Fuqua Insights Podcast: What Happens When Too Many Ideas Compete for Too Little Attention?

Cassie Beddick  00:03

Welcome to Duke Fuqua Insights, a podcast where we explore faculty research and the actionable takeaways for business leaders at every level. Every business leader has to decide which ideas deserve attention, which trends to follow, which data to trust, and which advice is worth acting on. But when anyone can publish, post, or produce instantly, the challenge is no longer just finding information; it's knowing what has been meaningfully filtered. I'm Cassie Beddick, an MBA student at Fuqua, and today I'm joined by Professor David McAdams to discuss what happens when too many ideas compete for too little attention. His research shows that as content becomes cheaper to produce and easier to share, ideas may receive less scrutiny, and the quality of what we collectively know can decline. Professor McAdams is a Fuqua professor and economist whose work uses game theory and economic epidemiology to understand how information, behavior, and innovations spread. Thanks for being here today, professor.

David McAdams  01:10

Thanks, Cassie.

Cassie Beddick  01:11

So we can go ahead and jump right in. You've studied game theory and how information spreads. What drew you to this area of research?

David McAdams  01:19

Well, I am a game theorist. I've been teaching application of game theory here at Fuqua for 17, 18 years, and game theory is about strategic interaction. So the choices people make, how they impact each other, and how they change people's incentives. And so you can apply game theory in all sorts of ways. And over my career, I've studied several applications, but for the past decade plus, I've been really focused on this question of how information spreads, which is really central to economics. You know, you don't talk about it too much in say core courses, but if you think about it, how do we grow as an economy? We develop new technologies. We have to deploy them. So somebody has to learn about them. The information has to spread. In fact, last year's Nobel Prize winner Joel Mokyr won for sort of exactly this insight. He studied historically how the Industrial Revolution came about and traced it back to this social community that would share ideas and build on each other. So ideas just don't exist in a vacuum. They come about and they get deployed and utilized in a social context. And so that means socially people are making choices. They're interacting with each other. And so this has been a really fascinating field for me to think about through that game theory lens. Another example of why information spreading is important: think about marketing. Ultimately, we want people to buy a product, but you can't buy a product if you don't know about the product.

Cassie Beddick  02:56

Right.

David McAdams  02:57

So awareness of the product is the information. Okay, and you can encounter that information in many ways, including socially, by observing other people's choices, like maybe they bought the product or how they talk about it, or messages you hear on an ad. So thinking about how that information spreads is important. Now, how did I come to it specifically?

Cassie Beddick  03:18

Yeah,

David McAdams  03:19

I actually started in 2012, I became kind of intensely interested in the application of infectious disease. There was not a pandemic at the time; it’s a totally different story how I got involved in that, which is probably beyond our conversation…

Cassie Beddick  03:34

Another podcast

David McAdams  03:36

Yeah, and I have talked on other podcasts about that. But there, the information is the DNA of the virus. So you think about it like information can take many forms. It spreads by this virus getting into a new person, and that depends on choices. Am I interacting with you? Am I washing my hands? Am I getting a vaccine? These are all choices that impact how information spreads. But once I had spent several years in that world publishing papers in biology journals just as much as econ, it occurred to me that you know a lot of the methods I was learning about would be useful to study these core economic questions too, which often didn't have quite that infectious lens to them.

Cassie Beddick  04:25

So you talked a little bit about how you were working on that infectious disease side, and I feel like epidemiology is a word people normally think about that side of it. But your latest paper focuses on the field of economic epidemiology. So to people that aren't familiar with this concept, can you define that term for us?

