things to do with diet, there are cultural influences, there’s the fast food industry, there are the large cokes, there’s infrastructure, there’s public transportation, there are all these things that come into play that are affecting obesity in the United States and the thing is there’s no way any one person can understand all those dimensions and if you thought about, ‘Oh, here’s a policy that we’re going to implement’ in order to just evaluate, just the processes of value in that policy would require really a team of experts with diverse backgrounds, understanding, knowledge bases, that sort of thing, in order to recognize or even identify which policy is going to work and which policy is not going to work.
And yet twenty years ago or whatever it would have just been seen as purely a sort of a nutritional challenge versus a far more multifaceted challenge now, right?
Absolutely. Think about building a building. I’ve worked on some dissertations of people who work in construction management enabling engineering like building a ship and it used to be like, ‘OK, we’re just building a building’, and now that has environmental implications. There are access issues, there is how you deal with the technology, there are just so many dimensions on which to build, what sort of social networks will result as a function of the building structure, right? Things that we just never thought of in the past and so when you think about a design team, an architectural team for a new academic building, office building, I mean the fact that they’d be thinking about physical disabilities, environmental policy, social networks, it’s just astounding and if you don’t think about those things you’re not going to come up with a good building. Those are things you absolutely have to take into account.
And the data, there are lots of data in the book but maybe you could just sort of quantify the evidence that says that there is a diversity bonus. So, for a CEO or a Chairman of a Board listening to this, what’s the business case behind this?
So, the business case of this is fascinating. In the analogy I give all the time, and this works really well on Wall Street, if I’m investing in a portfolio and I have one stock that pays 10% and another stock that pays 6% then I get the average, I get eight percent, and so the reason I have diversity in my investments is to spread risk, right? It’s just hedging, right? When I have everything go bad, if I’ve got everything in one stock, if you have a group of people predicting – and here’s one of the data sets, it’s remarkable, some researchers at Duke looked at 28,000 predictions by leading economists over a forty year period, these are people within the European Union who predict interest rates, that sort of thing, if you take the very best economist she’s 10% better than an average economist in this group of let’s say forty people predicting interest rates in the European Union, so she’s 10% better. The second best is 9% better. If you average the best economist and second-best economist you don’t get 9.5% better, you get 18% better and that’s the diversity bonus. Now you think, ‘Wait a minute, why is it that with the stock portfolios I get the average but with predictions, I get a bonus?’ but the logic goes as follows. With predictions, sometimes the one person is high and the other person is low and then when you average them, they nail it, and so the fact that there’s any negative correlation in the predictions is going to make two economists far, far more accurate than the average of the two economists, so there literally is, mathematically when it comes to prediction, a bonus. We can do the same logic on problem-solving, the same logic on verifying the truth, the same logic on coming up with creative answers like, you come up with a set of answers, I come up with a set of answers and we’ll get the average, we’ll actually get a bonus because we get the union of our answers and we can often recombine them, so depending on the nature of the problem the mathematics differs a little bit, but in the types of high dimensional complex problems people deal with now, diversity literally gives you a bonus. When I talk to more public audiences I kind of describe this as like the introduction of the three-point line in basketball, like those shots are 50% more, and it’s absolutely a mathematical fact. The question is how do you get it and so what the book talks about and what I like to talk about when I go to organizations is, ‘How do we think about achieving the bonus?’
Let’s get to that in a minute but there’s a lovely example of the Netflix improvement of the algorithm which I think brings out some of those points. Maybe you can just give us a summary of that because it was a fascinating case study in the book.
