Big Data, Little Brain: Making Structural Components Known and Emergent Outcomes Clear with Paolo Gaudiano

Welcome to The Flux, where we hear stories from people who have asked what if questions to better understand the world and talk about how data can help tell stories that impact decisions and create an intentional impact on the future. This is your host, John Cordier, CEO at Epistemix. In a world where the flux capacitor from back to the future does not yet exist, people have to make difficult decisions without always knowing how the future will play out.

Our guests are people who've taken risks, made decisions when uncertainty was high, and who have assisted decision makers by using data and models. We hope you can turn lessons from our podcast into foresight, so you or your organization can make better decisions and create an intentional impact for others.
-
John: Hey there. Welcome to The Flux, where we hear stories from people who’ve asked “what if” questions to better understand the world and learn how data can help tell stories that impact decisions and shape the future. This is your host, John Cordier, today joined by Paolo Gaudiano.

Paolo: John, it’s a real pleasure to be on. I’ve been following your work for a while and have been eager to talk with you, so this is quite an exciting opportunity. Thank you.

John: Of course. You’ve had a lot going on lately. Why don’t you tell our listeners a bit about your current work, and we’ll start there.

Paolo: The majority of the work I’m doing these days is focused on a field that’s not seen too favorably right now diversity, equity, and inclusion. Most people hear that and think, “Oh, condolences, this is not a great time to be doing that.” I actually got into DEI about ten years ago, coming from more than two decades of work in applied complexity science.

About ten years ago, I saw an opportunity to use that background to quantify the financial benefit for organizations that are more diverse and inclusive. A lot of DEI work focuses on fairness, justice, and equity, or uses broad correlations that don’t tell you much because of the old correlation versus causation issue. What I wanted was a way to tell organizations and their leaders how changing the way they treat employees could improve their financial KPIs.

So I set out to do that and now run two organizations. One is Valeria, a mission-driven for-profit company focusing on measuring inclusion. The other is a nonprofit that does core research using agent-based modeling, applying it to business and societal problems.

John: Many of our listeners come from complexity science and agent-based modeling backgrounds, but you might be the first to connect that work directly to DEI. What was your “aha” moment when you realized complexity science could be applied to this space?

Paolo: That’s a great question. I actually started doing agent-based modeling before I knew that’s what it was called. During my master’s and PhD work in the 1980s, I studied neural networks and simulated individual neural circuits to understand things like how the retina translates light into signals or how humans learn to control arm movements. Those were essentially agent-based models.

When I became a professor in the early ’90s, I met Eric Bonabeau. We became friends, collaborated, and later worked together at a company called Ecosystem. That’s where I really started applying agent-based models to practical problems like consumer insights and personnel management.

Fast forward to 2015 Eric and I were doing R&D and consulting but were tired of chasing the next big project. Around that time, I attended yet another session on diversity and inclusion and was struck by how the personal stories people told about discrimination contrasted sharply with the vague, systemic “solutions” being proposed. I thought, how do you connect individual experiences with organizational outcomes? Then it hit me: this is exactly what agent-based modeling does.

So I built a simulation to understand how workplace experiences impact financial outcomes. That led to my first model in DEI, which grew into the work I do now. Interestingly, I never mention complexity science or agent-based modeling on my company website, even though that’s the foundation of everything I do.

John: When organizations use your simulation, what kinds of changes are they trying to test or understand?

Paolo: Interestingly, when I work with most clients today, we only use simulations during the sales process. I show them a video of a simulated organization with four layers entry level, management, VP, and executives starting 50/50 male and female. Under normal conditions, that balance holds. But if you introduce a small bias in promotions say men have a slightly higher chance of being promoted the organization quickly skews toward men at the top. It perfectly mirrors real-world data.

We even published a paper showing we could match gender imbalances across U.S. industries. When I showed this to a chief diversity officer, she wanted to customize it for her company. But I realized their diversity data alone was useless it didn’t tell us anything about people’s day-to-day experiences.

So we created a confidential platform where employees report specific workplace experiences that affect their ability to work. Initially, I wanted to feed that data into simulations for forecasting, but I realized people weren’t ready for that level of sophistication. Instead, we began using the data itself to measure inclusion.

