WEBVTT 00:00:04.972 --> 00:00:09.452 Hi, I'm Paul Verschuren today with my colleagues Jenna Bednar and Andreas Rupstorf. 00:00:09.672 --> 00:00:13.692 We're speaking with Robert Axelrod about the foundations of collaboration. 00:00:15.152 --> 00:00:18.572 Professor Robert Axelrod is a political scientist and the William D. 00:00:18.612 --> 00:00:23.012 Hamilton Distinguished University Professor Emeritus at the University of Michigan. 00:00:23.672 --> 00:00:28.632 Robert has been the author of many scientific articles and especially the 1981 00:00:28.632 --> 00:00:32.572 paper The Evolution of Cooperation, co-authored with Bill Hamilton, 00:00:32.572 --> 00:00:35.192 is considered a classic in the field. 00:00:35.892 --> 00:00:40.852 This is the Convergent Science and Instrument Forum podcast on collaboration. 00:00:41.532 --> 00:00:47.172 And today, me, Paul Verzsior, with my colleagues Jenna Bettner and Andreas Rupstorf 00:00:47.172 --> 00:00:51.532 are talking with Robert Axelrod on the topic of collaboration. 00:00:51.932 --> 00:00:55.452 So welcome, Robert. Thank you. Welcome. Thanks for inviting me. 00:00:56.032 --> 00:00:59.452 So, Robert, just to start off. 00:01:00.572 --> 00:01:06.672 Could you give us a short description of your professional trajectory to give 00:01:06.672 --> 00:01:10.252 us a bit of a sense where you're coming from and where you are now and all the 00:01:10.252 --> 00:01:13.592 steps in between that matter to our conversation? 00:01:14.492 --> 00:01:20.752 Sure. Well, I'm a political scientist who has long been interested in international 00:01:20.752 --> 00:01:30.832 security affairs and often use agent-based modeling and game theory to study 00:01:30.832 --> 00:01:33.292 and analyze such situations. 00:01:34.032 --> 00:01:39.732 I've also been interested in evolutionary biology and I've used some evolutionary 00:01:39.732 --> 00:01:42.812 concepts in my political science work. 00:01:42.812 --> 00:01:50.632 Work um my i'm at the university of michigan uh i've recently retired but i'm 00:01:50.632 --> 00:01:58.012 still active um uh my two major i had many collaborations uh, 00:01:59.472 --> 00:02:05.572 one was an evolutionary biologist william hamilton and more recently was an 00:02:05.572 --> 00:02:11.132 oncologist at cancer expert Kenneth Fienta. We can talk about those or others. 00:02:11.712 --> 00:02:14.352 Would you like more elaboration on my career? 00:02:15.852 --> 00:02:20.532 No, look, whatever you feel is relevant as an introduction. So it's really up to you. 00:02:20.592 --> 00:02:24.272 I'm not going to sway you one way or the other. But if you think that's enough 00:02:24.272 --> 00:02:26.112 as a start, then we're ready to go. 00:02:26.392 --> 00:02:30.812 So the topic is collaboration. 00:02:32.492 --> 00:02:34.752 So what is it and what is it good for? 00:02:37.056 --> 00:02:41.156 Well, in an academic setting where I am, 00:02:41.316 --> 00:02:48.216 collaboration often takes the form of two or more scholars or students working 00:02:48.216 --> 00:02:56.916 on a single project with the hope that they can do something that none of them could have done alone. 00:02:56.916 --> 00:03:02.276 And in many cases in my career, 00:03:02.516 --> 00:03:06.056 there have been interdisciplinary collaborations, I mean political scientists 00:03:06.056 --> 00:03:12.036 who worked with somebody from outside political scientists or even outside of social sciences. 00:03:14.336 --> 00:03:19.076 But in some sense, you now give an example of where it occurs, 00:03:19.496 --> 00:03:23.336 but in some sense the question would now be what are really the defining features? 00:03:23.616 --> 00:03:25.736 What makes something a collaboration? 00:03:28.556 --> 00:03:34.436 I think the basic defining feature is two or more people working on the same 00:03:34.436 --> 00:03:42.616 problem with mutual support, as opposed to completely independently sewing it together at the end. 00:03:44.396 --> 00:03:49.276 But then you could argue that, okay, we have cooperative systems and collaborative 00:03:49.276 --> 00:03:52.156 systems, so why don't we just give them one label? 00:03:52.156 --> 00:03:57.416 Because there in all cases we have agents doing things together right so it 00:03:57.416 --> 00:04:02.356 seems that there are more nuances here at work so what would these be well I 00:04:02.356 --> 00:04:06.256 think collaborative efforts um. 00:04:08.429 --> 00:04:15.149 Tend to be cooperative so i think collaboration is a subset uh but it is um, 00:04:16.769 --> 00:04:22.429 exploiting the differences between the participants uh the different things 00:04:22.429 --> 00:04:28.489 they bring to it the diversity uh they of talents and background and perspectives and goals, 00:04:29.309 --> 00:04:33.469 uh whereas cooperation where it's cooperation could be for example it's just 00:04:33.469 --> 00:04:36.569 uh i trade something with you and you give me money. 00:04:36.869 --> 00:04:40.789 And so cooperation is a much broader term. 00:04:41.289 --> 00:04:48.189 And certainly, we usually think of collaboration as necessarily human, 00:04:48.389 --> 00:04:52.509 although you could say that ants collaborate as well. 00:04:53.729 --> 00:04:59.409 But now, would that difference be, let's say that there's in the collaboration 00:04:59.409 --> 00:05:05.409 case, an explicitly defined goal in a corporation case, there's some happenstance 00:05:05.409 --> 00:05:06.769 interaction of exchange. 00:05:07.689 --> 00:05:13.249 Is that what you're going to? Yeah, collaboration typically involves, 00:05:13.389 --> 00:05:17.669 as I've said, a common purpose and the development of a common product, 00:05:18.589 --> 00:05:19.869 something that they achieve together. 00:05:20.129 --> 00:05:26.689 And it also typically involves some specialization, people having different 00:05:26.689 --> 00:05:29.269 things that they bring to the team. 00:05:30.649 --> 00:05:34.709 With your background, also, as you said earlier, your interest in evolutionary 00:05:34.709 --> 00:05:39.389 biology and your identity as a political scientist, 00:05:39.809 --> 00:05:46.389 do you then also see continuity here in this process of collaboration across 00:05:46.389 --> 00:05:48.989 different kinds of biological systems? 00:05:48.989 --> 00:05:52.129 Systems from let's say in case 00:05:52.129 --> 00:05:57.589 your cancer example cells interacting together differentiating in specific ways 00:05:57.589 --> 00:06:03.829 to interacting conscious and intentional agents like humans do you see a continuum 00:06:03.829 --> 00:06:12.109 there is that also what you what you are pointing to here yes uh well i'll give you an example um. 00:06:13.831 --> 00:06:24.251 And cancer involves cells, lines, developing the ability to overcome about seven 00:06:24.251 --> 00:06:28.611 or eight defenses that the host has to keep the cells in line. 00:06:29.091 --> 00:06:35.951 And the convention of wisdom was that one cell line based on a lot of mutations 00:06:35.951 --> 00:06:39.851 would develop these abilities to overcome all those defenses. 00:06:40.871 --> 00:06:47.111 I came up with an analogy much later, which goes like this. 00:06:47.151 --> 00:06:50.871 Imagine two thieves trying to rob a house. 00:06:51.131 --> 00:06:54.991 If one of them can pick the lock and the other one could turn off their wire, 00:06:55.231 --> 00:06:59.471 but they don't both have to have all the capacity to overcome the defenses. 00:07:00.731 --> 00:07:05.171 So that's a case of collaboration. It's also an example of cooperation. 