Oh, it’s just a great story, it’s that Netflix had this prediction contest and the idea was, ‘Let’s find the best data mining people in the world to see if anybody can improve our internal prediction software by 10%’. So, the idea is if you rent a movie from Netflix they’ll give you a score, they’ll say, ‘Oh we think you’ll give this new movie about Winnie the Pooh four stars because you have young children and then I’ll give it one star because I don’t let’s say, right? and this thing was called Cinematch. If somebody could beat Cinematch by 10% they’ll give them a million dollars, and the idea was that they were going to find who the best data scientist is. Well, a funny thing happened. The first thing is that it was all teams not individuals, but it was diverse teams that were winning and then no team could get to the 10% and so what happened was teams started merging with other teams that weren’t as good in order to get the 10%, and so this is exactly like the examples given by the economists. You get this team from AT&T, right, Bell Labs, it was the best. They bring in the team that’s third best in order to try and get to 10%, they still can’t, so then they bring in another team that thinks about the problem in a completely different way that’s fifth best and that gets them over the 10%, and what’s fascinating about it is the teams at that they brought in just thought about these movies in very different ways. So, these are the best and brightest people at Bell Labs and there were things that they didn’t think about. My favorite example in that whole narrative in the Netflix prize is the team they brought in last that got over the 10% was this Canadian team and one thing they realized is there are a bunch of movies like, if you know the movie Snakes On A Plane, or essentially any movie starring Will Ferrell, if you rate that movie right after you rented it you’re going to give it five stars, you’re going to say, ‘That was as an awesome movie’ but it’s kind of like drinking three Margaritas or something, if you wait a few days you’re going to be like, ‘Oh, that was a stupid movie’, right? And so there’s almost like, insurance people call it a hazard rate on these ratings and so what’s fascinating about it is that by the end of this tournament, the Bell Labs team merging with these other teams made 800 variables per movie which is just astounding, that’s just incredible diversity, and so when Neil Hunt in his ending quote of this contest says, ‘You know, we kind of started out trying to see the best data scientists and it turned out this was all about diversity not in some touchy-feely-Kumbaya-Oh it’s great to have a diverse team here but like measurable, quantifiable benefits from bringing in people who can think about this problem in different ways and the story of the Netflix prize is one benefit after another after another after another. Actually, there’s a website called Kaggle that runs these prediction contests on a monthly basis and what you see on Kaggle is this Netflix story just repeated time and time again and this is the key insight. If you’re stuck on a hard problem in your agency research part, there’s nobody smarter to call. I mean, Dr Bell, he’s just the smartest guy in the room, there are two years of data, he’s the smartest guy, he can’t go down the hall and say, ‘Hey, Einstein, can you help me with this?’. He is Einstein. What he can do is he can find someone different, that’s actually relatively easy to do and if you find someone different, he can improve.
So, Scott, just to clarify, you touched on it but let’s make it explicit. Diversity, there are two different types of diversity, you’re not just talking here about, I think you’d call it ‘identity diversity’ but by about ‘cognitive diversity’. Can you just say a bit more because I think many listeners might be getting what the economists refer to as diversity fatigue given it’s hitting them all the time in the media, maybe you could say a little bit about that.
Yeah, so, the space I try and carve out in the book is the following. The only way that diversity can produce a bonus is if it comes in some sort of cognitive form, it literally is the difference in how I represent the problem, right? So, people will have on their web page, ‘We value identity diversity’, – external differences – because people bring diverse perspectives but that’s not even clear what that means. What I’m saying in the book is this, is that what I mean by diverse perspectives is a different encoding of the problem so that could come in several forms. It could be that you think in standard Cartesian coordinates like horizontal and vertical displacement and I think in polar coordinates with an angle in a radius, those would be different perspectives and you can mathematically show that if we have different perspectives we’re going to do better at things like solving problems, making predictions, figuring out the truth, that sort of stuff. We also might use different categories or think of different variables, so when I say cognitive diversity I mean literally differences in how we represent the world, I mean different tools that we have, maybe you know dynamical systems, maybe I know agent based modelling, different knowledge bases, right? Even different metaphors that I’ve made apply in some sort of setting, so things like models, tools, perspective, those sorts of things. The only way that someone with identity diversity or not can come in and enable you to find a better solution to a problem is if different ideas enter the room, different ways of thinking enter the room. So, how does identity come into play? Well, identity often correlates with cognitive diversity. There’s no magic that can happen, it’s not like by bringing in some different identity group some magical thing is going to happen and we’re going to get a better solution. What has to happen is that different people bring in different ways of thinking and those different ways of thinking are giving you better solutions.
Yeah, and you touched on earlier the kinds of problems that are best suited for yielding a diversity bonus. What are they and are there certain types of business problems that you really don’t want to tap into this bonus that just gets in the way?
Yeah, I was joking about this the other day, I was giving a talk at the University of Nevada at Reno and I said, ‘If I’d have written this book fifty years ago, it wouldn’t have made any sense at all.’ Now, I mean if you look at the American economy, the British economy, European economy post-WWII, we’re building roads, washing machines, cars, literally like half of the workers are in these kind of routine manufacturing jobs, it’s kind of a command and control economy, it’s not a lot of problem solving. If you go to some of those same firms now, let’s take Caterpillar which makes giant farm equipment, it used to be they had the same size employee base that they had fifty years ago but it used to be most of their employees were working on a line doing routine jobs. Now most of their employees are out solving problems, they’ve got tons of designers, they’re trying to figure out how to build machines that work in all these different climates and different terrains and for different crops and all sorts of stuff, so robots are basically building the tractors now, so where these bonuses matter are on high dimensional complex problems. So, one of the things that I’ve found, one of the great experiences I’ve had over the last decade is I’ve got to go visit all sorts of different organizations and let me give one type of organization that is probably the best at this and this is hospital emergency rooms. So, if you walk in and you’ve twisted your ankle, they kind of do this complexity valued triage – ‘This is pretty simple, it’s a twisted ankle and he’s not going to die from it’ right? So, they give it to someone, they give you to someone and they do some routine thing like wrap it, somebody else checks, they send you home, right? It’s like Six Sigma land, everybody’s happy, you’re fine, right? If you come in and you’re presenting in some weird way, it’s just not clear what’s going on, they may show you, especially at a teaching hospital, you may see six, seven, eight, nine doctors, right? A personal anecdote, one time I somehow drove a piece of pasta that I was scraping out of a pot that came out of the dishwasher all the way down to the bottom of my fingernail.
