Join host Daniel Serfaty in the fifth and final part of this highly successful series on the science of teams as he talks one-on-one with Professor John Hollenbeck of Michigan State University. Colleagues and friends for more than 30 years, Daniel and John explore the magic of teams, from their early work together on the US Navy’s groundbreaking A2C2 program, to how COVID and remote work is changing the very nature of teaming, to a future of multiteam systems that marries human intelligence and artificial intelligence.
Join host Daniel Serfaty in the fifth and final part of this highly successful series on the science of teams as he talks one-on-one with Professor John Hollenbeck of Michigan State University. Colleagues and friends for more than 30 years, Daniel and John explore the magic of teams, from their early work together on the US Navy’s groundbreaking A2C2 program, to how COVID and remote work is changing the very nature of teaming, to a future of multiteam systems that marries human intelligence and artificial intelligence.
Daniel Serfaty: Welcome to MINDWORKS. This is your host, Daniel Serfaty. This week is part five of our five-part series on the magic of teams. It's been a very successful series from the feedback we've received from the audience. And so I am delighted to conclude that series today with a very special guest.
Professor John Hollenbeck is an old friend of mine, but he is also a university distinguished professor at Michigan State University, and the University Professor of Business at the Eli Broad College of Business there. His long-term research focuses on team decision making, self regulation theories of work motivations, and employee separation and acquisition processes.
What is unique about John is that he has the unique ability to blend very practical advice that I'm sure he shares with his students in management and in psychology, as well as a very deep and critical understanding of theory and what makes team work and what could make team work better. He has been honored by almost every single society in psychology and management, and I am here to bet that there is not one student in industrial organizational psychology or management that haven't studied in one of his books or read one of his papers.
John, welcome to MINDWORKS.
John Hollenbeck: Thanks for having me Daniel. When I got a call from you to do this, I can't tell you how excited I was. We go back together to 1991. We worked together for three years-
Daniel Serfaty: Oh, my God, [crosstalk 00:01:48].
John Hollenbeck: -and that was while you were only a teenager back then. But that was really kind of my first foray into both granted research and into research and part of the ATCT program working with you and you led that. I learned so much from you, watching you organize that group of cats that we really were. You were herding cats in the ring, as you know, you had kind of qualitative people and mathematical modeling people and lab people and field study people, and you just orchestrated that thing so beautifully. Everybody talked to each other. It was a project like nothing I've ever been on before or seen really since, in terms of the diversity of approaches that you brought to that.
John Hollenbeck: And I know it was tricky. You were really herding a lot of cats there, and I always admired it, because you were a teenager, I was only eight. So I looked up to you as my older brother.
Daniel Serfaty: Well, at the third leg of a young, very unique personality is incredible sense of humor, even making fun of me of herding cats in an academy industrial environment eons ago. Actually, we go back to that program because that was an interesting, almost meta program, in which we learn how to organize a team while developing principles of team organization amongst ourselves.
But John, you had a choice of career, graduating from a PhD program a few decades ago. What made you choose this particular domain of focusing on teams? As a field of endeavor, I know you have other focuses, but what about the teams aspect? Why that and not any other domains in management?
John Hollenbeck: Well I'll be kind of quite honest with you. When I first came out in 1984, my initial program of research was on goal setting and goal commitment at the individual level. Team research is very, very difficult to do, and we'll talk about multiteam system research a little bit later, which is even harder for me. As an assistant professor, I actually didn't feel that the teams area was a safe place for me to work. I needed to get tenure, I needed to kind of get that done. Individual studies were just, "Crank those out faster," and it was a little less complex. So for a young person it was kind of easier for me to get my arms around it, and I went up for promotion early. Once I got promoted, I started working on teams and I never looked back.
But I actually do feel like I needed a little bit of security behind me to go to into the teams area. And again, just to put it in perspective, I probably got promoted in '89, it was really 1990 that we got part of an ATCT grant. After that, I was doing funded research on teams the rest of my life.
But I have news to share with you today. We just got news of a $1.6 million grant from ARI this morning, and we'll talk about how that fits in, and Daniel, it's still along the same track that we were in 1991, my friend, in terms of kind of the things that we're studying. But that project with you was really kind of my first foray into teams and my first foray into funded research, and my first foray into the military. I was using words like 'boat'. They're like, "No, you should be using words like 'ship.'" I was using all kinds of offensive language that first couple of years, and [inaudible 00:04:57] helped us survive that first round and the next thing you know, it's history.
Daniel Serfaty: That's funny. I mean today, even myself, after working with the military for more than 30 years, I still make some terminology faux pas so to speak. But you learn, and people are pretty tolerant of that.
So you mentioned studying teams is very difficult. Can you unpack that for our audience? Why is that particularly difficult?
John Hollenbeck: Well, for one point, and we'll talk about multiteam systems, if I want to do a study and an individual has statistical power, I need 80 people. If I want to have the same level of statistical power in team context, I need 80 times 4, I need 400 people to do that study. So that's not a lot. Just coming up with another research participant, especially in the field context. You got to be in a context where, not only do you have 80 teams, but these 80 teams are not comparing apples and oranges. If you're comparing a basketball team or a football team to a software development team, you've got to kind of find 80 teams that are doing similar work so we can talk about who's performing well, who's performing poorly.
The other thing is that the team level, there are so many different levels going on. Not only is there individuals, but in many cases, those individuals make multiple decisions. And so the decisions of individuals are nested under individuals. The individuals are then nested within the team. And now within the team, there's dyads. And we just published a paper in JDP on triads, and it turns out triads are a unique, specific thing worthy of their own study. And so if you look at a chain, it's kind of like a microscope that if you dial in at this level, you see individual decisions. If you take it out a level, you see individual people making a bunch of individual decisions. Now you pull it out, you see there are groups of two people interacting with each other, and then three people... and then there's the team.
And now if you take this as a multiteam system level where we're working today, okay, now I need 80 multiteam systems. Dude, that's really hard. That's 15 times 80, and they've got to be doing something comparable. It's even hard to do it in a lab, because it's just generating a number of subjects. And this relates to the grant that we got today. We're trying to build a national infrastructure for multiteam research, where if you have teams and I have teams and [Debra 00:07:14] has teams, if you have 80, I have 80, if [Debra 00:07:17] has 80, we can get together, technologically we can run a multiteam system.
So we're trying to build this national infrastructure to lower the volume of the people trying to do multiteam system research. But just going from individual stuff to a team by itself is so much more complex, and I will tell you in 1984 as an assistant professor, I wasn't ready for it. As you can probably attest, I wasn't ready for it in '91 either when I started working with you. But 25 years later, I think I'm starting to kind of get it. And so it's a lot more complex.
Daniel Serfaty: Thank you. And we'll unpack all these notions of triads and multiteam systems for our audience a little later in the discussion. There is a myth almost, teams are almost mythical, especially in America with this notion of sports teams and the great teams that are more than the sum of their parts, et cetera. Does it introduce another level of complexity, this notion that there is some magic that are happening, because human beings are designed to work with other human beings. Something at that level?
John Hollenbeck: I love the term magic, because I think we were talking about this before, we kind of used the term magic. I do think there's a magic there because the chemistry... it's not what's happening at the individual positions or the individual or the dyads or the triads. It's all of those things kind of working in parallel so that in many cases things will happen. And it just looks like magic to you because like with a good magic trick, you're looking at the right hand, not the left hand. You're looking at the individual, not the team, or you're looking at the team and not this particular dyad within the team.
And so the magic occurs because you're looking at the right hand, at the action level. So breaking that apart is fun. But I got to say, there's two things about the metaphor when you use magic. There's both good magic and bad. Again, I really believe that there are decisions that are so bad, so irrational, and so illogical, you can only get them out of people in the social context. An individual working alone would never make this mistake, but I can get my MBA students, my executive development students, to make unbelievably stupid mistakes if I put them in a group context and set them up.
I'll give you one example. This would never happen to an individual. But you're probably familiar with the cognitive diversity research on framing, and how if you frame an issue in terms of, these are the things that will be lost, you literally get in a very risk-seeking part of somebody's brain and they become risk-seeking. If you take the exact same data and just flip it around and talk about what you could gain... I mean these are just inverse probabilities. This is the exact same thing. But you just framed it as a gain or loss. If you tell people, "Frame it as a gain," they become extremely conservative. And now you take that process, which is a individual process. Now you put it into a group context. We'll take my MBAs. I'll take four MBAs that are... We do a lot of surveys, so I'll take four MBAs that I know are really risk-seeking in general, as a pre-disposition, and I'll take four of my MBAs that are really cautious, as a pre-disposition.