00:07:05.171 --> 00:07:12.811 So I worked with a geneticist and an oncologist to develop this concept, 00:07:13.091 --> 00:07:16.571 see if it was even feasible. 00:07:18.011 --> 00:07:22.971 Later, it's been found to be valid in lab experiments. It actually happened. 00:07:25.011 --> 00:07:29.711 So that's collaboration within a tumor. 00:07:29.711 --> 00:07:40.431 In addition, you might say the same example can be used to study collaboration. 00:07:44.391 --> 00:07:50.351 I seem to have lost my picture here for some reason. I didn't lose your picture. 00:07:50.911 --> 00:07:57.391 But would you then say that collaboration defined in this way, 00:07:57.491 --> 00:08:00.431 also with the cancer example would fit in 00:08:00.431 --> 00:08:03.851 this broader category of an emergent phenomenon of 00:08:03.851 --> 00:08:07.151 sort of interacting elements that create 00:08:07.151 --> 00:08:16.171 a higher level structure yes absolutely but i was also going to add that that 00:08:16.171 --> 00:08:25.491 same example of tumors is a collaboration between uh three scientists uh bringing 00:08:25.491 --> 00:08:26.731 different things to the table. 00:08:27.651 --> 00:08:31.371 And so there's collaboration within the tumor and there's collaboration within 00:08:31.371 --> 00:08:34.411 the academic study of studying that. 00:08:35.291 --> 00:08:41.651 But sure, no, I understand that Nosara discussion in that sense is collaborative. 00:08:42.291 --> 00:08:47.871 But in some sense, if you now bring it to this underlying issue of the commons, 00:08:47.891 --> 00:08:52.651 right, that is supposedly playing a structuring role in human collaboration, 00:08:53.111 --> 00:08:55.971 what will be the commons in the case of cancer? 00:08:57.211 --> 00:08:58.991 Well, the cancer case, the... 00:09:01.912 --> 00:09:08.772 You might say the common evolutionary path of the cells in the tumor are to 00:09:08.772 --> 00:09:13.012 multiply and increase as they are in every biological system. 00:09:14.892 --> 00:09:25.612 Some of these cells can be evoking something called VEGF, which encourages capillaries, 00:09:25.672 --> 00:09:28.752 which encourages blood supply, which encourages oxygen. 00:09:28.752 --> 00:09:33.092 So one of them might be promoting oxygen in the immediate area, 00:09:33.272 --> 00:09:36.252 and the other one might be promoting some other nutrient. 00:09:39.232 --> 00:09:45.832 And so the common purpose, I mean, there's no common purpose defined by the 00:09:45.832 --> 00:09:47.472 evolutionary process itself, 00:09:47.612 --> 00:09:54.312 but the result could be at the emergent property of a malignant tumor. 00:09:54.312 --> 00:10:02.532 In the case of typically human collaboration, there's often a specific goal 00:10:02.532 --> 00:10:07.792 that either has been handed to them as team members or they develop themselves. 00:10:08.832 --> 00:10:12.792 But notice that the goals of the humans could be quite different. 00:10:12.972 --> 00:10:19.352 One of them might be after fame and getting credit and promotion in the organization. 00:10:19.352 --> 00:10:25.732 The other one might be an attempt to try to solve a problem and care more about 00:10:25.732 --> 00:10:27.352 the problem than their personal credit. 00:10:27.672 --> 00:10:34.632 So collaboration doesn't necessarily require complete agreement on the goals. 00:10:38.209 --> 00:10:44.329 Yeah, we've been active in the chat, Bob, because this is raising a lot of super 00:10:44.329 --> 00:10:46.049 interesting questions for us. 00:10:46.129 --> 00:10:54.309 So there's a couple of C words for, I think, that we're throwing around here. 00:10:54.309 --> 00:11:02.989 We're talking about cooperation, collaboration, the commons, right? 00:11:03.149 --> 00:11:06.789 But we can also talk about collusion, right? 00:11:07.249 --> 00:11:10.609 And the existence, I guess, of club goods, excludable goods. 00:11:10.829 --> 00:11:14.289 So when we're talking about a common interest or common purpose, 00:11:14.629 --> 00:11:19.949 I'm wondering to what extent, and it's interesting that we're talking about two different systems. 00:11:19.949 --> 00:11:26.529 The human system where there's rational forethought and then this biological 00:11:26.529 --> 00:11:28.769 system where there isn't. 00:11:29.569 --> 00:11:38.709 And so we're talking about whether there needs to be completely overlapping 00:11:38.709 --> 00:11:43.949 common goals or whether there can be some differentiation in that. 00:11:43.949 --> 00:11:48.109 And then also at the level of when we're talking about cancer cells. 00:11:49.089 --> 00:11:54.949 If the cancer cells are successful, they're actually going to kill the host, 00:11:55.229 --> 00:11:58.509 which is going to lead to their own death. 00:11:58.509 --> 00:12:02.429 So that's an interesting thing to unpack. 00:12:02.829 --> 00:12:07.329 And at the human level, an interesting thing to unpack is to the extent that 00:12:07.329 --> 00:12:16.069 this interest is held in common across all members of the group and includes 00:12:16.069 --> 00:12:19.049 also benefits to those outside of the group. 00:12:19.049 --> 00:12:22.789 So that's a lot of threads that I've just thrown out to you, 00:12:22.829 --> 00:12:25.869 but just because this is already a quite rich conversation. 00:12:25.969 --> 00:12:30.049 So any of that that you want to pick up on, and I'm sure that my colleagues 00:12:30.049 --> 00:12:33.709 will follow up on any piece that we want to continue with. 00:12:35.289 --> 00:12:38.549 We've already talked a little bit about the first three. On collusion, 00:12:38.769 --> 00:12:43.749 I see collusion as cooperation in the service of exploiting others. 00:12:43.749 --> 00:12:50.249 An example would be two major companies work together to set high prices. 00:12:51.349 --> 00:12:55.429 And so it's what we call a restraint of trade. 00:12:55.549 --> 00:13:04.269 And so that kind of business collusion is done often in secret so that the outsiders 00:13:04.269 --> 00:13:10.389 who are being disadvantaged don't even necessarily know the collusion is happening. 00:13:10.769 --> 00:13:12.949 But is that what the cancer cells are doing then? 00:13:13.749 --> 00:13:17.329 It's exploitation of others. Yes. 00:13:25.760 --> 00:13:30.460 Okay and of course as you said I can't fortunately cancer, 00:13:31.340 --> 00:13:37.520 evolutionary process as you said is a dead end mm-hmm mm-hmm and that's the 00:13:37.520 --> 00:13:41.440 tragedy of the commons for the cancer cells right yeah exactly so they're destroying 00:13:41.440 --> 00:13:44.700 the host that nurtures them, 00:13:45.320 --> 00:13:49.400 the better they do the more likely they're going to kill everything, 00:13:50.260 --> 00:13:53.960 that's well kill their environment fortunately it's not infectious so they don't 00:13:53.960 --> 00:13:56.140 kill others humans But they kill themselves. 00:13:56.680 --> 00:14:02.960 That's right. Yes, that's right. So it reminds me a little bit of climate change. Yes. 00:14:04.100 --> 00:14:10.500 Andreas, you wanted to step in. Yes. So, Bob, I would like to return to your 00:14:10.500 --> 00:14:17.200 article, Achieving Cooperation Under Anarchy, Strategies and Institutions from, I think, 1985, right? 00:14:17.440 --> 00:14:19.580 Which really sets kind of the 00:14:19.580 --> 00:14:23.480 ground for a lot of current thinking about collaboration and cooperation. 00:14:23.480 --> 00:14:29.680 And in that one, you were really using the powers of game theory as a way to 00:14:29.680 --> 00:14:32.920 outline the kind of dilemmas that one might end up in. 00:14:33.640 --> 00:14:38.500 Could you maybe start by telling a little bit about what would be the difference 00:14:38.500 --> 00:14:43.240 between being in a stack hunt or being in a prisoner's dilemma or in another 00:14:43.240 --> 00:14:46.800 of these situations when you are mutually trying to collaborate with each other? 