I shouldn’t laugh, it sounds painful.
Well, it was funny but there’s a question because it’s organic matter that had gone to the dishwasher so is it safe to leave in there or do they have to take off the nail, right? So, I saw six doctors and at first, I thought they were just harassing me for fun because everybody would say, ‘Oh so you’re a professor’. ‘Yes, I’m a professor’. Worst pasta injury ever. But the thing is that if you have something that’s complex they show you to lots of different sets of eyeballs independently and then they’ll come together as a team and they’ll say, ‘OK what do we do here, right?’ and what they don’t want to do is wait and see if they have to cut off the finger, so they end up making a decision, but firms have to do the same thing. You want to think of diversity, cognitive diversity as a strategic asset, you want to think, ‘I need people in-house that understand my problems that care about this, that share a sense of mission, but I don’t want all diversity all the time, I’m not turning the diversity knob to eleven on routine standard stuff, diversity probably kind of gets in the way. It’s the high-value problem-solving prediction, strategic planning stuff where you really need diversity.
Because I mean the other reality, I guess, is it’s not just good enough to fill a room with people with very different perspectives, it is, as it says in the preface, it’s hard work to actually get this group to gel. What have you seen in terms of what good looks like from leaders who understand it but also actually put it into action in their teams?
So, I think what good looks like oftentimes is if you’ve got a shared mission, this works so well. So, if you go to NASA, you know, today I’m going to speak at NOAA which is the weather forecasting service. People just love the weather, right? So, whether you look different, whatever, if you think about this differently and you’ve got good ideas people are cool with that, it’s really easy to bring teams together like, ‘Wow, that’s a new way to think about this and that works, that’s super interesting’ and ‘There’s a new way to graph this, wow, that’s super helpful’ and so there’s just this esprit de corps that just makes it work, everybody sees new ideas as better. I think where it becomes more difficult is if people don’t – let’s take a company, and this is a very well-run company, like Molex which makes computer cables, it’s a fabulous company I think they have to rely on really talented leadership in order to get people to buy into the value of diverse team solving problems because they’re making cables for medical equipment in computers. No one gets up in the morning and says, ‘Yes! I’m going to make cables!’ but you do get up in the morning and this is where Molex is just a fantastic company, you do get up in the morning and say, ‘I’m going to listen to my employees, I’m going to try and build teams that are going to come up with better solutions to our problems that can give us cost savings and how we design new factories, build new machines’, that sort of stuff, but I think that the problems they’re facing are much, much harder than the problems NASA is facing and if you take pharma where I’ve spent a lot of time, that’s kind of in-between. Some parts of pharma are like, ‘We’re working on oncological drugs that are going to cure cancer’ that’s exciting to get up to do but then at the same time it’s incredible, they’ll do some real command and control things like building a new processing facility, new production facilities and that’s kind of command and control, right
Yeah, but it’s so heavily regulated and you lose your freedom to operate it, it nails your company for a long, long time if you can’t keep the asset going. So, you touched on this just now, the idea of companies that are pushing the frontiers of knowledge, this response that they can really harvest this bonus but there are also companies that might have a clear mission even though it’s not particularly exciting but they also manage to tap into this bonus. What about, I mean, are there any particular industries or even geographies where you see this and where people get it, if you like, and take full advantage of this resource, this asset as you say?
It’s funny, I find both Silicon Valley and the government fascinating in this. So, Silicon Valley, I don’t know if you saw, last week there was an article I think it was both in the Journal and The Times about people who do neural network research especially something like neuro capsules and things like this have been being bought up at just huge salaries, a million dollar a year salary to go and basically work for Google, Ford, a handful of places to work on things like self-driving cars and language recognition. So, you can think about this in the following way. There are probably like seven, eight different approaches to AI that were out there. All of a sudden, this one looks like it’s going to work. So, you just have this buying spree where each of these three or four companies get started buying diverse talent within the realm of people who knew that craft. So, what’s interesting is you have this portfolio approach at first, all these companies had a few people trying different stuff, the academies trying all this stuff, now all of a sudden it looks like this is the one that’s going to work. You have, ‘OK, quick let’s hire twenty people who are all going at this in slightly different ways, put them in a room and have them work really, really hard on this and see if we can get a breakthrough. I think that’s just fascinating. The question there is, did they move too soon? Because there’s a certain diversity lost when you suddenly take the leading researcher from the University of Toronto and take the leading researcher from Colorado, take the leading researcher of Oxford and you let them talk to each other. Each one is now in a less diverse but still somewhat diverse silo. Government does the same thing. We have things like DARPA which really, unlike Al Gore, did invent the internet and DARPA which is the intelligence community’s version of that where get them as a regular basis like, ‘Hey, here’s a request for a research proposal for these really out-there things and I’m just trying to find diverse minds to work on stuff’. Pharma, Eli Lilly has this thing called InnoCentive that Procter and Gamble uses as well where if you can’t solve a problem you just put it out there with a wanted poster saying, ‘Ten thousand bucks if somebody can solve it this’, so I think that there are places that are really good at thinking about how do we open up our problems to larger audiences when they get hard.