You can set up the framing, "Oh, you're making an irrational decision. Relative to the probabilities, you're being way too cautious or you're being way too risky." But what happens in a group context is group polarization. That is if you and I start overconfident, you're 80% confident, I'm 80% confident, [Debra's 00:10:34] 80% confident, you put the five of us in a room together for 20 minutes, and then you come back and ask us how confident we are, it's like 99. I mean, you're literally polarized because, "Wow, Daniel, I didn't think of that. You're right. That's even better that what I thought of." Nobody has any negative information. We just kind of beat each other and beat each other, and if you go to my cautious students, they go just the other direction. They're afraid of everything. They won't get out of bed. "Oh, this is only going to be 20% successful." By the time they're done, there's no chance. It's .01.
Only in a group context can you take people that would be a little bit irrational in kind of diversity terms, make them unbelievably irrational, and then I have a slide about it. My students are blown away that number one, this happened, and number two, it was so predictable that the dude's got a slide about it, and the rest of the lecture is built on this error that we all just made that we didn't even see coming. That's why I believe it's magic, because I can make this happen over and over and over again, with every executive development. Every MBA group.
We do a thing called a $10 auction. I don't know if you're familiar with the $10 auction.
Daniel Serfaty: No, please.
John Hollenbeck: Okay [crosstalk 00:11:46]. You can only get this in a group context. Basically, you put a $10 bill in an envelope and say, "We're going to have a bidding war for this $10 bill." Now, in most situations, the key to an auction is figuring out what something's really worth, but you know this is exactly worth $10, and so this auction has an interesting set of rules.
If you pay the most, you get the $10. If your bid is the second highest, then you pay that bid, but you don't get the $10. Third, fourth, fifth, you're out of it. So I put that $10 in there, and we start. Usually it just sits there for about 15 seconds, and Daniel, I've got execs that are COOs in organizations. Eventually some exec will say, "Well, it's a no-brainer. I'll pay $1 for a $10 bill." Then another guy says, "Two. Three, four, five, six." They get through to seven, and then they start laughing, because they'll look at me and go, "Dr. Hollenbeck, you're a bad man. Seven plus six, you're going to make a profit on this $10 Ha, ha, ha." I get it up to nine, and as always, one of the students finally says, "Wow, this is a really great lesson. Yes, I will pay $10 for a $10 bill."
That guy always says it like he thinks it's over. It ain't over, because you go to the person with nine and say, "No, this seems odd." The decision confronting you now is you either eat a $9 loss for sure, or take a chance. But if you say 11, notice how I frame that as a loss? You eat a $9 loss, for sure. I just framed it as a loss and put this person in risk-seeking mode. Or take the chance that if you say 11, this knucklehead's not going to say 12, it's a $10 bill. You know what that guy says, every time? He says, "11," and then you turned to the guy holding the 10 and say, "I know this seems odd, but here's the decision that confronts you now. You need to eat a $10 loss, or you say 13 to prevent." And that guy says 13, and then these two people will go up. If you can get them to go to 19, they'll often freeze at 19. They'll have a hard time getting to 20, but if you push them over 20, they'll go to 29.
The most I ever got, I was actually doing an executive development in Kellogg, and it's the nicest people you'll ever meet. I got this thing up to 19, and I just wanted to see if I could push to 20. There was this woman, Sarah, and Sarah just wasn't willing to go to 20, and she was going against this guy, Frank. I remember their names. Her girlfriend said, "Sarah, we're not going to let Frank beat you." They take out their purses and started giving Sarah money. Now, every guy in the audience goes, "Well, that's just bullshit," and they take out their wallets. Daniel, I am watching this room full of people taking out tons of money so it becomes a battle of the sexes where they fight for this $10 bill. Keith [inaudible 00:14:26], one of my heroes, University of Illinois, actually got $1900 for a $20 bill one time. This is a record.
No individual working alone would ever do this, but if you put them in a group context, there's your magic. It's not good magic. It's bad magic, and in an MBA class, an executive development class, you kind of have to teach people escalation commitment and why you need to have circuit breakers on certain decision processes. You made the initial investment, but you don't make the re-investment decision because you're wasted. You're done. Somebody with a cold, hard, calculated heart will make the re-investment decision. There's a bunch of things you can do. [crosstalk 00:15:05]
But I'm not done yet, Dan. I'm going to give you one more, because once I get all these monies from these execs, I don't want it and so I want to give it back. So we play something called an ultimatum game. The ultimatum game goes like this. There's two people. You make an offer, and there's an ultimatum. We don't negotiate. I can either accept your offer or turn it down, okay? So, it's an ultimatum. Now, winner of the game is the person, and usually at the end of the $10 auction, I have 30 bucks I want to give away. I want to give this money away, and you make an offer, and the person who wins the money is the person that can take the most value out of that thing but still get the other person to say, "Yes, I accept it." I will tell you, I usually get it started.
You and I are playing. I go, "There's $30 in there, Daniel. I'll take 29 and leave you one." How are you going to react to that? I'll tell you, all my exec students, they're like, "No, that's unfair." And then they say, "No." Then you go down to the next person. "How about 28 and two?" "No." "27 and three?" "No." I will have people literally say no to $12. And then you say to them, "Do you understand that you just violated every rational economic principle in the world? Your choice is between $12 and zero, and you took zero." And then the exec will go, "Hell yeah, because the other guy got..." It's like, "No." We call that counting other people's money, rather than making decisions about your own money. You're making decisions about other people's money. There's so many of these things that you know in advance, you set up.
I want to give you one more, and then I'll jump to your questions, because this is a really important one. In my business, have you ever tried to teach the Challenger Space Shuttle? It's very difficult. Thank God somebody came up with the Carter Racing Team. The Carter Racing Team is a case where basically you walk an MBA team or execs through a situation where they have to decide whether or not they want to race in a NASCAR thing. It's the last Sunday of the year, here's been your history, the car's been breaking down, you're finding the people who sponsor you are upset, and they lay out all of the contingencies of if you race and win, yay, this happens. If you race and you're competitive, this happens. If you race and you lose, that's okay, but if you race and the car doesn't finish because you've been unreliable, that's a disaster.
You walk all of these MBA teams through it, and in the end I will tell you Dan, every single one of these teams? They race. Then you have the greatest moment when you flip the slide and go, "Congratulations, you just lost the Space Challenger." Then, you show the Space Challenger blowing up. If you try to teach the Space Challenger, everyone looks at that and goes, "Well, what a bunch of idiots. Didn't they see the O-ring data? Didn't they see what the temperature was? Didn't they see where the wind was coming from?" The Carter Racing Team is the exact same data, but now you have to detect it in advance, instead of explaining it after the fact. Again, for most of my students, when number one, we tell them, "You just lost the space shuttle," and then they knew that I know that they were going to launch it, and now let's talk about decision making errors under high stress context in the face of previous failures.
All of a sudden, they're a little bit more open to listening about it, where if you try to teach the Space Challenger as the Space Challenger, it's like, "No, I would never do that. What a bunch of idiots. Those guys are stupid. What's wrong with the NASA people? Aren't they trained?" No, no, no, no. You would do it. That's kind of the bad magic, that these things are completely predictable. You can only get certain bad things out of a group that you couldn't get out of an individual. That's a lot of the fun of it, too.
Daniel Serfaty: Well, thank you for sharing all these stories. They are basically cautionary tales about that magic. Maybe it's a black magic of team at some point, because it has to do with re-emphasizing why it's really important to understand team dynamics and to try to put in place the right structures, the right processes, the right interactions in order to prevent those kind of groupthink phenomena that you described earlier. Group work polarization, or other situations in which people, at that point, do not optimize their utility functions but are trying basically to establish, to maximize some other function that has to do with the social hierarchy in the team. Who is the person who's going to win the auction, for example?