00:14:53.507 --> 00:14:57.227 Well collaboration itself could be a prisoner's dilemma it could be that two 00:14:57.227 --> 00:15:03.567 members of the team each have an incentive to shirk that is to say let the other guy do all the work, 00:15:04.687 --> 00:15:08.687 but if they both shirk of course then the collaboration doesn't accomplish anything, 00:15:09.507 --> 00:15:17.007 and whereas um if they're in a if the collaboration goes on for some time uh 00:15:17.007 --> 00:15:20.087 each Either party might say to the other, 00:15:20.227 --> 00:15:25.307 well, if you shirk, I'll shirk, but if you're willing to give your contribution, 00:15:25.547 --> 00:15:27.107 I'll make a contribution too. 00:15:28.047 --> 00:15:36.927 We can even play tit for tat, which is to say, I'll start by giving some effort to this collaboration. 00:15:37.347 --> 00:15:42.387 Then if you cooperate by giving effort as well, then I'll continue so. 00:15:42.387 --> 00:15:48.947 But if you start shirking and not giving any effort, then I'll stop too, 00:15:49.047 --> 00:15:54.487 and then you'll realize that that was a losing proposition for both of us. 00:15:55.087 --> 00:16:03.967 So even within a collaboration, you can have a prisoner's dilemma between the participants. 00:16:05.287 --> 00:16:10.307 But I guess what this suggests is that the understanding of the situation is 00:16:10.307 --> 00:16:13.187 a critical part of a collaboration, right? 00:16:13.347 --> 00:16:17.747 Or the mutual understanding of the situation is a critical part of the collaboration. 00:16:18.887 --> 00:16:27.567 Well, they can have quite different conceptions of what it's about and what their goals are. 00:16:27.567 --> 00:16:36.447 They do have to appreciate, however, that the process is interactive. 00:16:37.287 --> 00:16:42.367 What I do and what you do jointly determine the payoffs to both of us. 00:16:42.507 --> 00:16:47.047 They have to at least understand that, so that when the other side... 00:16:51.592 --> 00:16:59.112 If you stop making effort, you come to realize that the other side might stop 00:16:59.112 --> 00:17:01.732 too, and that's better than both of you working. 00:17:03.252 --> 00:17:06.312 But you could have quite different understandings of the situation. 00:17:06.592 --> 00:17:10.272 For example, as I said, you might have different goals and you might have different 00:17:10.272 --> 00:17:13.352 understandings of what resources are available to the collaborators. 00:17:13.592 --> 00:17:14.932 You might have different understandings 00:17:14.932 --> 00:17:18.192 of who your audience is and what you're trying to accomplish. 00:17:18.192 --> 00:17:23.032 For example, in many organizations, two members of the same team might report 00:17:23.032 --> 00:17:30.272 to different bosses, and each one of them has the goal of satisfying their own boss. 00:17:33.532 --> 00:17:35.712 And so they don't necessarily have 00:17:35.712 --> 00:17:40.792 to have the same goals or even the same understanding of the situation. 00:17:40.792 --> 00:17:50.312 They have this minimal need to appreciate that if they don't cooperate, 00:17:50.552 --> 00:17:51.852 the other side probably won't either. 00:17:54.512 --> 00:17:59.092 And of course, critical to the game's theoretical approaches is the identification 00:17:59.092 --> 00:18:01.532 of a payoff matrix, right? 00:18:01.912 --> 00:18:06.032 But in a lot of kind of, I guess, real life situations, the payoff matrix is 00:18:06.032 --> 00:18:10.732 not quite so clear cut. There is a lot of uncertainty with it. 00:18:10.952 --> 00:18:14.832 And can you say anything about from your kind of perspective, 00:18:14.952 --> 00:18:20.092 what do you do or what sustains collaboration then in situations where the payoff 00:18:20.092 --> 00:18:21.812 matrix might be unknown? 00:18:25.688 --> 00:18:30.928 Well, typically, it might have a certain amount of uncertainty and error and 00:18:30.928 --> 00:18:35.108 noise in judging it, but it's rarely completely unknown. 00:18:35.308 --> 00:18:39.188 If it's completely unknown, then you'd have no reason to collaborate because 00:18:39.188 --> 00:18:40.788 you don't know whether to do anything or not. 00:18:40.788 --> 00:18:44.068 Um so i 00:18:44.068 --> 00:18:47.468 i i accept that uh some uh 00:18:47.468 --> 00:18:52.928 uncertainty about exactly what it is is common and and some notion even of whether 00:18:52.928 --> 00:18:57.248 the other guy cooperated recently uh you're never sometimes you've never been 00:18:57.248 --> 00:19:05.548 sure of that um so there could be misperception and noise um but there has to be, 00:19:07.268 --> 00:19:13.248 some understanding that your choices affect the other side and their choices affect yours. 00:19:13.448 --> 00:19:17.288 And just for the example for the prisoner's dilemma, you just have to have an 00:19:17.288 --> 00:19:21.888 understanding that you're the right order of the payoffs. 00:19:22.028 --> 00:19:26.088 So you have to have, in a prisoner's dilemma, what defining feature is that. 00:19:28.668 --> 00:19:33.728 The best payoff for you is if you shirk and the other guy does all the work. 00:19:33.808 --> 00:19:39.148 The next best is if both of you work, and the next best is if neither of you 00:19:39.148 --> 00:19:44.108 work, and the worst is if you do all the work and the other guy shirks. 00:19:44.468 --> 00:19:49.388 And so that's all that you really need for the prisoner's dilemma is just an 00:19:49.388 --> 00:19:51.668 understanding of the order of those possibilities, 00:19:51.968 --> 00:19:57.628 which is pretty common, namely that you do better by shirking, 00:19:57.648 --> 00:19:59.668 but you both do better by cooperating. 00:20:01.418 --> 00:20:06.498 Now, Bob, do you feel that these kinds of models from game theory alike we now 00:20:06.498 --> 00:20:09.298 describe in the person's dilemma are still valid? 00:20:09.618 --> 00:20:13.558 I mean, they also have been criticized in caricatures, if you want, 00:20:13.678 --> 00:20:14.958 of human decision-making. 00:20:15.078 --> 00:20:18.898 Herb Simon already started with this, right, with this whole notion of bounded rationality. 00:20:19.398 --> 00:20:23.398 There's also more and more evidence that humans are not optimizers always. 00:20:23.538 --> 00:20:25.298 We don't optimize really for outcome. 00:20:25.598 --> 00:20:30.498 We might have surprising deviations from that. So do you feel that that's still 00:20:30.498 --> 00:20:36.338 an effective model to look at the kinds of collaboration that we would like 00:20:36.338 --> 00:20:39.158 to optimize in everyday life and in our world? 00:20:39.538 --> 00:20:44.478 Well, traditional game theory and how it began does assume rational actors who 00:20:44.478 --> 00:20:49.658 have the ability to do foresightful analysis, look ahead, if I do this, 00:20:49.698 --> 00:20:51.318 you do that sort of thing. 00:20:51.318 --> 00:20:55.518 So that's true but evolutionary game 00:20:55.518 --> 00:20:58.938 theory has been able to drop the assumption of rationality 00:20:58.938 --> 00:21:04.098 so for example these cancer cells are not rational they don't look ahead they 00:21:04.098 --> 00:21:11.838 don't do any calculation um but uh but they can still cooperate from an evolutionary 00:21:11.838 --> 00:21:16.658 perspective which is to say what because whatever works well means that they reproduce more. 00:21:17.358 --> 00:21:23.518 And so the prisoners of land, for example, and game theory in general, 00:21:23.558 --> 00:21:31.418 no longer require rationality. It's sometimes a useful assumption in order to make predictions. 