David McAdams  04:44

Yeah. So, epidemiology is this really broad word in medicine. It's just how understanding population health outcomes, and there's a subfield called infectious disease epidemiology, which is specifically about how infectious disease spread. And economic epidemiology emerged in the early '90s, and the first application was infectious diseases. Just they were a little lazy, and they called it economic epidemiology instead of economic infectious disease epidemiology. But I've been pushing for a broader view of this term because I think that yes, infectious disease is one application, but there's all these other applications, including the ones we talked about, that they haven't really used the methodology that was developed for those infectious diseases. So the way I would frame it, and again, this would not be your Wikipedia definition, but what economic epidemiology is about is how the choices people make affect how information is created, how it spreads, and how it's used, and the focus traditionally is on that middle word, which is how it spreads. But a lot of my work is thinking about well, the way information spreads affects well the incentives to create it because information doesn't grow on trees, right? And so you're anticipating how your information will get out there, but also how it will be used. So, can people learn if, say, information is… can you learn if something is true or false? And that'll affect how you act. So, we want information to be used well, right? And whether people are able to use it well depends on how it spreads. So that's why we need these models that use this economic epidemiology.

Cassie Beddick  06:25

Nice, thank you. That gives us a good frame of reference. So, building on that idea, how should we think about the ways ideas spread and compete for attention?

David McAdams  06:35

Yeah. So I was thinking about maybe a good way to start is with contrasting with kind of the usual way of thinking about ideas and how they reach people. There's this concept called the marketplace for ideas. Have you heard this?  

Cassie Beddick  06:51

I have.

David McAdams  06:51

What do you think when you think of that?

Cassie Beddick  06:53

I think that you know we are in a society where everyone can have ideas, and so we're kind of picking and choosing among the ideas that are available to us which ones we want to.

David McAdams  07:07

Okay, so you can choose which ideas you like best, and there's also this sense in which the best ideas will win, right? And if you think about, implicit in that concept of marketplace, what is a marketplace? It's a place you go where everything is, and you pick apple, orange, and they're right next to each other, and you can just directly compare them. But if you think about how you actually encounter ideas, it's not really a marketplace. Ideas are spread around. In particular, ideas exist in people's brains, different people's brains in different places, so by definition, they're not all together in a marketplace, and social interactions are where they get exchanged. And when I say social interaction, I could mean us talking, but could mean also some other way of communicating, like you know, you make a post and I see the post, or you know, you do something, and I see the results of what you did, or something like that. So, if we step away from that conventional model that just ideas exist out there, and we think about how they're spreading, that leads us to focus on these other choices people make. You know, and in this particular paper I wrote recently, the main decision that I'm focusing on is how intentionally or how intensively do you go out there and seek out information. But there's a lot of other choices people make too, like how intentionally do you really take care to only tell other people things that are true? So are you investing your time and energy in figuring out what's useful for someone else, or are you just saying it? You know, and economics also focuses on intention, like what are your goals, you know, and that can affect, I think because goals can also be designed, we can change what motivates by rewarding someone for certain behavior and so on.

Cassie Beddick  09:03

So you said… you were talking about what we're focusing on. Attention. So attention is a big part of this work. Why does attention matters so much in determining which ideas survive, spread, and get ignored?

David McAdams  09:17

Again, I think it's helpful saying why is it useful to think about attention? And attention in economics was really brought into the world of economics by this guy named Herbert Simon, who won the Nobel Prize back in the 70s. And because at that time computers were just coming online and there was just like this explosion of information, it's like ‘oh wow, well this will allow us to live in a utopia,’ like people were thinking. And Simon is like, you know, it's great to have all this information out there, but this creates a new dilemma. He said, information's not free, but he said when there's an abundance of information, there's a scarcity of attention. And what he meant by that is like, I don't have this infinite capacity to just think about everything, and so his work really focused on the role of information processing. It's like what you really need is not like someone to go out there and collect gazillions of bits of data. What you need is someone to take existing data and turn it into something simpler and more actionable. So there's this concept of information overload, which he kind of… his work brought to attention. And speaking of attention, I've kind of forgotten your question. So can you remind me?  

Cassie Beddick  10:36

Of course. Why does attention matter so much in determining which ideas survive, spread, or get ignored?  