Now, in a lot of situations that I know our audience is going to find themselves, work teams, those things may not happen to the extreme that you can orchestrate in your MBA classes with your MBA students, but they do happen all the time. We see very often in meetings things deteriorate, and when you look at them in posterior you say, "Well, the team forgot what they were trying to do. They got into another situation." We know from history that this notion of establishing consensus too fast for the sake of consensus is actually dangerous. The Cuban Missile Crisis is a classic example that people talked about, where all these advisors basically reinforcing each other's mistaken beliefs.
John Hollenbeck: Yeah. I've got two things on that, before we leave that, because the Cuban Missile Crisis is kind of an interesting example. I do feel that we team researchers through a lot of predicting after the fact, and we often blame teams for the kind of things that we're saying here. In many cases where you don't have the counterfactual evidence. The things that I've been talking about, we know what the rational decision was. We know what the counterfactual evidence is, but in so many team contexts, because you didn't go a particular direction, you don't even know would have happened had we gone in that direction. It really does kind of promote, and the Cuban Missile Crisis was [inaudible 00:20:36] really got people used to this paradigm where some really smart person would go into the archives of some decision fiasco and then dissect it. Just like the Carter Racing Team, tell you all the things these idiots did wrong. We've got to really be careful.
That's why we need a science. We need a science where you kind of have to make your predictions in advance, and easy to predict the future after it's happened. Yogi Berra. The future's hard to predict in advance. I think we've really got to check... That's why scientifically, I kind of believe in quantitative science. I'm always a little leery of qualitative studies where people go in knowing what already happened, or kind of going in there with a particular angle. You've got to predict the future in advance, and so what I try to do with my classes is to show that some of this science is so magical. I can predict it in advance. I can build a whole lesson plan around it. That's how irrational you're about to be. Again, that's the fun of it. That's the magic of it. I love to teach this stuff. I love to research this stuff. I love going to work every single day. I can't wait to find out what we're going to screw up next, and then kind of fix it and move on. Yeah, I'm totally fascinated by all of that.
Daniel Serfaty: I stand corrected. I didn't mean that those historical example... I know that they're being taught as paradigms of decision making mistakes, or misunderstanding of the situation in teams or in groups. My point is that, and I remember we worked on some of those projects in the past in which when you do the forensics of something that was disastrous, where lives were lost, as we have many examples. In the military, for example, or in industry, and you interview the folks that were in the middle of that decision making process that are now [inaudible 00:22:23] being almost accused of being characterized as having made a mistake, but once you immerse them back in the same situation, they are all pretty adamant that, "Given what I knew at the time, with all the uncertainties at the time about the information and the time I had to make a decision, I would do exactly the same thing today." [crosstalk 00:22:45] for us to do Monday morning, quarterbacking on some of those critical decisions, that's really what I'm saying.
John Hollenbeck: I agree. And there was another. We were talking about language in the military, and the one part of the language in the military I've learned is the expression, "It happened on your watch." Now, we're not necessarily saying it was your fault, and so you might want to come back and go, "Yeah, it happened on my fault, but it wasn't my fault because boom, boom, boom, boom." Then they go, "We didn't say it was your fault. We just said it happened on your watch," which kind of still implies [crosstalk 00:23:16] fault. Just an interesting use of that language. It happened on your watch.
Daniel Serfaty: [crosstalk 00:23:21] the difference between authority and responsibility. We won't get into that right away. Maybe if we have time toward the end.
So we've discussed different aspects of the magic of teams and what sometimes we have to understand, that teams can witness extraordinary performance out of teams that we wouldn't have predicted, and sometimes it's the other way around. If you look back at more than 30 years of your own research, but also research in the field on some major concept or major findings that you think, "Wow, that really changed the paradigm. That really changed my understanding about teams." Can you share a couple of those with the audience?
John Hollenbeck: Yeah, I think there's two that really kind of jump out at me. Again, we've talked about it a little bit. Just the multilevel nature of teams, and how if you're not looking at the right place, you'd miss everything. If you're looking at the individuals, but not the dyads. If you're looking at the dyads, but you don't understand the triad, you've missed the whole thing, because you're looking in the wrong place.
It's so hard to be looking at two places at once. The most impossible thing to call in basketball if you're a referee is whether somebody got fouled on a three point shot or not, because to know whether it's a three point shot, you have to be looking at their feet. Where were they? To know if they got fouled, you have to be looking at their hand. It's literally impossible for you as a referee to tell [inaudible 00:24:43] as a basketball, to make that call, because you can't be looking in both places at once. That's one of the fascinating thing about teams, that it can be happening at all these different levels, and if you're not looking at the right place, and you got to have a little bit of flexibility, you got to go in. The plan was to look at this place, but nothing happened there. Let's look at the dyads. Let's look at the triads. Let's look at the team as a whole. Whatever. You kind of need that ability to explore, but right now within the field, there's kind of a lot of pushback about hypothesizing after the results are known.
That's considered not cool, that you had [inaudible 00:25:15], and now you're kind of just sniffing around, exploring. And sure, if you explore 100 things, five are going to be statistically significant at .05. You can see where excess snooping is a bad thing. Not snooping at all is a bad thing, too. How you find the right balance between this is hard data to collect, [inaudible 00:25:37], you've got to give us a little room to breathe. We're not going to be able to predict every single thing in advance. If we could, we wouldn't be able to publish it, because it's already such an established part of the knowledge base that their teaching it in MBA programs. You can't do that now. That's been done. We know that already. If you're on the frontier of it, you don't always start out looking at the right place. You've got to have the freedom to be able to look at these different kinds of levels. I would say that was thing number one.
The other thing, and again, it was probably [inaudible 00:26:03].
Daniel Serfaty: Before you go into that, John, just a clarification. I want to hear about the second milestone, so to speak. Is that just a question of where you put the spotlight, or is it also a question of the granularity in a sense that you cannot predict the behavior of material just by looking at one molecule or one atom?
John Hollenbeck: Absolutely.
Daniel Serfaty: What is that? Is that...
John Hollenbeck: Daniel, I just love that example for two reasons. One, we worked at the Facility for Rare Isotope Beams, which is the newest linear accelerator being built in the United States. The United States builds a new linear accelerator every 20 years. It's an $800,000,000 project, so it's kind of a big deal. I'm working with the project director, who is a nuclear physicist, and we're trying to explain to him the importance of a triad. There's these two people, and you might think these two people have a relationship. You can try to predict what's happening between these two people. Either their individual characteristics, or the characteristics of their relationship. I want to talk about a [Simmelian 00:26:58] triad. A Simmelian triad is there's a third person that has a strong connection to both of those people, and it totally changes that relationship. You cannot understand that dyad without understanding this thing.
Now, I say this to Thomas Glasmacher, who's the director of the Facility for Rare Isotope Beams, and his whole face lights up. He goes, "John, John, John, that's just like boron. If you put boron next to hydrogen, nothing. But if in the same thing you put carbon, boom. All of a sudden this thing..." And just to watch his face light up, and he immediately as a scientist understood that yeah, triads are a different thing. In his organization, we studied the scientists in his organization and we plotted the informal organization chart and it allowed us to count all the dyads and the triads. If I had you, Daniel, and I said, "Okay, Daniel, who are the important dyads in your life?" You could probably do it. A lot of people don't even know the triads that they're a part of, because you may not know that this person is related to some other person.
Yet, triadic influences happen all the time. I've got a close colleague who's a department head at Arizona State, and one of my graduate students is an assistant professor at Arizona State. Yay! Now, they've got a dyadic relationship. They know me, one's a department head, one's a young assistant professor. But I talk to both of them every single week, and they both had me as a major professor, and they are both tied into the Michigan State mafia. You cannot understand the relationship between those two people unless you know the third-party guys. We're doing a lot of work with third party guys, and that was something that I think now, we just have the paper coming out in JAP, that's another one of these things that's kind of invisible to people. They just look at the individuals, the dyads, whatever. But nobody's paying attention to the triads, and yet that could be where the magic occurred. If you're looking at the dyad and it's happening at the triad, you missed the magic.
Again, why teams are so much fun, and why it's so complex, you only have so many things you can measure my friend, in a study. You want to measure a bunch of individual difference variables, you've got a lot of that, but that doesn't give you a lot of room to measure dyads. If you measure the dyads, and if you want to get stuff at the team level, so many variables. So many levels, and so few degrees of freedom. You have to make really hard calls going into this thing. Where do I really believe the action is? Because if I spend all my degrees of freedom at the wrong level, I'm sharing a bunch of null results. So you got to build in enough opportunity that you measure things at the dyadic or triadic level that if something happens there, you at least have a chance to get it and that's where the challenge is. It never gets boring and it never gets easy. And it never gets predictable.