In terms of some managers and some leaders listening who are thinking, ‘Well what does this mean for me? I’ve got the same employees, I’ve got my team, I’m not going to switch them’, maybe there are some different nationalities represented, how could those individuals, are there things that you’ve seen that are really effective at tapping into the latent diversity in a group.
So, here’s the one thing that when I talk to business leaders all the time, I try to get them to think about avoiding what my friend Cian Levine calls the ‘siren call of sameness’ which is that, ‘he that looks like me and thinks like me is smart’. So, instead of thinking of a person as kind of having an ability between zero and a hundred or something, or an IQ between 80 and 160, you want to think of people as tool boxes and sets of experiences and ways of thinking and if you have some rubric that you apply to people and then assign some score to them implicitly, what often happens is you end up surrounding yourself with people who look and think like yourself and surrounding yourself with those people limits the diversity you have in the room and you just miss things. So, one of the examples I give in the book just briefly is Chipotle which is a fast food restaurant or convenience food restaurant chain, McDonald’s bought a big chunk of them because they didn’t like the McDonald’s people because the McDonald’s people are too corporate, they kind of had this ugly divorce and then right after that what happened is Chipotle just started poisoning people all across America because their supply chain was terrible but they basically kicked all the supply-chain people out because the supply chain people weren’t all organic environmentally. Well, the thing is, I’m an environmentalist too but the thing is, it’s not clear that these engineers thought about the world from an efficiency mindset, they didn’t like that efficiency mindset. Well, you can’t kick those people out of a room or you’re going to start poisoning people because you’re going to lose the quality control that you need so I think that avoiding this siren call of sameness and avoiding this idea of mapping people, picking apart smart people and instead thinking, ‘Am I surrounding myself with interesting people who challenge me?’ One of the other things that creates bonuses, and Catherine Phillips who wrote the afterword for the book talks about is, ‘Just being in the presence of someone different from you causes you to think differently’. So, let’s suppose you’re designing a building, now we add a third member to our team and this person is in a wheelchair. We’re going to think about the building totally differently or this person is blind. I have a friend who is blind who loves architecture and people are confused by that but he’s like, ‘Oh, the sound in here is so great, it feels so great, or it’s so easy to navigate’ Think about designing a building when there’s someone in the room who is blind or a garden when there is someone who is blind. You are going to think of different things as well, so when you think about how you’re populating your teams, if you’re sitting in a room with, you know like we were talking about the Volkswagen leadership team, if you’re sitting in a room with eleven 54 to 57 years old German engineers who all went to the same Max Planck Institute, there’s some stuff you’re going to miss. You might not realize women carry purses, people have dogs, all sorts of things that just never occur to you.
We talk about this kind of corporate monocultures, the bigger your organization the smaller your world, and ‘avoiding the siren call of sameness’ is another way of saying the same thing versus the other end which is a diverse, rich, sustaining ecosystem of multiple perspectives
Right, and there are limits. The thing is, again, if you want someone to say, and this is where I think the diversity fatigue fits in and sets into people, the idea is not that you want all diversity all the time and the idea is that diversity again is this kind of a strategic asset, but you’ve got to have it in-house and you’ve got to think about it like ‘We have a vibrant organization with lots of people who are seeing and experiencing the world in different ways.’ Let’s take the financial crisis of 2008, all these bankers were concentrated in New York, they weren’t aware of how many people had these massive mortgages and how many people were flipping houses and how many people were extended to ten times their income levels because they weren’t interacting with those people. Its kind of like political scientists in the United States were shocked with Trump winning the election and people were shocked by the Brexit vote. That’s because if you’re sitting in Cambridge or you’re sitting in Princeton, New Jersey or Palo Alto, California, sipping a latte, how can you possibly understand what a voter in rural Michigan, rural Iowa, Manchester is thinking about the world? Each one of us is seeing through a very narrow lens, and on high dimensional complex problems you want to have a broader way of thinking. Andrew Haldane, who is the Chief Economist of the Bank of England, he’s doing things like bringing in psychologists and sociologists and people who do natural language processing into this world of financial data within the Bank of England, and look, if you’re hiring one person to think about interest rates you’re not going to hire a sociologist, you’re hiring sixty, and this is just like straight economics. The 59th economist is probably worth not that much at that point, they’re not going to add anything different, they’re just going to nod their head and say, ‘Yeah, markets work. Prices are efficient’ right? I mean you might want to bring somebody else in who says, ‘People play roles’ or ‘People follow the crowd’ just having a sociologist to introduce new ideas or even just to force the economists to make their arguments carefully. Because there’s a sociologist in the room is going to lead you to a better solution.