David McAdams  10:43

So okay. Starting point is we don't have infinite capacity. But then you might think to yourself: well, as technology is changing, the cost of attending to lots of information is also changing, so maybe the utopia world could come about. Maybe, as the cost of attention goes down, we could end up living in a world that's approximately the marketplace for ideas, where everyone can pick and choose and figure out—right? But the problem is that people are making choices, and the choices they make will change as the environment changes.

Okay, so, in game theory we have this concept called equilibrium or Nash equilibrium. So, we need to think about how our choices are engaging and interacting with each other, and so it could be that as the cost of attention goes down, we get closer to the utopia, or it could be that we remain trapped in some sort of equilibrium that's very far away from that.

Now I should say I didn't necessarily ask that question in the most recent paper. That's like I guess an open research question of whether, as the cost of attending things goes down -- but just speculating here -- there's this thing called the free riding effect that you might have heard about. That when someone else can do something, well, you have an incentive to let them do it and do less yourself.

So, thinking about that social interaction, thinking about the equilibrium outcomes, having a systematic, careful way to think through these effects and their secondary effects--that's kind of the value of having a theoretical framework.

So we may go to talk a little bit about the work I did, but you can use a systematic framework to address other questions as well, and I think that's sort of useful, especially in a fast-changing world. So, like, as every day the technology is able to do something I couldn't do yesterday, the theoretical framework is helpful for guiding you to better predict and understand where things are going to go. So what I've been trying to do is build those theoretical frameworks that people can then use.

Cassie Beddick  13:11

Nice. So you've already mentioned a couple times about your paper on the research that you've been working on. Let's maybe get some cliff notes. Hit the highlights for me. Tell me what this paper is about.

David McAdams  13:24

Yeah, fundamentally, what I'm thinking about is how our choices interact with each other, right? So in particular, let's say you learn about something. Okay. Now you might then tell me. Now I know about it and I might then tell someone else. Now we could just say, well, let's imagine that as a mechanical process where things just automatically spread. Okay, but it's not a mechanical process. When you encounter something, you're going to think about it, and you may or may not put it in your memory. You may or may not put it in that part of your memory that you tell people at parties, right? So there's some choices going on there, maybe more or less conscious choices, but there's choices going on there, and that affects what you share. And then I will encounter what you tell me, and I'll say, "Oh, Cassie just told me this thing. She's a pretty discerning person. Maybe I should know this thing.” But I might think, well, that's dumb. Like she made a mistake this time. I'm not going to tell. So, what emerges are these social transmission chains, where as information goes farther down a chain, on average it gets more reliable. Okay, so there's a sense in which you can think about information not as a marketplace, as a whole bunch of chains that are reaching you. So you learn about things, and if you draw those chains, it looks kind of like a tree. That's what we call a network. Okay, so this network emerges, and the main thing I focus in this particular paper that I wrote most recently is the volume of information that's coming in. If once in a while, you get like this little, little drop of knowledge, you're like, oh my god, I got a piece of information. You're going to treat it differently than if there's this like torrent, this hose of information being hit with all the time. And in particular, because you have limited information, you won't be able to process everything in the torrent. Okay, and when you talk to people, they're getting such a torrent; they won't be able to attend as much to what you told them. So what's going to end up happening is the social chains, the transmission chains, will be shorter, and so the information that is in people's minds will be less reliable.

So this model is capturing a trade-off between the volume of information that comes into a society and the quality of information that people know, so people know more, but they're dumber.

So that's kind of the equilibrium outcome. And so -- and there's various implications for that -- what does that really mean? And I should also just say, as a theorist, the point of theory is not to try to build a model that exactly matches the world, because if you think about a little while, you'll be “you know what, David? Yeah, that transmission chain -- I can see that's an aspect of the world, but there's other aspects of the world that matter.” There's other things that happen, and that are missing from the model and that you may come up with a couple of those, and we can chat about them. And those are future research directions.