Daniel Serfaty: But isn't there some kind, again if we stay even within the realm of research, I think it has direct implications of the way you design work teams, project teams at work if there is a combinator or explosion of dyads and triads as in, the number of people in a team, no matter how they organize it [diarchically 00:29:53] or hierarchically. If you have a team of six people, you have a lot of triads in there and you have even more dyads. And so the question is, is there a way, does the theory, the multi-level theory or any other model, give us a way to watch for those two or three triads that are really important or that can explain most of the performance of the team?
John Hollenbeck: Yeah, having it in our 1995 paper, which funded out of the A2C2 program, those were the first papers that actually looked at multiple level. A lot of people thought multi-level analysis came out in 2001 and that's when people really started getting good at it. In 1995, we were doing it and because we were in a lab context with people randomly assigned to conditions, and everything was [inaudible 00:30:35], didn't have a lot of pressure on the analytics. But the analytics get super complicated when all of a sudden all these things are correlating together and you got to find where is the variance on that.
In our '95 paper we just said, "Okay, what is the single most important thing at this level? What's the single most important thing at this level? What's the single most important at this level? How do those combine to the team level?" That was one of the initial forays into it. What's the most important thing at the decision level? What's the most important thing at the individual level? What the most important thing at the dyad? That was a way to kind of keep the number of variables to three, but then [inaudible 00:31:07] you say, "Okay, what's the five most important things in every one of these levels?" You've just run out of degrees of freedom when you have too many variables and not enough research participants. So that's a challenge.
Daniel Serfaty: Yes, well that's certainly food for thought for our audience here, because there is a subset in our audience that are researchers in teams, but there is probably most of the people in our audience work in different teams or study in different teams, and they are probably sensitive to that notion of the complexity of the dynamics when you look at two people. But when you look at three, and then you look at multiple threes in a team but [crosstalk 00:31:40]
John Hollenbeck: And it's not fair, it's not fair. If you've got a real life job, I mean you got enough balls in the air that you got to juggle. You're not like me, I don't even have a full time job. I mean, I can study these things at levels, levels, levels, but if you've got a full time job, you can't study... You might not work at the dyadic level up until you been in the team where this is a really good team except for Frank and Sarah. Frank and Sarah, they always go off, every meeting is... Okay, now that's a guy who figured out that dyads are important. And that, "Hey man, Frank and Sarah are on their own level, but whatever you put these two together on a team..." Okay, that is a person who now all of a sudden sees dyads in a way that they didn't see before.
For a practitioner, it's usually when something goes wrong, that this should have been a really good team. Why is this not a good team? Oh, I'll tell you why. Because Frank and Sarah won't let us get from point A to point B. Now they see it. And so it's definitely unfair to ask people that are doing real life jobs to be able to manage all the complexity of this, but this is why we have graduate education. This is why we have executive education. It's to point people to these kind of specific things you can use. That's established knowledge base and, as you know as a researcher, half of my life is dedicated to pushing the future knowledge base that many people will be teaching tricks 20 years later from stuff that we did in [inaudible 00:32:59].
Daniel Serfaty: Yes, you mentioned there are two things, two milestones, so to speak, that you can think of that really, redirected or changed the way you were looking at team. One is that notion of multi-level and the multi-level theory that you proposed to the world. Was there another one?
John Hollenbeck: Yeah. multiteam systems, which I think we'll talk about more.
Daniel Serfaty: Define them for our audience [crosstalk 00:33:20].
John Hollenbeck: Yes, okay. multiteam system is three or more teams that work interdependently with each other. You can imagine why the three-alarm fire in the old days, three-alarm fire meant three different fire companies showed up. You would take the north, I'd take the south, and he'd take the east. We'd just cross our fingers and hope to God we weren't spraying each other with water. There was a real lack of coordination.
So multiteam systems are teams that have to work together. They're increasing. Same reason that we could talk about historical reasons why organizations aren't built around teams now more than individuals? They needed to have greater scope. They needed to have greater specialization, and teams allowed you to have more scope and more specialization than individuals. You just keep pushing on that frontier, and now there's stuff that the team can't do either. The team doesn't have enough scope or specialization. Now you create the multiteam systems.
Once we started working on these, this was late in my career, I started looking back. I came to the conclusion that we were building a science of stand-alone teams. That is, teams that work, five of us, we work independent of anybody else. We're a problem-solving team, we're a project team, we're this team. But we never interact with any other team. We just got to solve our own problem. If you think about why that is, from a researchers point of view, we can find 80 teams that might be doing the same task. You can bring 80 teams into the lab.
So we were looking where the light was good, because you can do this. But once we started getting into the multiteam systems, Daniel, it's like all of these things that we consider best practices turn out to not be best practices if teams are interdependent. I would teach my MBAs, "Your teams need to be empowered. They've got to make their own decisions, they got to have the adaptability to break right if [inaudible 00:35:02], if they break left, they're going to go left. Stay in a pattern. But you got to empower these teams."
Okay, in a multiteam system, empowerment looks like unpredictability. Like, "Dude, you're never where you're supposed to be. We planned that you were going to go left here, and next thing I look up and you're going right. What's that?" "I was empowered." That'll go great. Your empowerment is destroying me. One of the things that we frequently learn that relative to what we teach our MBAs about empowerment, these multiteam systems need to be a little bit more centralized than that. Implicit coordination. We teach that in teams, implicit coordination is great, because me and Daniel and Debra, we've worked together for so long, we don't even have to do a lot of talking to each other because I know what Daniel's thinking, Debra knows what I'm thinking. You watch us and you see this tremendous coordination without any communication.
In a multiteam system where I don't know what your team's doing and why? You're clearly opaque to me. So the lack of communication makes it really hard for me to coordinate with you because I don't know what you, Debra, and John are really thinking about because you never really articulate it. When I ask you to articulate it, no offense, but you're not that good at articulating it, because it's implicit cohesiveness. We build cohesive teams, baby! That's our job. In a multiteam system, teams have to sacrifice for other teams, and the more cohesive the little component teams are, the more they love each other. The more they are unwilling to sacrifice for the larger part of the multiteam system.
In teams, open communication structures, everybody should be allowed to [inaudible 00:36:36]. Multiteam systems, man, just the number of links. 15 people? 15 times 14 divided by two, that's 90 communication links. We can't have 90 communication links firing. You need a level of communication discipline that we haven't seen. I got a dozen of these things that, when I look back, we've been building a theory and science of stand-alone teams. And the minute you put a team in an interdependent context, some of that's bad advice. We are trying to rewrite the rules of teamwork for teams that are part of multiteam systems. It's not just me. It's people like [inaudible 00:37:13] Church, Steve [Zakarao 00:37:15], John Epp. There's a lot of people trying to rewrite their rules of teamwork for multiteam systems and we have a long way to go.
Daniel Serfaty: Thank you for that insight. Indeed, it's not just that it becomes more complex when you look at multiple teams that have an overlap or an interdependency. The very same principles of good teamwork can actually be reversed, as you said, when it comes to working in harmonious multiteam systems. We're all part of that. What I like with the multiteam systems, as a system engineer myself, it's very appealing, this notion of system of systems. But from a human perspective, a human performance perspective, I think it's fascinating because you can look at it almost as the teams that are working with each other to some interdependency in tasks, but also in peoples.
Which implies that an individual worker, manager, engineer, is part of multiple teams as an individual. If you focus on individual, and therefore, her or his behavior is adapted to those multiple teams because they migrate between the different corners of that multiteam system. That by itself is interesting in terms of selection, training, and certainly team composition.
John Hollenbeck: And again, just for our listeners, I know they want to keep a distinction between people that are on multiple teams. This is multiple-team membership. I can be on three different teams and that definitely creates challenges. But a multiteam system doesn't necessarily have to have that. I'm only on one team, you're only on one team, Debra's only on one team, and we're good. Now if you combine that with not only do we have that, but Debra's also on this other team and you're on this other team, and so we have both multiteam membership embedded within multiteam systems.
Daniel Serfaty: That's the story of my life as a CEO, John. That's what I do. [crosstalk 00:39:05].