Yeah, fascinating. Now as you say, I have in mind the scene, have you seen the movie The Big Short?
When they go off to Vegas to do some proper on the ground research and a rather different world they discovered from the one they were existing in Connecticut and that’s exactly the point.
No, it’s exactly true, and the thing is, let’s take the case of political scientists who thought of Trump voters as racist people and then they go out and they meet Trump voters and they find out that they are in interracial families or they’ve adopted children from racial groups and they’re like, ‘OK, wait a minute that doesn’t make any sense.’ But if you’re just sitting from afar looking at correlational data that’s a very different thing than actually going out on the ground having an understanding of the life experiences of these people, and these people are not only voting and buying houses but any product you make, anybody in any business community, any business you’re in eventually that means going to consumers and that means going to consumers and you need to understand those consumers. High tech things as well though, you need to have you – there is new knowledge being produced all the time and if you’re not bringing in new people knowing those new techniques or thinking of people who are in related industries who might have ideas about, ‘Here’s how we solve this’ or ‘Here’s how we’re going at this’, you’re just missing out on ideas, if you miss out on ideas you’re missing out on improvements and the market will eat you up.
Now we talked a little bit about putting together teams and boards, what about the individual contributor in an organization whose work is complex, they’ve got difficult problems to solve, what can an individual do to expose themselves, to increase their own sort of cognitive diversity, and I know I’m kind of bending the rules a little bit because the way you define it is slightly different from that but are there certain things that individuals can do or maybe don’t, but let me be specific, what do you do, Scott, to continue to remain fresh in this area?
Yes, so first off, there’s this fabulous book called Expert Political Judgement by Phil Tetlock where he looks at people who are what he calls hedgehogs, single model thinkers, they think markets always work or they think they’re socialist, they get a single view of the world and he looks at literally ten thousand predictions over several decades by people and he finds that these hedgehogs are actually worse than random. So, if you just have one way of looking at the world, you’d be better off flipping a coin and randomly guessing.
This is the ‘man with a hammer’ syndrome that we talk about, right?
Yeah, and the people who are actually reasonably good at making forecasts and predictions are people who use lots of models. But in that book he finds out that they also, because they have this kind of Whitman problem, you know, ‘Do I contradict myself while I’m large enough to contain contradiction?’ So, this book that he wrote after that called Super Forecasters, he basically took these people he calls foxes who have lots of ideas and he taught them the basic laws of probability so that their rules would kind of sum to one, and then he shows that they are good forecasters and then teams of these super forecasters are even better forecasters. So, one thing to do is to think about, ‘Am I keeping myself exposed to new knowledge bases, new ways of thinking, in terms of keeping yourself fresh, but the other you can do is – you don’t want every person to necessarily be super diverse if they’re working on a team because then nobody’s deep so what you want to think about is you want to think about, ‘How am I going deep in the thing I’m supposed to be an expert in?’ because here’s where I think it’s kind of misleading. We tend to think that there’s a single path to going deep. One of my favorite charts in the book that I had to fight for like crazy is this wonderful map of the mathematical knowledge based network where you start over at addition and goes to multiplication then division so it starts out looking kind of like a ladder, but then it just branches off to dynamical systems to logic all of a sudden it looks like a giant tree of mathematical knowledge so if you think you’re deep within any one area – let’s just take neuroscience, last year there were sixty thousand papers or seventy thousand papers published in neuroscience, there are many, many ways to go deep so continuing to add knowledge bases even things like – here’s one thing that I do that’s just kind of fun is, I’ll randomly in an area, for example, I was speaking to weather forecasters today so in the last week I just started looking at weather forecasting journals and just reading abstracts of these journals and then downloading some PowerPoint presentations, watching PowerPoint presentations of leading researchers that do weather forecasting so not TED talks because those are kind of one thing but the experts, and you can often glean, here’s how they’re thinking about these particular things. So, in weather forecasting they have these spaghetti graphs where they show all the different paths that the hurricane could take and they’re talking about, what techniques do they use to use these spaghetti graphs in the best possible way, and that gets you thinking, ‘Are there spaghetti graphs in the stuff I do?’ Well, the guys study path dependence, cultural path dependence, is there a spaghetti graph equivalent, right? And that then becomes – so what you think of it, let’s think again with cognitive diversity, perspectives, tools, those sorts of things, the spaghetti graph is a perspective, it’s a way of coding information. What the weather forecasters have done which I’ve learned over the last week and a half just is that they’ve then got a whole bunch of tools they’ve developed to apply to these spaghetti graphs. So, it’s kind of like, can you put your data in a matrix? If you can, you can open a big giant linear algebra book and apply a whole bunch of theorems to it. So, one of the things I think is really cool is with any one sort of way of representing the world or data or a problem, there’s an associated set of tools with that and so exposing yourself to different representations, I just find super, super useful for me because it enlarges how I think about the world. Another quick example, the opioid crisis in the United States, it’s like ‘How is this happening, why is this happening, what’s the cause?’ and what’s been fascinating to me is that there aren’t a lot of models, no one’s been able to organize the data very well in terms of what’s really going on and I finally found like last week someone had just this really simple flowchart of the people who have pain, 80% percent go on these opioids. Of those who go on opioids, some percentage quit within three weeks. Those that don’t who stay on, 10% percent of those switch to heroin or something, I’m like ‘10% percent, oh my gosh’. So, it’s kind of like a markup process way of thinking, a systems dynamics way of thinking about the opiate crisis, you know that’s not only way but it’s like wow, that really helped me get an understanding of what’s going on but then it also leads me to think, what other ways might I think about this, right? So, I’m constantly out there on the prowl for thinking of new representations.