00:21:33.218 --> 00:21:36.958 And sometimes the predictions are wrong if the players aren't rational. 00:21:41.037 --> 00:21:46.817 But there's a downside to that, right? Because then we sacrifice the goal-directedness, 00:21:46.837 --> 00:21:52.297 the coordination of human agents in collaborative contexts. 00:21:52.777 --> 00:21:57.977 And we are back into a more behaviorist perspective on how human behavior comes 00:21:57.977 --> 00:22:01.777 about, which also has not been a complete success story. 00:22:02.757 --> 00:22:07.937 So there is a price you then pay if you bring it into this pure evolutionary, 00:22:08.677 --> 00:22:10.337 non-rational perspective. 00:22:11.037 --> 00:22:14.917 Or do you feel that's just a necessary step? Well, for humans, 00:22:15.037 --> 00:22:20.777 another version of evolutionary, another aspect that's not rational is trial and error learning, 00:22:21.077 --> 00:22:25.657 where you don't necessarily look ahead, but you experience whether you're doing 00:22:25.657 --> 00:22:28.737 well or not, and you change behavior if you're not doing well. 00:22:28.737 --> 00:22:37.777 So trial and error learning is another form of adaptation that humans can do, and of course always do, 00:22:37.937 --> 00:22:43.737 that's different from rationality, but it's often a very effective thing. 00:22:46.557 --> 00:22:49.977 Go ahead, Jenna. Sorry. Oh, no, thank you. 00:23:18.657 --> 00:23:22.537 You know, we can have biases, we have fears, et cetera. 00:23:22.777 --> 00:23:27.297 And so, you know, what role does that play? To what extent could we incorporate 00:23:27.297 --> 00:23:33.977 that into our game theoretic analyses successfully, make it work for us, 00:23:33.977 --> 00:23:37.077 not against us, as we're trying to understand human collaboration? 00:23:39.795 --> 00:23:43.695 But let me give you a specific example. 00:23:46.435 --> 00:23:53.555 Last week, I was, no, three weeks ago, I was talking to Chinese and Russian 00:23:53.555 --> 00:23:59.195 academics and think tank type people, 00:23:59.255 --> 00:24:04.335 an ongoing discussion of cyber conflict and how to avoid instability. 00:24:04.335 --> 00:24:07.835 Stability and the uh 00:24:07.835 --> 00:24:11.415 point of what i said was beware of 00:24:11.415 --> 00:24:16.195 vengeance so this is one of the psychological features of humans that jenna 00:24:16.195 --> 00:24:23.395 was talking about which is uh when we feel insulted or exploited or disgraced 00:24:23.395 --> 00:24:30.975 or punished without reason uh we often have a very powerful psychological need for vengeance, 00:24:31.335 --> 00:24:37.175 which often then we act on without much regard to the costs and benefits. 00:24:37.435 --> 00:24:40.535 We sort of put all that future-oriented thinking ahead and say, 00:24:40.635 --> 00:24:44.095 well, vengeance really feels good. I need to do it. 00:24:44.895 --> 00:24:46.415 And that's very dangerous. 00:24:47.455 --> 00:24:51.895 And if you're in a realm like cyber conflict where cyber weapons are used at 00:24:51.895 --> 00:24:57.895 scale, there's a lot of possibilities that people will feel the need for vengeance if attacked. 00:24:59.655 --> 00:25:05.435 And the implication, of course, is that we should really be aware of that when 00:25:05.435 --> 00:25:08.835 we try to understand the consequences of our own actions. 00:25:10.055 --> 00:25:16.435 So is it enough to say beware of vengeance, or is there then some guidance about 00:25:16.435 --> 00:25:18.835 how to avoid it? Oh, yeah. 00:25:18.895 --> 00:25:26.895 I mean, the guidance is to take into account what the other side might regard 00:25:26.895 --> 00:25:28.915 as requiring vengeance. 00:25:31.555 --> 00:25:37.075 Well, here's a good example. Pearl Harbor. The Japanese attacked Pearl Harbor 00:25:37.075 --> 00:25:40.855 and then knew that the United States was stronger. 00:25:40.855 --> 00:25:47.075 Stronger, but they didn't have a clear understanding of what would happen next, 00:25:47.195 --> 00:25:51.075 except that for six months they would be able to conquer everything in sight, 00:25:51.215 --> 00:25:53.095 but after that they'd be short of resources. 00:25:54.695 --> 00:26:02.715 I think that the United States' reaction to Pearl Harbor, among other things, 00:26:02.875 --> 00:26:04.615 was a strong desire for vengeance. 00:26:06.015 --> 00:26:09.595 That kept us going. uh and in 00:26:09.595 --> 00:26:12.335 a way where the japanese thought well maybe americans will let 00:26:12.335 --> 00:26:16.955 bygones be bygones they'll take the philippines and indonesia and malaysia and 00:26:16.955 --> 00:26:22.595 indo-china but then we'll let it we'll just let it all um regardless of the 00:26:22.595 --> 00:26:27.535 fate of complete uh but in fact we didn't of course and i think part of the 00:26:27.535 --> 00:26:31.955 reason we didn't was vengefulness so i think the uh. 00:26:34.007 --> 00:26:40.187 One policy application is to be cognizant of what the other side, 00:26:40.367 --> 00:26:46.007 in this case, America for Japan, what the other side might regard as vengeful action. 00:26:46.107 --> 00:26:50.047 And if you do it, you better do it with your eyes open. Otherwise, you better avoid it. 00:26:51.007 --> 00:26:56.367 But Bob, there's something that I cannot fully understand. Maybe you can explain 00:26:56.367 --> 00:27:04.267 that, because vengefulness, the wish for revenge, is a very individually defined emotional state. 00:27:04.727 --> 00:27:08.627 And now we describe it at the level of a nation state. Well, 00:27:08.647 --> 00:27:12.187 because it's shared by lots of individuals, yeah. 00:27:12.987 --> 00:27:18.447 Yeah, no, I think it's perfectly reasonable to say the United States was in a vengeful mood. 00:27:18.507 --> 00:27:21.587 Or after 9-11, we were in a vengeful mood. 00:27:21.587 --> 00:27:28.247 And of course we invaded Afghanistan and eventually killed bin Laden part of 00:27:28.247 --> 00:27:30.527 the motivation of Afghanistan was certainly vengeance, 00:27:31.467 --> 00:27:37.607 fine I understand but the point is of course there are many layers now between 00:27:37.607 --> 00:27:42.967 that right so we have institutions we have larger groups of humans that all 00:27:42.967 --> 00:27:48.467 coalesce into something like a nation state so how would that percolate between 00:27:48.467 --> 00:27:51.187 these levels of course you can say well they respond to this 00:27:51.587 --> 00:27:54.607 to the sense of vengefulness in the individual citizen. 00:27:55.227 --> 00:28:00.347 But still, how is that then incorporated in all the institutional layers that you have to go through? 00:28:00.807 --> 00:28:03.707 Well, in this case, I mean, sometimes that can be very complicated. 00:28:03.767 --> 00:28:07.147 Like you're trying to pass a law to slow down pollution. 00:28:07.487 --> 00:28:09.827 You know, you have to take into account all those layers. 00:28:11.167 --> 00:28:20.267 But in some of the cases I've cited, it's very straightforward because in both 00:28:20.267 --> 00:28:25.687 of those examples, was American citizens shared the experience of being attacked. 00:28:26.907 --> 00:28:37.427 They shared the experience of feeling that it was not an appropriate response to anything they did. 00:28:39.868 --> 00:28:47.888 And the leaders felt the same way. And so there was strong support in both those 00:28:47.888 --> 00:28:51.308 cases at all levels. So it didn't get complicated. 