Most organizations measure diversity, which is an emergent outcome. Poor diversity doesn’t tell you why it’s poor or how to fix it. Measuring inclusion gives you the underlying behavioral data that leads to those outcomes. For now, we use simulations mainly to motivate data collection, but eventually, we’ll combine both for strategic planning.

John: That’s fascinating. You mentioned earlier that trying to manipulate emergent outcomes directly often fails. What are some examples of that where people try to fix the symptom instead of the root cause?

Paolo: The DEI world is full of those examples. The most classic is focusing solely on hiring diverse talent. I’ve been warning since 2018 that this approach would cause backlash. It’s like walking into a cold house and lighting a match under the thermostat it’ll read warmer, but you haven’t fixed the drafty windows or open doors, and you might burn the house down.

When companies focus only on hiring, they upset existing employees and bring new ones into environments not designed for them. The real issue is usually in retention and workplace conditions. Similarly, with pay gaps, many try to “fix” disparities by simply raising salaries for underpaid groups without addressing the systemic causes. It’s a symptom fix, not a solution.

People tend to think linearly “I don’t have enough women, so I’ll hire more women” but complex systems don’t work that way. Simulations help people see this clearly and challenge their assumptions.

John: How do you help people begin thinking this way understanding emergent systems without necessarily diving into technical modeling?

Paolo: I’ve been teaching a course for years called “Managing Complexity in Business and Society.” I introduce the distinction between structural elements and emergent properties. Simple simulations like single-lane traffic models help people see how complex behaviors emerge from simple rules.

For executives, though, it’s often about showing proof examples of similar organizations where this thinking worked. Once they see results, they’re more open to understanding the concepts. Most clients don’t care about methodology they just want solutions that work.

John: Where can people interested in complexity science or agent-based modeling start learning more?

Paolo: Start with NetLogo. Download it from Northwestern University’s website and play with the models. It’s hands-on and teaches you intuitively how simple rules can lead to complex outcomes. Explore the model library to see the diversity of applications.

For those interested in theory, I recommend the book by Uri Wilensky and Bill Rand. Santa Fe Institute and the New England Complexity Science Institute are also good resources. My own upcoming book aims to be a more practical, applied guide to complexity thinking in business.

John: Let’s talk a bit more about measuring inclusion. Since diversity is easy to measure statistically, how did you define inclusion in a measurable way across different organizations?

Paolo: That was a challenge. I approached it backward from the individual level up. I think of diversity as a company’s balance sheet: it’s a snapshot, but it doesn’t tell you how you got there. To run a company, you also need cash flow and profit-and-loss statements.

Measuring inclusion is like measuring cash flow it captures the day-to-day experiences that impact satisfaction and productivity. Measuring equity, or merit, is like your P&L it shows which organizational processes are fair or imbalanced over time. Diversity is the emergent result.

So inclusion data tells you what’s happening to individuals, merit data tells you whether your systems are fair, and diversity tells you the outcome. You can’t fix diversity without addressing the first two.

John: Brilliant analogy. Now for our closing questions. We always ask guests about a “What the Flux” moment something you wish you could go back and redo.

Paolo: Early on, I came into this space thinking too narrowly from a gender-only lens. I wanted to show the complementary value of men and women at work, similar to Scott Page’s work on the value of diversity. But I quickly realized that view was too limited. I needed to think more broadly about all kinds of inequities and how they connect.

On a personal level, I’ve changed directions many times from aspiring marine biologist to engineer to neuroscientist to professor to entrepreneur. I’m comfortable making sharp turns when something excites me or surprises me. Lots of flux moments in my life.

John: I love that. Finally, what would it take for agent-based modeling to become as mainstream in business as machine learning or other data science tools?

Paolo: I think people underestimate how much domain expertise agent-based modeling requires. Most data-driven methods rely purely on statistical patterns what I call “big data, little brain.” Data scientists analyze patterns without truly understanding the domain, then hand results back to experts to interpret.

With agent-based models, you must combine domain expertise with modeling skill. That makes the results richer and more intuitive. You can literally see your own mental model come to life in the simulation. When people realize that, they see it’s not just data it’s their understanding of how the system works made visible.

John: That’s fantastic. Paolo, thank you for being on the podcast. We’ll definitely have to continue this conversation.

Paolo: Thank you. It’s been a real pleasure.