And just to be clear for the audience, you don’t need to be a polymath or a deep academic to do this. I mean, this plays out in, I guess, you talked about the US election but this plays out in exposing yourself to completely different perspectives on social media, for instance, switching out your feeds or picking up a completely different journal in a different industry versus having to go deep into some of the subjects you talked about, right?
Right. So, I teach an online course called Model Thinking where now a million people have taken it, it’s crazy but it basically says like, ‘Here’s some very basic models’ like a linear model or, ‘Here’s a basic of a box and arrow model’ or ‘Here’s what a contagion model looks like’ so nothing uses anything more than very simple algebra. My belief is that there’s probably a lingua franca of analytic tools in terms of ‘Can I look at a regression output? Can I understand the link between a straight line and something that has diminishing and increasing returns? Do I know what a threshold effect is? so there’s probably somewhere between twenty and forty, I wouldn’t call them technical terms, but just concepts like feedback loop, linear, diminishing returns, network, things like that, degree of a network, you can’t bluff and if you know those things you can read almost anything. So, one of things I’m trying to do in this model thinking course, but it’s not just me, a wonderful book by Anne-Marie Slaughter called The Chessboard & The Web about international relations and she said, ‘You can’t do international relations unless you understand some basic notions of what a graph is, what centrality means’ and she describes in really simple terms.
What you’re talking about is a different angle on Charlie Munger, he reckons that between eighty and one hundred mental models do the bulk of the heavy lifting in any kind of problem-solving. Now, he’s focusing on investments but he’s also a living polymath in the sense of he’s big into architecture, law, and psychology but it’s reassuring, I guess, I’m even more reassured by you saying there are between twenty and forty of these models that cover the bulk of it essentially.
Absolutely, there are not that many ways but you can think about things like, ‘Is it a random walk? Is it linear? Is it random coin flips?’ There’s a whole bunch of stuff that-
Is it random, what? I’m sorry, I haven’t heard that term.
I’m sorry, random walk. So, our stock prices are random walk, do they randomly go up or down, this is the basis of something called the ‘efficient market hypothesis’, but you can ask, ‘Is a soccer match, is scoring in soccer completely random or in football, or is it in basketball, is it random?’ and you can use these models as lenses to try and make sense of how the world works. One of the things that I think is, my friend Michael Mauboussin has this book called The Success Equation-
Now, we’re going to get into Santa Fe very quickly but he’s what, I think, he’s not chairman, what is he, CEO?
Yeah, he’s Chairman of the Board of Trustees, it’s an institute, he has this book called The Success Equation, which he and ran a conference called, ‘How do you distinguish skill from luck?’, and one of the interesting, and this is really important again it’s very simple, if you think as a manager you’ve got to look at performance and suppose you just fit a very simple equation, that performance is, A times skill plus one minus A times luck, so there’s some proportion of skill and some proportion of luck. If the luck proportion is big, you don’t want to be giving necessarily huge bonuses based on performance because the thing is a lot of its luck. You also don’t want to be necessarily promoting people who did really well just this past year because a big proportion was luck, whereas if it’s mostly skill then you can know if somebody’s performance has suddenly dropped off for the last month, you probably want to intervene right away, there’s all sorts of management practices and understanding that can come just from understanding the proportion of outcomes that are skill and luck. That’s a simple equation. One of the things that Michael would say in his experience at SFI is just a great example of this, as is the Aspen Ideas Festival, as is Davos, but putting yourself in positions where you’re out there seeing people present ideas, frameworks, ways of thinking, challenges to the status quo is a good way to sort of maintain cognitive diversity. It’s also, Steve Jurvetson said, he’s the famous investor, he surrounds himself with people who challenge him and think differently and this is also the basis of things like the old Delphi method from RAND, right? You might put someone in the position of being the person who’s the contrarian but I don’t think this is just about being contrarian, I think you just want people to think about the world in very different ways.