00:28:51.508 --> 00:28:53.428 You didn't have to worry about the role of Congress. 00:28:55.068 --> 00:29:00.008 You didn't have to worry about interest groups and so on. 00:29:00.088 --> 00:29:04.648 And because there was such a widespread understanding of the situation, 00:29:04.788 --> 00:29:12.968 it's calling for vengeance. and there was already identification and loyalty to the nation. 00:29:13.308 --> 00:29:17.908 So when Pearl Harbor was attacked, it wasn't just Hawaiians that felt attacked. 00:29:18.088 --> 00:29:19.608 It was all Americans felt attacked. 00:29:19.888 --> 00:29:23.188 And likewise, 9-11, it wasn't just New Yorkers that felt attacked. 00:29:23.768 --> 00:29:30.668 There was already this strong common understanding that the whole of the United States was attacked. 00:29:30.668 --> 00:29:37.108 And so to summarize that, sometimes the situation is simple enough that you 00:29:37.108 --> 00:29:39.328 don't have to worry about all the intervening stages. 00:29:39.528 --> 00:29:42.288 They all operate smoothly. 00:29:42.928 --> 00:29:45.608 But Bob, there's another aspect to this, right? 00:29:45.668 --> 00:29:51.568 So Japan was in a systematic war of conquest already since the 1930s. 00:29:51.868 --> 00:29:56.288 Also, when they attacked Pearl Harbor, they were performing many military operations 00:29:56.288 --> 00:30:01.868 around the Pacific. And their goal was to knock out the U.S. 00:30:01.868 --> 00:30:04.968 Of the war in one blow because they hoped to find all the ships there. 00:30:05.368 --> 00:30:09.428 And in the meantime, in parallel, the United States was in a competition, 00:30:09.548 --> 00:30:11.748 if you want, also for resources in the Pacific. 00:30:11.888 --> 00:30:15.928 So I could also give an alternative interpretation where I say, 00:30:15.988 --> 00:30:20.828 well, at the nation-state level, it was also a competition for control and resource. 00:30:20.828 --> 00:30:24.448 But to mobilize the citizenry, 00:30:24.548 --> 00:30:30.308 we use the sense, we enhance the sense of vengeance to mobilize the individuals 00:30:30.308 --> 00:30:34.028 in our society to actually volunteer to go to the army and sacrifice things 00:30:34.028 --> 00:30:39.468 so that we can achieve the goals we have as a nation state, which are not the same as vengeance. 00:30:42.828 --> 00:30:48.548 Well, I agree that the goals involve resources and power politics, but I'm just saying that, 00:30:49.228 --> 00:31:02.528 the motivation to declare war on Japan on December 7th and not a month before or not a year before, 00:31:02.748 --> 00:31:08.808 the situation from the American point of view didn't change dramatically in 00:31:08.808 --> 00:31:11.408 in the weeks leading up to Pearl Harbor. So, um. 00:31:16.362 --> 00:31:23.462 And so the reaction was definitely triggered, you might say, by that event. 00:31:23.662 --> 00:31:29.782 Now, of course, what we did about it did require leadership, 00:31:30.002 --> 00:31:35.002 including decisions such as how much of American resources should be devoted 00:31:35.002 --> 00:31:39.542 to the European theater to fight the Nazis and how much should be devoted to Japan. 00:31:39.542 --> 00:31:46.902 Yeah, that does require leadership, and you might study the institutions in 00:31:46.902 --> 00:31:48.922 order to understand that choice. 00:31:50.382 --> 00:32:00.442 So as we're teasing this apart, it almost sounds like there might be two different 00:32:00.442 --> 00:32:05.022 kinds of psychological processes going on. 00:32:05.022 --> 00:32:12.882 Maybe at the elite level, we've got a rational calculus of the appropriate actions 00:32:12.882 --> 00:32:15.822 for the state to take, but because it's a democracy, 00:32:16.062 --> 00:32:24.402 we need the support of the public, and so the elites can use emotion to get the public on board. 00:32:24.702 --> 00:32:33.382 Would you say that might be? That often happens. The clearest case may be the Spanish-American War. 00:32:34.842 --> 00:32:46.162 Also, the Mexican-American War, where the leadership led the way for the public 00:32:46.162 --> 00:32:48.142 to support the goals of the war. 00:32:48.142 --> 00:32:54.202 And it wasn't sort of spontaneous from the bottom up the way it was in 9-11 or Pearl Harbor. 00:32:56.122 --> 00:33:02.722 And certainly, the leaders do try to mold public opinion. 00:33:02.742 --> 00:33:09.602 I remember Roosevelt had a famous line to one of the people urging him to take action. 00:33:09.642 --> 00:33:13.042 He said, well, you've convinced me, now go out and convince the public. 00:33:14.383 --> 00:33:19.163 So that they will force me to do it, or in a way, please support me doing it. 00:33:20.003 --> 00:33:25.163 Yeah, so there are certainly complex interactions between the public and the leadership. 00:33:25.383 --> 00:33:29.763 And once you get into that, then you might well need to get into the institutional 00:33:29.763 --> 00:33:35.003 structures that mediate that difference, such as elections or the legislature. 00:33:36.143 --> 00:33:42.663 Yeah, and the public, I mean, it may be that they're unable to see as clearly 00:33:42.663 --> 00:33:47.103 as the elites, um, are, are interested in acting or not acting. 00:33:47.423 --> 00:33:51.463 Um, but then it's also probably the case that an awful lot of information cannot 00:33:51.463 --> 00:33:52.423 be shared with the public. 00:33:52.463 --> 00:34:00.043 So there's, there's, um, uh, um, covert information that just can't be released. 00:34:00.343 --> 00:34:04.583 Um, do you think, you know, of course, in, uh, in our country, 00:34:04.643 --> 00:34:10.503 Bob, um, we're all sort of, um, thinking about the implications of the the United 00:34:10.503 --> 00:34:12.043 States pulling out of Afghanistan. 00:34:12.723 --> 00:34:19.583 Do you think that this dynamic that you've been describing with Japan is happening 00:34:19.583 --> 00:34:25.723 in reverse in Afghanistan in the sense that the public just can't see our common 00:34:25.723 --> 00:34:29.243 interest in being present in Afghanistan anymore, 00:34:29.423 --> 00:34:30.903 and the leadership isn't able 00:34:30.903 --> 00:34:35.123 to make a convincing, compelling argument to bring the public along? on? 00:34:36.783 --> 00:34:40.283 Well, in this case, I don't think the leadership of the Biden administration, 00:34:40.863 --> 00:34:43.243 sees value in continuing the war. 00:34:43.383 --> 00:34:49.723 I don't think that they feel compelled by a public. 00:34:50.123 --> 00:34:54.383 I think they felt that they had quite a lot of choice about whether to declare 00:34:54.383 --> 00:34:56.883 a date by which thing we withdraw. 00:34:59.363 --> 00:35:05.543 You made a comment about the fact that the elites are rational, 00:35:05.863 --> 00:35:07.663 and I would say not necessarily. 00:35:08.143 --> 00:35:18.343 They sometimes make choices that, after the fact, we think were made for bad reasons, 00:35:18.543 --> 00:35:23.563 and not just bad information, but just not very carefully thought out. 00:35:23.563 --> 00:35:27.503 I mean, I've already mentioned Japanese case. I thought that was an irrational decision. 00:35:32.766 --> 00:35:41.326 Well, so if I ask, another thing to make a comment about Jenna's statement that 00:35:41.326 --> 00:35:45.106 sometimes the leaders can't share information, that's true. 00:35:45.426 --> 00:35:49.766 And then one of the factors that come into play is trust. 