But anyway, that's the essence of the idea of the paper, which is, and it's motivated by changes in the world that are causing information to come at us a lot faster, and also to come from sources that may be less reliable.

So if the sources are less reliable, we're going to rely more on this social filtering process. That's what I call it. We're gonna -- if people are really good at figuring out what's true, it doesn't matter that there's all this crap hitting us all the time. But the more crap hits us, the less we're able to deal with it. And sadly, what does that mean? The less we're able to deal with it, why should I even talk to you? So the transmission chains get even shorter if we don't talk to each other. So that's a vicious cycle, and so that can lead to our collective knowledge being even worse.

Cassie Beddick  17:10

What if, with the whole torrent of information, what if people are also not good at being able to pick out what is real because they know there's lots of misinformation out there?  

David McAdams  17:21

So that's the thing, that's the thing. So, misinformation is not in this particular model, but I have studied it in other papers. So in this particular model, the information that comes in from you is just a given. So what are the implications for the creation of information? So if your goal as a misinformation creator is to harm the society that you are injecting information into -- well, hey, this is good times because guess what? If you put in extra information -- okay, even if it's equally good as before, okay? Let's just take that, then the society won't be able to process it as well. And they're going to collectively trust their knowledge less. But even better, if you can pour in a bunch of crummy information, or information that makes it harder for people to discern what's reliable or not -- one feature of real information is it's tagged with various labels, you know. Like I know when I buy. I mean, I know what a Big Mac is. I go to McDonald's. I don't need social learning to figure out whether what a Big Mac tastes like. But when we're talking about non tangible things like stuff you encounter online, if there's all sorts of these creators creating things that look kind of like it but are actually crummy, it's going to undermine your confidence in the real thing. And if you think about the next step, well, I'll ask you: Once people have less confidence in the good stuff coming in, how's that going to affect the game among the creators, how are their incentives going to change?

Cassie Beddick  19:05

Well, I think it depends on what their goal is.

David McAdams  19:07

Okay, so we have creators making really good content, like the news, people who are really going out there and interviewing like four people and only publishing one quote. Okay, like they're really doing a deep dive.

Cassie Beddick  19:18

Gotcha.

David McAdams  19:18

How do their incentives change when someone else is pumping out tons of things with artificially generated quotes, does it give them more incentive to do their job?  

Cassie Beddick  19:28

I don't think so because they're working really hard.  

David McAdams  19:31

Probably frustrates. Because there's another phenomenon in news where a journalist publishes an article and then -- like 30 seconds later, or an hour later -- someone publishes a copy of it, or a summary of it.

David McAdams  19:46

Okay. And then they get the eyeballs, so they get the money. So in that case, you're motivated by money. Let's say, and someone steals your eyeballs, you now have less money. You have less incentive to create articles in the first place. Or if you create them to do a less good job -- make two instead of one really good one -- and in the same way, I'm saying that as this sort of channel of information, the hose, has more stuff coming through it, each individual idea spreads less far. Okay, reaches fewer people, and people trust it less. So each time it gets exposed to a person, they're less likely to adopt it, or put it in their brain, or buy it. Whatever your interpretation is of the information, that reduces their incentive, and that's a vicious cycle.

So if you're a misinformation player whose goal is to harm society, as opposed to one whose goal is to make money or something, then it's good times for you. There's a clear playbook for disrupting society.

That said, once society sucks, so once the equilibrium is bad enough, that creates a great great incentive to change the game.

Cassie Beddick  20:51

Interesting.  

David McAdams  20:52

I mean, this model is not about the solution. What I did in this paper was basically look at this sort of anarchic situation where there's no labels -- like when you send out information, you can't say I wrote it as Cassie, and everyone knows I have a personal reputation and I would never publish something false, right? And reputation just one mechanism. There's lots of mechanisms to try to solve a problem, so I'm not saying that we're doomed for a world where the misinformation agent wins …

Cassie Beddick  21:22

Good!