Yeah, fascinating. And I guess, the Santa Fe Institute, I guess, you’re a visiting lecturer there, how does it feel stepping in there, well, as it’s described, I think, as a mecca of multidisciplinary studies?
No, it’s great. Part of it, the air is so clean there and you’re on a mountaintop, it’s pretty great, and you share offices which is really fun but no, it’s one of things, there are physicists, computer scientists, biologists, and David Krakow who is the president there, we just had an off-the-cuff fascinating conversation the other day. At what point using information theory, at what point do you think of something as a separate entity and at what point do you consider it a collection of independent things? So, we think of a person as an independent entity but we would think of a swarm of bees or an ant colony as a bunch of ants but the thing is an ant colony comes really close to being like a superorganism, and so can you use statistics and information theory to have some sort of measurement of to what extent is something an organization and what extent is something a collection of individuals? Then you can ask interesting questions, there is a great book by Ed Hutchins called Cognition in the Wild about how no one person drives a ship. Think of a modern ship it’s not like there’s somebody at the helm steering the wheel, it’s a collection of about twenty people all interdependently making choices that steer a ship into port. You could then apply these techniques and say, ‘Is this twenty-individual people or is there really some sort of superorganism that’s steering this ship?’ and one of the great things about Santa Fe and I think David as president is fabulous in this way is that your listeners natural reaction is, ‘What does that do for me as a business leader? Why do I care about that?’ and David would say, ‘That’s on you’. We’re doing the ideas here, we’re having a great time, we’re thinking about what is an organism and what’s not, what is skill and what is luck, what is exceptional performance and what is not exceptional performance. How do you put people on Mars? We’re going to take a big questions and puzzle and measure and create frameworks, you take it home and put it to work, but the idea is if you’re exposed to those things, if you’re exposed to the notions like path dependents, tipping points, between the centrality in a network and power in a network, you can then go back and do things like download your email network and say, ‘Wow, this person seems to have a lot of power with an e-mail network, I don’t want to lose her’ and stuff like that.
Well, and, of course, if you’re running a public company, the numbers are pretty clear, right? The McKinsey study around diversity I think giving a 35% performance for upper quartile versus lower quartile for instance. As we said at the beginning, there is plenty of evidence out there or a body of evidence building as to the economic value associated with adopting or embracing more cognitive diversity essentially.
That’s right. Absolutely.
Super, super. We could go on for some time but also, we’re going to have to stop. So, Scott, I sent you three questions before the show. First one – what have you changed your mind about recently?
This is a really good question because you’re often not asked that. What I’ve really changed my mind about recently is the importance of getting people to pause and think. So, I think I used to prize quickness a lot more than I have over the last three or four months. I’ve recognized that people I know who exercise really good judgment – you mentioned Charlie Munger earlier, where you talked about arraying his experience on a lattice of models, as I’ve thought back on the people I think are just really amazing leaders and thinkers that I’ve met, is their ability to pause and in some sense embrace the diversity I’ve talked about and say, ‘OK, let me think about all the ways I can think about this and not rush to judgment’ and so I’m now less impressed when I used to be much more impressed even two years ago, people who right away would give you an amazing answer and think, ‘Wow, she is just so quick’, and now, I’ve changed my mind and now I’m kind of more impressive people who sit back for a second and you kind of see them think and then they say, ‘You know, here’s I’m thinking about this right’ or who’ll write me back two weeks later and say, ‘Here’s what I’m thinking about this’.
We had a previous guest, a friend of mine, Kevin Cashman, he’s written a great book called The Pause Principle about that actually and I’ll send you a link afterwards but it’s great stuff, great stuff. Excellent, OK. Second one – do you have a personal work habit or practice you can share with our listeners that’s helped make you more effective?
Yeah, oh my gosh. So, I have two kids now that are bigger than me, they’re like seventeen and sixteen, and this is the fourth book I’ve put out, The Diversity Bonus, and I’ve found that unless I carve out a particular big block in the day that is absolutely my time to write, the book doesn’t get written. You can’t, especially if you’re doing anything deep, you can’t find a half hour here, twenty minutes or forty minutes there because it literally takes ten or fifteen minutes just to get your mind back in that spot, right? You know before I had kids when I used to teach at CalTech, I had eleven hours a day to do research, I constantly had these two or three hour blocks, it didn’t even occur to me that I had two or three hour blocks because it was just how I rolled but now when my life is so full I realize that the only way I can really advance as a thinker is by carving out an hour and half, two hours a day that are my time to think and I can’t do in my office because when I’m in my office people come and find me, so I have secret hiding places that I won’t tell you. I’m lucky that Ann Arbor is like one giant coffee shop at some point, there’s like seventeen coffee shops, so I have a couple of coffee shops right next to campus that I walk down to and hide out in. The other thing I used to do is that I used to listen to music and just not think when I was walking the dog. Now, we have a new dog and I literally, on that half hour walk with the dog, think through how I’m going to use my hour and a half.