00:35:50.106 --> 00:35:54.826 Does the public trust the leaders? If they say this is the thing where the United 00:35:54.826 --> 00:35:57.306 States should do, trust me, do they trust them? 00:35:57.306 --> 00:36:06.466 And I think a major loss in the 00:36:06.466 --> 00:36:10.666 last four years in the United States is a decline in trust of government, 00:36:10.806 --> 00:36:13.206 for good reason. 00:36:13.846 --> 00:36:18.166 Well, it would be great to develop this notion of trust a bit further, 00:36:18.226 --> 00:36:24.786 because it seems to be critical both when dealing with an uncertain situation or an uncertain payoff. 00:36:24.786 --> 00:36:29.986 Of, it also seems to be critical in, so to say, linking up the individual level 00:36:29.986 --> 00:36:31.886 and the institutional level. 00:36:32.926 --> 00:36:38.466 So could you say also theoretically a bit more about what are you thinking about 00:36:38.466 --> 00:36:41.106 trust in terms of importance for collaboration? 00:36:41.286 --> 00:36:45.166 And how has the importance of trust or the understanding of the role of trust 00:36:45.166 --> 00:36:47.006 changed over the last years in your thinking? 00:36:50.006 --> 00:36:54.886 Well, typically, collaboration does require trust. It requires trust that the 00:36:54.886 --> 00:37:00.566 other person will give an honest effort, show up and do their part before you 00:37:00.566 --> 00:37:02.486 actually find out if that's true. 00:37:02.626 --> 00:37:08.346 You can commit yourself to participating in this activity, trusting that it 00:37:08.346 --> 00:37:11.346 will be mutually valuable. 00:37:16.886 --> 00:37:22.026 In international affairs, in the cyber context, 00:37:22.386 --> 00:37:27.826 the Russians and Chinese have told the Americans that trust is perhaps the most 00:37:27.826 --> 00:37:31.926 important thing in order to make progress in avoiding war and escalation. 00:37:33.886 --> 00:37:39.026 And game theory has done some work on what is required to build trust, 00:37:39.266 --> 00:37:42.986 but I think it has a lot more to offer. 00:37:44.366 --> 00:37:49.826 And certainly collaboration, as I said, requires a certain amount of trust. 00:37:50.726 --> 00:37:54.106 And trust, I guess, is also related to another critical issue, 00:37:54.166 --> 00:38:00.446 which has to do with the temporal extension of the collaboration or the exchange, doesn't it? 00:38:00.466 --> 00:38:04.586 Because trust seems also to be about projecting into a future. 00:38:05.286 --> 00:38:08.046 So is that an area to develop further? further, also theoretically, 00:38:08.286 --> 00:38:13.566 the relation between, you can say, future thinking and trust? 00:38:14.386 --> 00:38:20.206 Absolutely. In fact, one of the major conclusions of my work on cooperation 00:38:20.206 --> 00:38:26.886 was the importance of what I call the shadow of the future, 00:38:27.026 --> 00:38:34.566 the idea that both sides know that there's going to be a continuing relationship in economic terms, 00:38:34.686 --> 00:38:38.666 that they don't discount the future or too much that they find the future is 00:38:38.666 --> 00:38:43.566 worth worrying about and worth sacrificing today by taking a chance on trust 00:38:43.566 --> 00:38:46.446 in order to see if you can get the long-term benefits. 00:38:48.586 --> 00:39:02.446 So trust definitely can be built up by an appreciation that you're in this for the long term. 00:39:04.906 --> 00:39:10.266 So, Bob, we looked at a number of key features now, right? 00:39:10.326 --> 00:39:15.286 So the last one being trust, and then also the shadow of the future. 00:39:15.806 --> 00:39:20.066 But if we want to put these all together in a comprehensive framework, 00:39:20.466 --> 00:39:26.206 we could argue that also that collaborative system, as we also just described, 00:39:26.366 --> 00:39:32.166 going from citizens through institutions to a president, in the end has some sort of architecture. 00:39:32.446 --> 00:39:36.146 Where we have a functional decomposition, we have protocols of communication, 00:39:36.746 --> 00:39:41.466 we have mechanisms in which incentives are distributed through the system and so on. 00:39:41.626 --> 00:39:47.666 So would you see it in those terms? Do you think we can look at collaborative 00:39:47.666 --> 00:39:50.706 processes from that perspective of an architecture? 00:39:51.006 --> 00:39:54.566 And if so, what would the main features of that architecture be? 00:39:56.982 --> 00:40:01.962 Well, one of the main features of the architecture is the extent to which the 00:40:01.962 --> 00:40:09.342 goals are set by an outside agent, such as the boss within an organization. 00:40:11.922 --> 00:40:17.602 To the extent that two members of a collaborative team are working for the same 00:40:17.602 --> 00:40:22.982 boss and have the goal of making the boss happy with their work, 00:40:23.122 --> 00:40:29.562 that's a pretty strong architecture and a pretty strong way of setting goals and monitoring them. 00:40:30.042 --> 00:40:36.262 But often, participants in the collaboration don't have a common boss that sets 00:40:36.262 --> 00:40:39.802 the terms of the collaboration, and they've got to work that out together. 00:40:41.502 --> 00:40:45.522 Sometimes explicitly, such as, 00:40:45.542 --> 00:40:53.962 for example, if there's a treaty between two countries or a contract between 00:40:53.962 --> 00:40:56.562 two companies. I'll provide this if you provide that. 00:40:57.562 --> 00:41:01.442 So sometimes it's formalized between the participants themselves, 00:41:02.282 --> 00:41:07.562 but often it's less structured by a good deal. There's a handshake. 00:41:07.822 --> 00:41:12.062 Let's work on this together. And there's no formal contract and there's not 00:41:12.062 --> 00:41:16.982 necessarily a detailed definition of what this is that we're working on or how 00:41:16.982 --> 00:41:18.422 much you'll do and how much I'll do. 00:41:19.862 --> 00:41:25.442 Sometimes there's a collaboration, there's an explicit statement that I'll provide 00:41:25.442 --> 00:41:29.642 this if you provide that, based often on specialization. 00:41:29.882 --> 00:41:32.702 So, for example, when I collaborated with evolutionary biologists, 00:41:32.962 --> 00:41:39.562 William Hamilton, to develop the biological implications of the prisoner's dilemma. 00:41:41.418 --> 00:41:45.798 We had an understanding of what his strong suit was, and he could provide lots 00:41:45.798 --> 00:41:48.638 of examples for biology for what I was talking about. 00:41:48.858 --> 00:41:53.898 And I could provide some of the results of my computer tournaments and analyze them. 00:41:54.018 --> 00:42:01.118 And so we shared a common interest and understanding of both evolutionary theory and game theory. 00:42:01.358 --> 00:42:04.998 And that helped us work together effectively, even though our backgrounds. 00:42:06.038 --> 00:42:09.358 I mean, he knew a lot about Beatles and I knew a lot about war. 00:42:10.358 --> 00:42:13.838 Those two knowledge bases didn't overlap. Right. 00:42:14.058 --> 00:42:19.658 But now if I understand you right, Bob, then you're saying there is a hierarchical 00:42:19.658 --> 00:42:22.558 structure of goals and sub-goals, if you want. 00:42:22.998 --> 00:42:28.518 But now for the agents in the system to pursue those goals, we have something 00:42:28.518 --> 00:42:34.038 like trust, which also means by implication, we assume that all these agents have a theory of mind. 00:42:34.038 --> 00:42:38.358 Because that's why you need trust because you have a model of the other agent 00:42:38.358 --> 00:42:40.858 that says, well, they can do bad things, they can do good things, 00:42:40.958 --> 00:42:44.658 and I just make the assumption they will do good things. That's trust, right? 