John Hollenbeck: And again, what is the level of complexity? One of the questions that we were talking about before is what's the single most important thing with multiteam systems? I tend to punt at questions like that, because I'm a college professor. But I will say this, whether teams or multiteam systems, you got to boil this down. The single most important thing in teams, and this is kind of ironic, single most important thing in a team is individual accountability. Like, "Daniel, do your own job. Debra, do your own job. John, do your own job." Especially if we have our own specialization, so that I really can't do your job. Or that I have to do my job differently because of the way you do your job. All of that stuff creates this bad magic where these five individuals are so much less than the sum of their parts because of that.
In a multiteam system, we're kind of seeing the same thing in that the most important thing in a multiteam system is, each team has to focus on their own job. Don't worry about the multiteam system. Just focus on your own job. You focus on your own job, that'll put us in a position where maybe we can gain synergy. But if you don't do your own job, I must. We're going to watch the Super Bowl in a couple of days, okay? Offenses and defenses need to support each other. The other team's not going to score. If I'm the defense, and you're my offense? Guys, go put the ball in the air. Don't let their defense score, because if you let their defense score we're not going to... Or if I'm on offense and I can score 30 points? It's like, "Yo, defense. All you got to do is get a few turnovers."
So the teams can kind of support each other, but if they don't start from doing your own job, and I know my defense is bad so I'm going to have to score 35 points? All of a sudden we're doing things that we wouldn't normally do as a team, because of those knuckleheads. And then those knuckleheads are doing something because they don't trust us. Do your own job. Single biggest thing is accountability. That was at the team level or the multiteam system level.
Daniel Serfaty: I am personally a believer of that. When we come back from the break before we jump back into your new ARI program, I want to jump back about some comments you made about sports team. But I am a big believer of that. As you know, we used to have the Patriots, the greatest team on Earth here in New England.
John Hollenbeck: That's just Tom Brady, I hate to tell you.
Daniel Serfaty: Well, but he was believer of at least the mantra of military which was, do your job. You do your job, we decompose the jobs in such a way that we will put them together as a team.
John Hollenbeck: Exactly.
Daniel Serfaty: Just do your job. That was the mantra for many years.
John Hollenbeck: At the highest level, trust [inaudible 00:41:33] to have a good plan. So just do your own job. If you don't trust that, then all of a sudden you start not doing your own job. If you go into the Patriots training facility, that expression, "Do your own job," is everywhere. Every time you turn a corner there's a sign that says that. I think there's some real [inaudible 00:41:49]. I think if I look at organizations right now and you would think that because I'm a team researcher, I want teams everywhere. But I really believe that right now, I look at a lot of organizations, their biggest problem is open embeddedness. People are part of too many teams. They make a decision that, "Oh, it might be nice if this person was there," because that person may [inaudible 00:42:09].
Or they create a multiteam system. Maybe these teams should meet. Now you've just committed 15 people to have to meet because it might be worthwhile for these teams to coordinate. Often, after making a single mistake when there was a lack of coordination, but 19 times out of 20, these guys are well coordinated. One out of 20, there was a lack of coordination and now we have the meet.
The one thing I would tell your audience, the people who are practitioners, be stingy with how you create teams. Be stingy with who's on the team and who's not. The test is not it would be nice to have that person, versus, "No, this team really can't function without this person or this person's specialty," because you might think, "What's the harm in putting Debra on this team?" Oh, we're putting another team. What's the harm of putting Debra on this team, too? Hey, you know this team over here? Debra might have some interesting views on that. And often it's not the same person. It's three different people who don't even know other people are putting Debra on teams [inaudible 00:43:10] people.
And the next thing you know is Debra's running around from one meeting to another, and if she's not taking notes or [inaudible 00:43:17], you shouldn't be at this meeting. If you ever find yourself in a meeting that when it's over, you didn't take notes and you didn't talk? You shouldn't have been there, because the biggest problem with over-embeddedness is that it prevents individual accountability. I can't do my own job, because I'm going from meeting to meeting to meeting to meeting. My job is teaching. I didn't write a word today, because I went from this to this to this. Or I didn't get my homework graded, and I'm a teacher, because I was going from this to this to this.
If the single biggest thing that you need is individual accountability, the single best way organizations can support that is not creating over-embeddedness. If somebody tells you, "Dude, I'm on too many teams," they might be right. So try to avoid that [inaudible 00:44:01].
Daniel Serfaty: I think this is a beautiful, right on, literally right on, concern that many enterprises have these days. More these days, in the times of COVID, than in any other days. Because precisely the barriers of forming teams have disappeared. It is basically how many people you can get on Zoom. And so, we do that... I leave that every day by guiding my team leaders, my managers, to be, I like the term stingy. Don't create a team for everything, and when you create a team, be very stingy on the number of people you're going to bring to that team, because otherwise you hit a fragmentation level that basically start having decreasing return on productivity.
It's a real, real problem today, and any guidance that had come from the researchers or leaders in teams research regarding how to form those teams and how to form them in a way, just in that right middle when there are just enough teams and enough people on the teams, but not too many. Right now, it's an empirical. We have some empirical rules, but it will be great to be guided by theory.
John Hollenbeck: Richard Hackman, 5.4. Ideal team size, 5.4. That's five adults and an eight-year-old. I actually don't know what that eight-year-old does, but it can't hurt. And just to reinforce, there's two things that need to happen at any high-level organization. Yes, you need collaboration. That's true. But you also need individual concentration. If you're working on a complex job and you need to be accountable, you often need to be able to shut the door, turn off the phone, turn off your email, because what you're doing is difficult and it requires concentration.
What I would tell managers, and I've seen some organizations [inaudible 00:45:43] budget, that if you just spent collaboration dollars on me, you just put Hollenbeck in a meeting that he didn't have to go to before. But you owe me some concentration dollars. Where are you giving me time back that I can concentrate on my job? Because left to your own devices, or the uncoordinated devices of 17 different people that put me on 17 different committees, and don't know it? You just destroyed my ability to do my own job.
So anyway, enough about over-embeddedness. [crosstalk 00:46:11].
Daniel Serfaty: With that, I think it's right on. I hope our members of our audience are going to heed that advice. We'll be back in just a moment. Stick around.
Hello MINDWORKS listeners. This is Daniel Serfaty. Do you love MINDWORKS but don't have time to listen to an entire episode? Then we have a solution for you. MINDWORKS Minis. They are curated segments from the MINDWORKS podcast, condensed to under 15 minutes each, and designed work with your busy schedule. You'll find the Minis, along with full-length episodes, under MINDWORKS on Apple, Spotify, Best Prout, or wherever you get your podcasts.
I want to ask you a question as we talked earlier about the magic of team. The last question before we move onto the futuristic multiteam systems, and artificial intelligence and everything else. You come from a long tradition and a family of coaches and coaches of teams, and you've seen, especially in the sports area which is the number one metaphor in America when people talk about teams, immediately people talk about sports teams. They know the great sport teams, and then the sport teams of history.
Tell us a little bit your perspective about that. Your own observation about what, especially in the sports area if you decide to choose that, what has made good teams great?
John Hollenbeck: Again, in terms of how I got started and interested in teams, as you said, my family is coaches. My father was a coach, my brother is a coach, my son's a coach. We're constantly talking about things like that, because it's always in the forefront for us and almost all the discussions, Daniel, are about synergy or process loss. How this team was so much better than you'd think, or how this team was so much worse than you would think. That's about half the conversations we have about teams.
And then when we try to dissect it after the fact, and I do think sports resonates with people because for a lot of people that's something that they share, it's very public, and so they've seen it. In my MBA class, I always talk about the 1972 US basketball team as an unbelievable team. I actually have their names written down here, so it kind of helps me remember them. Here we go. Tim Duncan, LeBron James, James Wade, Carmelo Anthony, Allen Iverson, Amar'e Stoudemire, Carlos Boozer. Every name I just mentioned there is in the NBA Hall of Fame. Hall of Fame. They lost three games. They got beat by Puerto Rico, Lithuania, and Argentina.
Somebody joked that the gross national product of Lithuania was smaller than the combined salaries of the US Olympic team. I don't know if that's true or not, but that team's been dissected. And then of course the flip of that is 1976 US hockey team that won the gold medal, and not a single one of those guys went on to be a pro. Literally. I mean, to really make a living as a pro, let alone Hall of Fame. And so again, these are just extreme examples of that.