David McAdams  21:22

… but they could spur us to overcome some of the challenges associated with doing a solution, which can require collective action or someone to step in and do something.

Cassie Beddick  21:35

Let's talk about AI for just a minute because I think that is impacting the amount of information that we have available to us, and it's easier to produce it, access it, summarize it. Like you mentioned, what does your work suggest about the role of human judgment when deciding which ideas to trust, especially if we don't know if something was made with AI or not?

David McAdams  22:01

Right. So there's a lot going on in your question. You asked about human judgment. Let me start by just thinking about the AI.

Cassie Beddick  22:08

Yeah.

David McAdams  22:08

Okay. And you said three things. You said produce, access, and summarize.

Cassie Beddick  22:17

Correct.

David McAdams  22:18

Okay. So let's just think about each of those separately, how it's changing the game. Okay, so with AI you can now produce content. Okay, so first and easiest point I can make. Okay, now more stuff can come out of the hose.

So if there's a producer out there and the cost of making stuff is going down -- so this is actually something I have explicitly in the model. So as the cost of making things goes down, what's going to happen? Well, because more stuff will be coming out, people's trust will fall. Okay, and they will share things less, and the information ecosystem can shrivel up. Now it turns out, at least theoretically, it's possible that it could shrivel up so bad that the revenues go down to information creators so much that overall production collapses as well. So that sort of thing can happen.

Cassie Beddick  23:10

Wow.

David McAdams  23:10

That's assuming the misinformation. These AI people are in it to make money …

Cassie Beddick  23:15

Right

David McAdams  23:15

… as opposed to ruin us. Okay. But more broadly, what I call -- I don't know if I use this word, veracity -- but how true the stuff is that's in circulation that could go down because this process of what I call social filtering those chains …

Cassie Beddick  23:32

Right

David McAdams  23:32

… like how many times has it gone through a filter like a sieve? Well, there's going to be less of that. So, the producing side, I think AI could be bad news, especially if it's not producing higher quality things than before.  

Accessing information though, is a little different because, like I mentioned, now there's a network that's created by our interactions. Every time we interact, interact with like a line drawn between us, you can connect those lines. But if everyone goes to Claude, now suddenly everyone is having a social -- let's put it in quotes -- interaction with Claude, and so now the network is what we would call a star.

Cassie Beddick  24:15

Okay

David McAdams  24:15

So there's someone in the middle. It's more like a koosh ball. Have you ever played with those? All these little lines coming out. So there's one point in the middle, and all the other lines are just coming out of it, and so that's a very different structure, and it could be very effective if the center of the star has great information.

All right, and this actually relates to another famous scientific study. This guy, Francis Galton. Ever heard of?

Cassie Beddick  24:43

I have not.

David McAdams  24:44

Wisdom of the crowd.

Cassie Beddick  24:45

Oh, okay.

David McAdams  24:46

So he did this famous study right around the turn of the 20th century. So a long time ago, and he went to a state … it wasn't a state fair, it was in England … but he went to like a state fair, and they had a thing where… in some movies they have like they guess the weight. You ever seen Babe?

Cassie Beddick  25:04

Yes.

David McAdams  25:04

Okay, the way he wins the pig is he guesses the weight of the pig – he is the closest. What they did is they have everyone guess the weight of an ox. Whoever guesses the closest, like get the ox. Okay. And what Galton did was he collected all the guesses and averaged them, and he found that the average was closer than like anyone's guess, or really close. And so that led to a theory that predicted that if a lot of people have information, some people are overoptimistic, some people a little too pessimistic, but as long as their information is diverse and not biased, and so on, there's like a mathematical result that says their average will be relatively good.

So if you're the center of a star, and let's say you can gather information from all the people you're connected to as well as give them information, then you could be ‘wow! Okay, 1000 people have tried this new shaver and they like it.’ Now, when someone asks for a recommendation for a shaver, I can give them a good recommendation. So now we're not relying on social interaction as much, but more on centralization.