00:42:44.878 --> 00:42:50.018 So now that already gives us elements of this architecture, but now you have 00:42:50.018 --> 00:42:52.478 protocols of communication, right? 00:42:52.518 --> 00:42:56.498 That both has to run within layers and between layers. 00:42:56.698 --> 00:43:00.798 So do you have a specific view on what the features of that kind of communication 00:43:00.798 --> 00:43:04.038 structure then should be key to maintain your goal structure. 00:43:06.795 --> 00:43:13.675 Well, it's often said that a major impediment to interdisciplinary collaboration 00:43:13.675 --> 00:43:18.995 is that the two sides don't speak the same language. 00:43:20.135 --> 00:43:24.375 So when I started working on cancer, I certainly didn't speak the language of 00:43:24.375 --> 00:43:25.335 the cancer specialists. 00:43:25.675 --> 00:43:29.715 And there was a huge variety of specialized terms. 00:43:31.155 --> 00:43:38.075 You don't have to master the 6,000 terms that are in the index of a major cancer textbook, 00:43:38.295 --> 00:43:48.115 but you do have to learn a lot about what p53 is, for example, and why it's important, 00:43:50.335 --> 00:43:58.095 and VEGF and a million, several hundred other things. and so developing mutual 00:43:58.095 --> 00:44:02.215 understanding is often part of the collaborative process. 00:44:04.875 --> 00:44:09.615 Andreas? Yeah, so I think this is very interesting that the moment we started 00:44:09.615 --> 00:44:13.275 out with architectures, but the moment we go to the scientific collaborations, 00:44:14.255 --> 00:44:18.435 we are, so to say, back again in this more anarchistic realm where there is 00:44:18.435 --> 00:44:23.195 not a clear hierarchy of a shared boss and a shared goal and yet people figure 00:44:23.195 --> 00:44:24.435 out ways of collaborating. 00:44:25.395 --> 00:44:30.515 So is it maybe time to think a little bit more about what does it take for these 00:44:30.515 --> 00:44:33.635 interdisciplinary collaborations to actually work? 00:44:33.855 --> 00:44:39.095 Is there still kind of a commitment to some overarching shared goal or shared 00:44:39.095 --> 00:44:45.935 cause, be it the advent of knowledge or the increase of knowledge that stabilizes it? 00:44:46.015 --> 00:44:50.175 Or what does it then take for these agents who are in very separate systems 00:44:50.175 --> 00:44:53.315 actually to create this space in the middle between them. 00:44:53.635 --> 00:44:57.395 And I think the interdisciplinary collaborations are really interesting instances of that. 00:45:01.403 --> 00:45:05.803 Well, in the cases of my collaborations, which I think are typical, 00:45:06.083 --> 00:45:11.143 although they're much more heavily interdisciplinary than most collaborations are, 00:45:11.303 --> 00:45:22.183 the project started with the shared goal of achieving a new understanding of 00:45:22.183 --> 00:45:25.603 some, a better understanding of some phenomena in the outside world, 00:45:25.823 --> 00:45:31.403 such as say an arms race or cancer, 00:45:31.843 --> 00:45:42.963 and that the intention usually among academics is that this would lead to at 00:45:42.963 --> 00:45:47.683 least one publication as a concrete output of this collaboration. 00:45:50.343 --> 00:45:55.823 Sometimes part of the goal is just having fun. I find it fun to work with people 00:45:55.823 --> 00:46:01.043 that know a lot of things that I don't know, and to understand their perspective. 00:46:01.843 --> 00:46:07.923 And a lot of science, of course, is driven by curiosity and the desire to kind 00:46:07.923 --> 00:46:10.143 of play with ideas and concepts, 00:46:10.383 --> 00:46:15.823 obviously with the purpose behind that of advancing understanding and knowledge, 00:46:16.043 --> 00:46:26.883 but not necessarily a material purpose such as getting a raise. 00:46:29.499 --> 00:46:35.479 So, in your background for us, you shared your presidential address to the American 00:46:35.479 --> 00:46:39.839 Political Science Association called Political Science and Beyond. 00:46:40.379 --> 00:46:44.879 And in that one, you also talk about, you know, the four ops that you need to 00:46:44.879 --> 00:46:46.139 cultivate as a researcher. 00:46:46.459 --> 00:46:49.959 It's reading up on a variety of fields. It's teaming up with others. 00:46:50.359 --> 00:46:53.279 It's about loading up on research related to your problem. 00:46:53.559 --> 00:46:57.539 And it's about lighting up when you need to escape from the problem for a while. 00:46:57.539 --> 00:47:03.979 So, with four principles, in a sense, also count as general principles that 00:47:03.979 --> 00:47:07.839 are useful to initiate collaborations because each of them, 00:47:07.859 --> 00:47:11.999 in a sense, seems to assume that you need to bring yourself into a particular 00:47:11.999 --> 00:47:14.379 state to allow these kind of collaborations to happen. 00:47:17.899 --> 00:47:22.999 Well, yes, and you certainly, I call those things the four ups and they're on 00:47:22.999 --> 00:47:26.479 my website and people want to check out what they mean. 00:47:31.679 --> 00:47:34.739 I think the most interesting way is the last one 00:47:34.739 --> 00:47:38.179 the lighten up and what are the points i make which i 00:47:38.179 --> 00:47:41.559 have to found in myself is that uh when 00:47:41.559 --> 00:47:47.339 either in a collaboration or working on your own you get to a dead end uh uh 00:47:47.339 --> 00:47:51.819 you beat your head against the wall uh it pays to lighten up it face to back 00:47:51.819 --> 00:47:58.239 off and sleep on and turn to some other project for a while in order to get 00:47:58.239 --> 00:48:01.459 out of the current mode of thinking that you're stuck with. 00:48:02.999 --> 00:48:07.279 So, Bob, that makes me think a little bit more about what would cause a collaboration 00:48:07.279 --> 00:48:10.119 that ought to continue to fail. 00:48:11.479 --> 00:48:18.279 And so, you know, or another way of thinking about that is what makes a collaboration robust? 00:48:19.179 --> 00:48:21.239 So what is it... 00:48:23.135 --> 00:48:28.195 So in your sense, what do you think about what causes something to fail or what 00:48:28.195 --> 00:48:33.635 would, alternatively, how could you structure it so that it would be able to 00:48:33.635 --> 00:48:37.875 continue despite being hit by, you know, in this sense, 00:48:38.015 --> 00:48:43.755 you know, maybe people's distraction from other projects that are pulling them away? 00:48:47.475 --> 00:48:52.715 Well, I'll give you one example of a failure that I was involved with, 00:48:52.795 --> 00:49:01.535 which is this story about cooperation among cancer cells done with Dr. 00:49:02.095 --> 00:49:08.795 Kev Pienta. So when we sent that off to a major journal, we got two reviews. 00:49:08.795 --> 00:49:13.995 We got two reviews and of course what we were doing was somewhat speculative 00:49:13.995 --> 00:49:20.995 and one review says it's impossible what they're proposing this kind of cooperation and the other journal, 00:49:21.095 --> 00:49:25.115 the other reviewers say oh everybody knows this already, 00:49:26.975 --> 00:49:33.215 so honestly that was a failure it was a failure to get the thing published in 00:49:33.215 --> 00:49:36.895 that journal and it was also a failure of our um, 00:49:38.815 --> 00:49:43.795 ability to communicate what we were trying to do to show that it wasn't impossible 00:49:43.795 --> 00:49:46.055 and it was different from what people already knew. 00:49:46.255 --> 00:49:53.455 And the way we proceeded was to give up on that journal and take responsibility 00:49:53.455 --> 00:49:59.595 for explaining what's new here and why it was at least possible. 