The reason I love this is because anytime I give this example in class, and our MBA program is very, very international. [inaudible 00:49:06], it's very, very international. I will tell you, somebody will come up and they've all got their own story about the 1987 Chinese ping-pong team.
Daniel Serfaty: Oh, from their own country.
John Hollenbeck: From their own country. And they're like, "Well..." And they'll talk about it. You know, I'm not a big ping-pong guy. This person's obviously super into ping-pong. And the 1987 team was so much less than the sum of their parts for reasons I don't know, but this Chinese person is talking to me and it's like... Or the 2005 Canadian curling team. I don't even know where this story's going, but I know it's going to be a story of synergy and process loss. "Oh my god, the 2005 Canadian curling team, they shouldn't have done anything! These guys didn't know what they were doing."
And then you put them together. The 2009 Italian [bocce 00:49:46] team, again, I don't even know where this story's going. But the minute this person wants to talk about the 2019 bocce ball, I know it's going to go one direction. I just think this is a trans-cultural phenomena where we all get it. We all get the synergy process loss thing, and we've all got our own favorite example of it. And even people that don't really follow sports will have an example of it of... Another one is you see weddings. The five maids of honor at this wedding are the most beautiful, wonderful people in the world. But man, this was [inaudible 00:50:20]. Or that these five groomsmen are the five greatest guys. This should have been the greatest party of all time, but all we got was a fight in the parking lot.
Everybody's got the story of individuals that just came together and man, you didn't get anything to predict it. At Michigan State, I always talk to my department head about the perfect meeting. A perfect meeting for Michigan State is absolutely what you thought was going to happen at that meeting, happened at that meeting. No surprises. No dynamics. No synergy, no process loss. It's just a meeting, let's get in there, get out of there, nobody get hurt, [inaudible 00:50:55].
Anyway, yeah. Sports teams are definitely attractive for that reason. It just really resonates with people.
Daniel Serfaty: Thank you. That's funny. People describing what's the best meeting ever. That would be a great discussion [crosstalk 00:51:08]. Let's turn the dial a little bit toward the future and maybe it is linked to a new form of multiteam systems. This last year of confinement, Zoom work, distributed operations, has basically created this notion of, and I believe we don't have a word in the English language yet for that. A network, a team, a multiteam organization, that because we have lowered the barriers of composing some work organizations, we created these new entities. They're amorphous. Some of them are more structured than others. I wouldn't call them a team, but I don't know how to call them.
My son plays video games with folks he's never met in his life, and he plays them every week. So they meet every week to play together as a team. But maybe it's not a team. It's a [inaudible 00:52:05] of social structure that has been enabled by technology. Are we looking at the dawn of a new way by which human organized to accomplish the goal?
John Hollenbeck: Yeah. Well Amy Edmonson has actually coined the term 'teaming' to get away from the word team. Really what you see at organizations is teaming. It's much more verb, and the word teaming, if you think about it, means it's just teaming with this or teaming with that. Again, it is these really kind of unstructured teams. For your audience doesn't know this, but Daniel and I, another thing that we share is, we both have twins. I went in our basement, my twin boys were playing Halo with, like you said, a bunch of strangers. People were swearing. It was like, "Who are you playing with that are swearing?" And they were like, "Oh, we're playing with these guys from Australia."
So they're in my basement in East [Lancing 00:52:54] playing Halo with a bunch of guys from Australia. For all I know, they could be a real-life SWAT team. Who knows? But like you said, the barriers are so low now that all of a sudden we're playing with different people. Yeah, I definitely think that's part of the group over-embeddedness problem. The fact that it's all technologically mediated makes it harder, too, because a Zoom meeting has a lot of features that, for evolutionary reasons, are not really good for people in the eye context, not good. The head sciences aren't right. It doesn't really simulate being in a room with three-dimensional people in a way, and it's very, very tiring for your human brain to try to process the fact that this is not a normal group situation for your human brain, but your human brain's trying to make it like it is one. It's extremely fatiguing.
Daniel Serfaty: Why is that tiring?
John Hollenbeck: It's just not natural. If we're in a group meeting, all eyes aren't on me. But if I'm in a Zoom meeting with nine people, I'm looking at nine faces all of which look like they're staring at me, even when I'm not talking. You all are staring at me, when you're not. Where if we're in a real-life conference room, I can see that you're looking this way, Debra's looking at [Irash 00:54:02], Dan's on the internet, these two guys are talking to each other. Again, it's just not natural and because it's not natural and we're trying to make it natural, it becomes tiring.
I do think what we're learning with Zoom meetings as [inaudible 00:54:13] is size. Group size is not the number of people, it's the number of communication links. So a group of five is five times four divided by two. That's 10 communication links. Okay. You double that. 10 times nine divided by 2. That's 45 communication. A Zoom meeting with 10 people on it, especially if you follow my rule that if you're not taking notes and you're not contributing to the conversation, you shouldn't be there? That's tough. 10-20? Forget it. The communication links just kind of explode.
I do think we're relearning lessons about team size, and that team size needs to be a lot smaller. I think the other thing we're learning yet again, this is Richard Hackman, and Amy Edmonson is a student of Richard Hackman, so I think she gets this more than the average person. But he was a big fan of boundaries, team boundaries. A team's got to have a hard boundary, and people can't just come in and out of this team like it's Grand Central Station and pop in and pop out, because it's very dysfunctional. So he was a big fan of not just small teams, N=5.4, by the way, was his number. But also teams with really strong boundaries.
So like, okay. We may want to let the other people in. But not routinely, and I do think that the concept of teaming recognizes that we've sacrificed a lot of that and that we just throw teams together all the time. It takes a long time to build a team. They've got to to through stages of development. If we interrupt that every time, we shake them up. And so it's ironic because people believe in teams, and that's why they're constantly forming them, not recognizing that they're engaging in very self-defeating behavior because they keep stirring up the teams that they're trying to build where they just really need to leave them alone and respect their boundaries, and maintain their boundaries.
Daniel Serfaty: I think this notion of boundaries is important. I have so many questions about that. I want to make sure that I don't ask all of them. But I think since I've seen that as you have probably evolved the notion of maintaining team size for meaningful interaction, and by meaningful I just don't mean productive interactions but also meaningful in terms of does it enrich me to interact with that person? In more and more meetings I go to now, virtual meetings, they'll use breakout rooms to make those teams, to control basically team size through Zoom device to break people up into four or five people teams, has been used almost empirically by people because they wanted to get some work done because they realize that team size have to be controlled.
John Hollenbeck: I do a lot of online teaching, and one of the first things they teach you in online teaching is you can't just go, for my MBA class, me against 40. You must break out into breakout groups, because otherwise it's so much harder for the students to have... The students think there are 40 faces looking at them, which is not true. But you go into these breakout rooms. I not very good at technology, and I go to a breakout group it's like, "Oh man. I hope they come back." I lost two MBA students, the FBI's looking for them. We went into a chat room, they never came back and Daniel, I hate to say it. We still don't know where those kids are.
Daniel Serfaty: They are lost in cyberspace somewhere.
John Hollenbeck: They're lost in cyberspace. And so yeah, chat rooms are important but make sure you get every single kid back because otherwise there's going to be some hard questions asked.
Daniel Serfaty: I want to turn the page on this one, because I think that at the end of the day, I wonder, it's really a research question almost or a philosophical question even, more than a research question. I wonder if perhaps our generation concept of the social structure called teams, which has been essential to our professional, personal development, professional success, is basically disappearing. It's disappearing because our next generation, your twins, my twins, are much more tolerant of surface-level connection, and because they are tolerant of that surface-level connection, they are not superficial. That's the way they conceive of that connection with that coworker in Finland, or that coplayer in Australia.
They can sustain many more of those connections because they are not as deep as we think they should be. Therefore, I think this notion of teaming will be interesting for the next generation of researchers to see whether or not there is actually an age different between the digital natives, the people that grew up with the network, and people who did not.