Cassie Beddick  26:11

Okay.

David McAdams  26:12

And on the other hand, if somehow one person tries the shaver and that causes Claude to recommend it, guess what? The second person is going to try it, and any input they give back has been polluted by the fact that Claude gave it to them.

Cassie Beddick  26:26

Yeah.  

David McAdams  26:27

And that's called an information cascade – a different concept. But like very little information could be embedded in there. So, kind of depending on how the game goes… let me say the AI is definitely changing the network. It's definitely changing how social interactions happen, and if we go to Claude more and we don't talk to people, then that would also change the game. Yeah. So that's a big way, and then finally, you said summarizing information. Again. there's intelligence here. So so maybe there's a way that AI could just profoundly change the game so that my model is not relevant at all, right? And then could potentially lead to really good outcomes. So, but just thinking that through, there's a lot of possibilities. I think there's a lot of room for future research thinking about specific specifically how AI changes this game.

Cassie Beddick  27:19

Nice. It's very interesting. You know, we've been using a lot of AI at Fuqua to help us out. So let's do one more question for MBA students and business leaders. What is the practical takeaway from this research?

David McAdams  27:37

Yeah, I think in terms of practical takeaway, one thing I would say is you don't want your organization to be like my model.

Cassie Beddick  27:46

Okay.

David McAdams  27:47

Okay. So, as I said, I did the exercise of let's imagine a sort of anarchy world where people just interact with each other. A finding is that social interactions are good. They tend to socially filter things, and so knowledge is a little better, but another key finding is that you're never going to get to perfect. Even if people interact a ton, a ton, those chains go on almost forever. And the reason is, as things get more trustworthy, people do less filtering. So if a source … if information is coming from trusted source, you're going to do less, you're going to be less discerning yourself. Yeah. So you're you're not going to get the ideal outcomes unless you sort of intentionally design the information environment. How are those interactions happening? Of course, it is the one is most related to my work. But also, like, what and how are people able to exercise judgment? Like, what information do they have? What tools do they have? There's the Galton, but what you want to be more like is these other types of models that work. So there's the Francis Galton model. So that's the wisdom of the crowd, and what that would suggest is that if you have a decision that people have diverse information on, so it's not like a situation where one person can figure it out, but you want to gather the insights of a group. You want that game to be more like his game, where everyone writes down their guess independently, and then you get to average them somehow. So you don't want one person to step up in the meeting and like just say their point of view because they're going to be affecting everyone else. If you said, "I think this ox is 200 pounds. That's probably way too low. That's way too low for it. I was like, Cassie, you have no idea what an ox weighs.  

Cassie Beddick  29:27

I've never seen an ox.  

David McAdams  29:28

You’re like a digit, a number off, right? But so that's one model. Another model I think is, you know, when you publish a paper in Science, like you have an incentive to deeply, deeply research it. Okay, so the paper is really good, and you have someone evaluating it who knows what they're doing, and then you have a process to. Spread it really widely if it succeeds, and so if you have a decision in your organization, it's a high stake situation. You know, it makes a lot of sense to delegate and incentivize someone to really be invested in getting that answer right, and then having some evaluation. And if it's viewed as good, just to spread it to the group. So anyway, my model is more like a benchmark, and you don't necessarily want to live in that benchmark where you're just relying on people chatting with each other at the coffee. But there is value to that, and you do want to bear in mind that a lot of benefits happen. Ideas, you know, combine and they get shared, and the ones that are shared a lot are more successful. So you do want to listen to what people are talking about around the water cooler.

Cassie Beddick  30:48

Gotcha. Well, amazing. Thank you so much for being here, professor. I know I've certainly learned a lot, and I know our listeners will too.

David McAdams  30:56

Cool. Thank you, Cassie.

Cassie Beddick  31:03

Duke Fuqua Insights is produced by the Fuqua School of Business at Duke University. You can learn more at fuqua.duke.edu/podcast.