Interesting, brilliant. I was thinking of, I don’t know whether you’ve read Cal Newport’s book about deep work?
Because given that we’re all hugely time constrained and everyone’s multitasking, to be able to do that and to really get yourself into the moment and go really deep for a couple of hours a day is a source of competitive edge because so few people are able to do that these days.
No that’s right, that’s right.
Lovely, great. And final question – what’s your most significant failure or low and what did you learn from it and how have you applied that, Scott?
So, you know this diversity work has really been both the low and the high and that is that – my background is this. I started out getting a PhD in mathematics, I switched to game theory because I thought it was super interesting, I worked with Roger Myerson and Leo Hurwicz, who won the Nobel Prize applying game theory to the design of mechanisms, I had a job at CalTech, it’s kind of like, what could be better as an academic? And then I got interested in this complex system stuff and one of the things I saw was this amazing disconnect for people in complex systems diversity played these functional roles, it made systems better, more productive, more innovative, that sort of stuff, whereas most people talking about diversity were talking about it in a sort of politically correct identity sort of space and had political overtones but no sort of basic value add, so I started doing work and started writing technical papers in this area and I could not publish them in the very, very top economic journals. It could be just below the top but when I’d submit it at the top I literally would get reports that would say, ‘Though this paper may have a large echo in the literature it seems tangential to the main focus of economics’, and these were papers about diverse teams solving problems, problem-solving the economy and that’s still kind of the case.
And what was going on there? Is that the ‘Structure of Scientific Revolutions’ story that it just hasn’t reached a tipping point yet?
I think it’s partly that, it’s partly because I’m using different techniques than people used and so then what I did is – you know, I’m in the American Academy of Arts and Sciences, I’m a college professor at the University of Michigan, it’s not like my career suffered, but what I had to do is I had to do an end around and publish in the Proceedings of the National Academy of Sciences, I had to write books which is something that I wasn’t prepared to do, I had to say, ‘OK, look, I’ve got to get these ideas out there’ and so what I did is started writing books which is not what a standard economic theorist does. If you look at my tribe of people in the economy who write mathematical papers, we do not write a lot of books that are, I don’t want to say popular, we don’t write books that are intended for a general audience over policy makers, so I had to do an end around but I have to admit, it was super frustrating to have to submit papers, in fact I’d just planned a paper with my wife in the American Political Science Review which is the flagship journal talking about cultural diversity and its functions and how institutions create diversity but that paper took three years, it got nothing but positive referee reports but it was still the case that the editors of the journal were like, ‘Well, we don’t typically publish stuff like this’ and we were like, ‘But everybody loves it’ and they were like ‘Well’, we’re going to send it to a couple more people’ but eventually they had to accept it so maybe now I’ve gotten over that hurdle but that’s – we wrote the first draft of that paper in 2001. Right, so that’s a huge hit but I’m not complaining because it’s given me this other life that’s been way more interesting. As much as I love writing mathematics on chalkboard and I do, if you said you’ve got an hour to live I’d probably go write about mathematics on a chalkboard while eating a burrito or something actually, but you know, it’s what I love to do but this is been a challenging thing for me, going out, meeting with business leaders, government people has been a great gift and joy, and having done work that has been relevant has been this amazing choice so yeah, huge setback from what I thought I would do should be like write mathematical papers that got lots of citations and people would say, ‘There goes Scott Page, he’s a really smart mathematical theorist’, I didn’t get that life.
So, it was wonderful to talk to you, Scott. Now, where can people get in touch with you?
So, there are several ways to get in touch with me. One is, they can just Google me and pop me an email if they want. The best way to interact at first though is, this model thinking course that I teach through Coursera. If you just type in ‘Scott Page Model Thinking’ there are hundreds of videos that you can just pull off Google or you can take the entire course. I think one of the best sorts of introductions to me individually is watch a couple of those videos and get a sense of where I’m at, and then I’m happy to answer emails from people at LinkedIn, Facebook friends with lots of people, sort of communicate generally. I have a book hopefully coming out at the end of the year on models, this model thinking thing, it’s mostly done, Basic Books and I are just trying to decide whether it should be six hundred pages or eleven hundred pages, we’ll probably settle on eight hundred, but when that comes out I think I’ll probably have a more active Twitter public profile but now I’m happy to take emails from people.
OK, and we’ll put links to all of that in the show notes here, but yeah, Scott, it’s been great to talk to you, I’ve really enjoyed your work and as I said, in the beginning, I think this is relevant for a lot of our listeners just given the complexity of their worlds and the reality of trying to solve those kinds of problems with one hammer or one tool and the opportunity that you’re thinking and the evidence supporting it gives them is I think very, very compelling so I really do appreciate your time today.
No problem, thanks a lot. I really appreciate it myself.
Have a good day.
Yeah, thanks a lot. Bye.