John Hollenbeck: I hope you're right. I will say that as an evolutionary psychologist, that brain that you have in your head and the brain that I have in my head was basically developed two million years ago, and it doesn't respond overnight to changes in technology. You have a hunter-gatherer brain. So do I. So I worry about it. I will tell you my son, Tim, one of my twins, he thinks he has 10,000 friends. Well Tim, see if any of your 10,000 friends are going to come help you move this week, because I'll bet you go over 10,000, see if anybody will let you use their pickup truck. He doesn't have 10,000 friends.
I had a student come in because one of my classes, I take their technology away. Especially I have a class of freshmen, and it's kind of like Independence Day when they see the aliens and it's like, "Well, if you take away their technology, they're just like humans." That's what I'm learning about freshmen in college, because I'm 63 years old. Even my kids are 31. These are aliens to me. But you must take their technology away, because you're teaching in a class and a kid's on his phone, he's on the video, and you hear the ESPN jingle going off. So i just shut it down. I go, "For 80 minutes, the 40 of us are just going to be together as individuals, shut off from the outside world, talking to each other."
One of these students came up to me. He says, "I understand where you're coming from. I understand we have to concentrate. I understand we have to focus." He goes, "But you just don't understand how good my generation is at processing parallel information coming from many different angles." He goes, "Dr. Hollenberger, you just got to give me a chance."
I was like, "Son, you had me right up until that Dr. Hollenberger thing, because we've been in class for five weeks and my name's Hollenbeck. I put it up on the board every single time." And so maybe we're seeing an evolution in human history, but I'm not seeing it in some of my students quite yet. But anyway.
Daniel Serfaty: Dr. Hollenbeck, you just perhaps proved that maybe precision is not a value in the future. [crosstalk 01:00:32] in family names, that's less [crosstalk 01:00:35].
John Hollenbeck: Just that big ugly guy that's up there talking. His name's not important. You know how I'm talking about.
Daniel Serfaty: We're talking about the future and the future evolution. I know that you want to ask some questions, but more and more, we are looking at teams. Again, we have to invent a whole new vocabulary, I think, because we are unable and constrained by our own language. More and more we are studying around, my colleagues and I, with teams that are made basically of different entities. Some of them carbon-based, or human, some others artificial. They can be robotics, they can be artificial intelligence bots, they can be different kind of entity. In fact, for the past year, we had a new employee at Aptima. Her name is Charlie. She was the object, and the subject, frankly, of the first episode of this podcast series on MINDWORKS.
She's artificial. She has somewhat a personality. She helps doing all kind of things now, more and more. She was designed to be a panel member in a conference, but now she's also members of proposal teams. She co-wrote a chapter with her co-creators on artificial intelligence. Kind of interesting. And so we are creating those new teams for the military, for the hospitals, for even research and development companies like my own, that are made of different types of intelligence.
Do you think we are entering a new area, or is just same old, same old? In a sense that whatever we know about teams, even multiteam system at some point, is going to apply whether or not some of those members are not human?
John Hollenbeck: Again, I do think this is kind of a new area. I don't think a robot or an AI is treated by other humans like a human. So I do think this is kind of a new area. It's not like, "Oh, this is a team, but now this team has a child on it. Or this is a team, but now this team has a genius on it." I really do think it's qualitatively different, and I know you guys at Aptima have way more experience than I do. I have one experience in AI. Basically it was a company that was looking to compose teams, because you can image so much of the work is outsourced now. A lot of work is outsourced as individuals.
Yet as organizations are increasingly built around teams, you might want to outsource a whole team. So the idea is this organization's going to basically be an outsourcer and so they wanted to have an AI to learn how to compose really good teams from a bunch of individual outsourcers. Okay, we have all these outsourcers that we can draw from. Who's the best team from a team chemistry point of view?
What the AI is trying to do is what makes teams cohesive? What makes people good performers? And the criteria was like leader evaluations, or team member evaluations. Do you feel like this team did a good job? Do you feel like this team is cohesive? Did you feel like this person was doing a good job? [inaudible 01:03:23]. And so, the AI was trying to learn what goes into getting good evaluations and poor evaluations, but the thing was like a precocious child. It was learning all kinds of things you didn't need to teach it.
If there's bias in the supervisor evaluations, the AI learns the bias. The team is all homogenous group of these, and this person's an outlier, this person's a token. The AI learns that tokens don't work. The people that are building this thing, they say, "Oh, it's so beautiful because the AI doesn't have any human biases. It's just an objective..." It has human biases if you teach it! A child's not born with that. It has to be taught.
That was just my first experience, but we're increasingly looking at multi systems to reduce the number of people. It's tempting to replace people with AI things and so I would love to hear about what you guys have learned. How do people react to it? Are there a lot of differences to it? I mean, what's been your experience?
Daniel Serfaty: That's a good question. We don't have enormous experience. This is a nascent field, this notion of so-called human-AI teams. The first thing that we are discovering is that the paradigm or replacement is a wrong way to think about the problem. If you just say, "Oh, we're going to replace that node on the diagram with an AI," yes. That can work for some kind of tasks. I assume that that's less interesting. The more interesting insertion of AI in teams, as opposed to human-AI teams, is having AI do functions that you and I and team researchers have dreamed about somebody doing that function. Maybe an AI that is roaming the team, roaming the information system of the team, and find coordination opportunities, and then suggest that. Or find collaboration opportunities. Kind of an eye in the sky AI that can reallocate tasks, can reallocate information, can even suggest new types of or new forms or new directional collaboration that is not happening in order to, say, optimize a mission.
John Hollenbeck: Almost like matchmaking. These two people should get together. They don't even know each other, but if they did, wow would they hit it off. Kind of like that?
Daniel Serfaty: Something like that, and why would an AI like that be able to do that? Because it's an AI that has learned, that has absorbed a lot of data about teams. That's why I think my suggestion, this is just one insight that is emerging right now, because most people look at one human, one AI, and how do we optimize a dyad? As you taught us a few minutes ago, dyads is interesting. Triad is much more interesting. So just focusing on the one-on-one and then generalizing to larger teams is a dangerous thing to do.
We are discovering that, also with AI, the question I have is actually a plea to the team researchers in organizational psychology, in management sciences, et cetera, that these essential questions for our future, for the future of work, are too important to be left only to the AI engineers. Because in the example you gave, if AI is just given a free rein to do whatever they want, they're going to do exactly that. Principles of good work and good collaborations are going to disappear, and who knows principles of good work and good collaborations? Researchers like you. So I think it's very important that this field takes a plunge and say, "Okay, we're going to study that," because as I say, it's too important to be left to [inaudible 01:06:57] to a developer.
John Hollenbeck: Let me give you a really good example that reinforces that. We have a paper published on wearable sensors. We do a lot of things with wearable sensors now in terms of collecting data. There was a wearable sensor that they did a lot of social metric work to kind of capture who was talking to who, who was physically located, whatever. They were generating their own measures and whatever, and we were asked by the National Science Foundation to look at that and see what their applications were for both business and research.
The thing that we recognized right away were these engineers were developing an entirely separate science of teams. And so they would have wearable sensor measures of cohesiveness and what cohesiveness is. It would define it by what this wearable sensor's doing. Of course, we couldn't get it to correlate with any measure of cohesiveness that we have. We would measure groups and we would use the wearable sensor measure of cohesive, and it was like, virtually they had almost every construct they had, there was a construct that existed in the science, and none of those correlated.
You would literally have a science of teams being developed by the wearable sensor people. A 50-year science of teams in the pages of Journal of Applied Psychology. And those worlds don't come together at all. Your point that you can't leave it to the engineers? Oh my god, these guys didn't understand the most basic idea in psychometrics. Why would they? But they didn't understand a single basic thing about psychometrics. They would have all these metrics that all of a sudden they would just take a bunch of numbers, multiply them together. One of the equations had pi in it, 3.14159. As far as I could tell, the only reason that we were multiplying this thing times pi was because it makes it look scientific. But then, that was a separate variable for some other variable, and when we tried to do a factor analysis, the system crashed because we found out that actually these two variables were the exact same thing except this one's multiplied by pi.
And so they correlated 1.0, and it was like, "Guys. I'm sorry." It was kind of like they generated all of these supposedly psychometric constructs, and they don't understand. It's not their fault. What they do is hard. You can't know everything. You can't know all that stuff and know 50 years of psychometric theory.
Daniel Serfaty: I allow myself to say that sentence. It's too important to be left to engineers. I am an engineer, as you know, and that's precisely why I'm saying that because the matters are complimentary but you cannot substitute, basically, that people don't treat each other [inaudible 01:09:24]. I think it's very important, I think is probably the most important, more so than the internet, perhaps, transformation of the future of work, of what work is going to mean in the future. This notion of seamless blending of intelligences between the realm of the artificial and the realm of the human.
I think it's even more important to think of it as an interdisciplinary enterprise, which brings us back to that first project when you and I, many years ago, where the reason magic happened in that project, you turn it into herding cats. But it was herding cats, but when the herd worked it was beautiful to watch because then you had network theory experts and we had industrial organization, psychology systems engineer, mathematical modelers, experimental psychologists, eventually all working around the same questions. They had different answers, or different methods, [crosstalk 01:10:20] question.
John Hollenbeck: And it took time. I gave you a lot of credit, because you were kind of the physical leader of that group. But Bill Vaughn, as you know, Bill Vaughn really deserves a lot of credit for that group [crosstalk 01:10:31].
Daniel Serfaty: ... because Bill Vaughn was not with us anymore was, a division leader at the Office of Naval Research would have the vision to bring together multiple universities, multiple experts, top experts in their field, to study the notion of adaptive organization and adaptive teams.
John Hollenbeck: And so he brought in the math modelers, the social network people, the [inaudible 01:10:52] theory people, the lab people, and he was patient and gave us time. He recognized in the beginning, it's just going to be parallel play. Each of these guys is working on the same problems, and there's not going to be a lot of interaction. It's just going to be almost like children playing next to each other. But over time, because he was patient, eventually these things fed into each other and we were mathematically modeling some of the theories that we were building.
I just can't give Bill Vaughn enough credit for that to have both the vision and the patience to let that thing happen, because it took a lot of time for that to happen.
Daniel Serfaty: Yeah, it takes vision and many of us have been working with the government for a long time, with different agencies within the government. From time to time, you have a visionary leader like that who can sustain and understand that these things take time. I think in what we're talking about, this notion of blended intelligence work and the future of teams in that environment, the paradox is that one side of the equation evolves at such a fast speed. I'm talking about the artificial intelligence, the deep learning where knowledge becomes obsolete within 18 months or something like that. Therefore, that's the big paradox or the big challenge for this new enterprise, I think, to be able to synchronize ourselves to the fact that the very concept, when we say artificial intelligence can do X, X can become 10X in the next year.
Therefore that changes the problem, and that changes probably the solutions, too.
John Hollenbeck: And then as you know, Daniel, we know where this goes. Eventually, the artificial intelligence recognizes that the only threat to its existence is the humans. Dude, it's like you're building Skynet. Do you even recognize that you're building Skynet? I mean, this movie always ends the exact same way. How come nobody sees it?
Daniel Serfaty: Maybe then we need researchers like you, and the next generation of researchers that you taught and you trained, to be able to prevent us to reach a Skynet kind of model. But John, I have time for one last question. I wanted you to just briefly [inaudible 01:12:55] on a prediction on what we just talked about besides Skynet as a prediction, which [inaudible 01:13:02]. How do work teams and multiteam system look like in 10 years? What are some of the things that you see?
John Hollenbeck: It's hard for me to look into the future in advance, but I will tell you what we're doing and what we're committed to, and what we believe. We do believe that the reason we have so many team-based structures in Western societies is because it's impossible for businesses to compete on cost. Because competing on cost happens in nations where the labor standards are so low, we can't possible go there. Therefore we must compete on differentiation, we must compete on speed.
All of the things that pushed us [inaudible 01:13:37] job [inaudible 01:13:38] individual who could work alone, all by himself, without having to talk to other people. If it doesn't get roboticized, it's going to be sent off-shore so far you'll never see it again. And so, there's just increasing pressures in Western societies to be faster, more creative, more differentiated. You cannot compete on costs. So that's where we're going. If that's where we're going, then we're going to be doing multiteam systems, and as I said before we have a science of stand-alone teams for really good reason, and that it's hard to do research on multiteam systems. That's a problem.
I am part of that problem. I will tell you specifically why. In the beginning, there was multiteam research and then multiteams were like two teams of two, or two teams of three. I will tell you, in the beginning, people that you know, smart people like Ted [Salus 01:14:26], people like Steve Gonsowski saying, "There's no such thing as a multiteam system. These things are just teams." They were right, because a two-person team is not a multiteam system. It requires size, and it requires specialization.
We did a program that was sponsored by AFOSR where we worked with captains and squad officers. We built one of the largest databases ever on multiteam systems with 15 or 16 people. We would go to AFOSR, I would teach in the morning. We had 450 people. And then we would run 31 teams, 15-person teams, in the afternoon. We would do that for three days, and we did that for seven years. It almost killed me. But this is not the kind of paradigm that the average person can do.
I was at a conference one time, and somebody said, I literally heard him oversay, "Oh, Hollenbeck ruined multiteam systems. Because now you can't do it with two teams of two, or two teams of three. You have to have 80 teams of 15 people, which nobody can do." He says, "Hollenbeck's literally put the industry out of business because he has a paradigm that nobody else can do. He has access to these 50,000 captains or whatever." And so I did feel a little bad.
We just got a grant today, I'm very excited [inaudible 01:15:35] about, with ARI. The purpose of the grant is to help the National Infrastructure for Multiteam Research. Lower the bar. This will be housed in the beginning at Michigan State, but later we want to distribute this. The idea is we will have a system that if you show up, Daniel, with 80 teams and Debra shows up with 80 teams, and I show up with 80 teams, which most team researchers can do? We are going to put you into a multiteam system. All of a sudden people that couldn't do multiteam system research before, can. All this data will become public, all of this data will be at a repository.
Obviously there's going to be some negotiation, because you have this angle that you want to study and Debra has this angle, and I have... But we will use those angles to try to create, "Oh, here's a good multiteam system. Daniel and Bill [inaudible 01:16:23]. They want to do this. We should put these guys together and we'll run 80 of those." [inaudible 01:16:27].
Now eventually, the first three years is just Michigan State, Penn State, and Arizona State. That's my colleague Jeff [inaudible 01:16:34] at Arizona State, and Steven Humphrey at Penn State. But a lot of teamwork is going on there. In the third year, we're looking to branch out and we're going to try to find three other universities that will partner with us. Beyond that, we'd really like to put this distributed. I'm 63, Daniel. This would be my legacy, baby. If we can create a national infrastructure for MTS [inaudible 01:16:52] that I can walk away from and that thing's self-regulating and self-sustaining and self-operating? We will have a scientific evidentiary base for multiteam systems that A) we do not have now, and we will not have without this kind of infrastructure.
Again, Greg [Roark 01:17:09] is basically supporting this. I do think he gets it. He gets that this is going to be a legacy that if we build it, they will come. If they come, we'll have something that we wouldn't have had four years ago. If I'm responsible for destroying MTS research, perhaps I can be at least partially responsible for lowering the bar and getting people back in it with the help of our friends at ARI.
Daniel Serfaty: I think this is such an exciting prospect. Thank you for sharing that with us. Congratulations, first, on obtaining this grant with your colleagues at the other universities. I think if anybody can do that, John, it's you, because you have enough reach to the right, to the left, in terms of the different types of research that are out there. This is part, for our audience, of the larger trend in scientific research in which massive collaboration, multi-institution collaboration, will provide the data and bring the data to a much larger, worldwide audience.
We'll have a podcast in the future about the future of those massive collaborations. But John, thank you so much for sharing all these wonderful stories with us today. Best of luck on that new enterprise. Maybe in a couple of years I'll have another podcast with you and how you succeeded.
John Hollenbeck: I want to thank you for the lovely gift of this microphone. I'm going to keep it and cherish it. I'm going to put it next to my Aptima mugs that I took from your office several years ago. I love the Aptima mugs. Just kidding, my friend. I'll mail that one back. I'm just kidding.
Daniel Serfaty: Thank you, John.
Thank you for listening. This is Daniel Serfaty. Please join me again next week for the MINDWORKS podcast, and tweet us @mindworkspodcast, or email us at [email protected] MINDWORKS is a production of Aptima Incorporated. My executive producer is Ms. Debra McNeely, whose name you've heard several times today. My audio editor is Mr. Connor Simmons. To learn more or to find links mentioned during this episode, please visit aptima.com/mindworks. Thank you.