00:50:01.315 --> 00:50:06.215 And so it took a certain amount of perseverance. But when you got reviews that 00:50:06.215 --> 00:50:10.315 said it's impossible, but everybody knows it already, that was pretty discouraging. 00:50:14.095 --> 00:50:14.615 Yeah. 00:50:17.255 --> 00:50:22.755 Bobby, if I could just then follow up on this one, because this notion of lighten 00:50:22.755 --> 00:50:28.575 up, as you say, seems to be really important to actually keep collaborations going, right? 00:50:28.895 --> 00:50:34.755 Now, if we are to return to the payoff matrix, is there a payoff matrix of lightening 00:50:34.755 --> 00:50:40.815 up? How do we think of something like the lightening up in terms of an economic game? 00:50:40.915 --> 00:50:45.255 Because it seems to open for a very different dimension that precisely is about 00:50:45.255 --> 00:50:49.675 exploring trust, exploring kind of a long-term being in the game together, 00:50:49.855 --> 00:50:54.455 exploring ways of, in a sense, creating a shared understanding or exploring 00:50:54.455 --> 00:50:55.315 the shared understanding. 00:50:55.655 --> 00:51:00.355 Does that have a payoff matrix in a more traditional sense? Or how do we think of it? 00:51:03.067 --> 00:51:08.887 Well, I would think of it as not in the payoffs, but in the discount rate. 00:51:09.067 --> 00:51:12.647 That is to say, how patient or impatient are you? 00:51:12.847 --> 00:51:22.067 How much does the future count as opposed to getting immediate awards? 00:51:22.307 --> 00:51:26.467 And if the collaboration, for example, is set up in a business organization 00:51:26.467 --> 00:51:30.527 where you have one month in order to increase profits or else you're a failure, 00:51:30.527 --> 00:51:34.147 Well, then, obviously, you can't lighten up. You've got to work hard on that problem. 00:51:34.887 --> 00:51:38.527 But if you take a long-term perspective because your discount rate isn't too 00:51:38.527 --> 00:51:42.387 high because you value the future, then you might come to the conclusion with 00:51:42.387 --> 00:51:46.227 your colleagues in the collaboration that what you're doing now isn't working. 00:51:47.347 --> 00:51:51.427 You ought to step back and, for example, examine the premises, 00:51:51.827 --> 00:51:59.027 examine why you structured the the search for a solution in the way you did, 00:51:59.167 --> 00:52:04.027 or whether there's other ideas that are completely different from the ones that 00:52:04.027 --> 00:52:05.387 you've been trying to build on. 00:52:06.987 --> 00:52:10.447 And so I see it not so much as a change in the payoff matrix as a change in 00:52:10.447 --> 00:52:20.787 the discount rate and awareness and a realization that to make progress, you might have to give up, 00:52:21.267 --> 00:52:25.207 making progress in the short run, which wasn't working anyway, 00:52:25.407 --> 00:52:32.207 in order to do some more exploration of distant possibilities that might pay off later. 00:52:33.803 --> 00:52:34.183 Thank you. 00:53:03.803 --> 00:53:07.743 Don't typically think about when we're doing game theoretic analyses generally 00:53:07.743 --> 00:53:11.443 in game theory. We think in terms of extrinsic rewards. 00:53:12.883 --> 00:53:16.383 But there's also an intrinsic reward. 00:53:16.743 --> 00:53:20.383 And here I'm thinking in particular about the Bach group, 00:53:20.643 --> 00:53:27.703 which is for everyone else, this is a group that Bob was a part for a long time, 00:53:27.783 --> 00:53:35.703 a beautiful collaboration in a sense of meeting of fellow scientists that I 00:53:35.703 --> 00:53:40.463 don't think had a specific purpose other than being a place where you could 00:53:40.463 --> 00:53:42.323 share ideas in progress. 00:53:42.703 --> 00:53:50.823 And so you didn't have a sense of a definite goal beyond that in terms of a common publication. 00:53:50.843 --> 00:53:56.143 And yet it was incredibly important to all of you who were participating in it. 00:53:56.423 --> 00:54:00.923 And so a big part of that was the connection between fellow human beings. 00:54:02.003 --> 00:54:09.923 So what role do you think this connectedness between humans plays in fostering 00:54:09.923 --> 00:54:11.723 collaboration, successful collaboration, 00:54:11.783 --> 00:54:16.063 and getting you over some hurdles and making it more robust? 00:54:18.223 --> 00:54:18.903 Yeah. 00:54:21.848 --> 00:54:27.608 You're quite right to emphasize the importance of human relationships, 00:54:27.848 --> 00:54:33.028 and in particular in collaboration, what could be fun of working with other people. 00:54:35.608 --> 00:54:41.588 There's been work, for example, on political mobilization, such as working for 00:54:41.588 --> 00:54:43.908 a candidate or a party or a cause. 00:54:44.948 --> 00:54:50.648 Because I was often found in that setting is that people get together for a 00:54:50.648 --> 00:54:57.368 purpose of accomplishing something, for example, women's rights or sustainability environment. 00:54:57.728 --> 00:55:04.308 But they often stay because of the personal rewards of working with like-minded people. 00:55:05.248 --> 00:55:09.648 So it can often be an extrinsic motivation for a common cause, 00:55:09.888 --> 00:55:14.148 but that the sustaining feature is that they enjoy working together. 00:55:14.308 --> 00:55:15.888 It's sort of like it could be a club. 00:55:16.648 --> 00:55:25.428 And what often helps promote successful collaboration is if at least one of 00:55:25.428 --> 00:55:28.808 the members explicitly attend to their interpersonal relations. 00:55:28.828 --> 00:55:34.428 For example, say to somebody, you know, you've You contribute a lot, 00:55:34.508 --> 00:55:37.248 but why don't you listen to other people a little bit better? 00:55:37.328 --> 00:55:40.988 Well, I'm not very good at being tactful, so that's a bad way to put it. 00:55:41.208 --> 00:55:47.868 But there's often been found in collaborative groups that one person takes on 00:55:47.868 --> 00:55:51.888 the role of a task leader, focusing on the task, 00:55:52.208 --> 00:55:59.188 and somebody else typically takes the role of a solidarity leader who is attentive 00:55:59.188 --> 00:56:04.808 to the interpersonal relationships relationships and works to keep those on keel. 00:56:07.292 --> 00:56:15.752 So, Bob, on the one hand, just a quick reaction to the issue of the lightening up. 00:56:16.912 --> 00:56:22.392 It's very reminiscent of this idea of incubation that goes back to the theory 00:56:22.392 --> 00:56:26.812 of creativity of Helmholtz, where it's like you expose yourself to something, 00:56:27.032 --> 00:56:33.892 and then there's a period of distancing, of incubation, after which insight would follow. 00:56:33.892 --> 00:56:38.932 And I find it a curious overlap, which might be worth pursuing on some other occasion. 00:56:39.652 --> 00:56:45.812 On the other hand, if I listen to you, I also feel that there's a collision 00:56:45.812 --> 00:56:49.452 here in conceptualizations and interpretations. 00:56:49.452 --> 00:56:54.612 Because when we talked about the cancer example, how it informs us about human collaboration, 00:56:54.992 --> 00:57:01.932 we actually went into a model which would deny intentionality and goal setting 00:57:01.932 --> 00:57:06.652 playing a role because it would play out as a biological mechanical process. 00:57:06.652 --> 00:57:10.252 However, when I asked you about the architecture of collaboration, 00:57:10.572 --> 00:57:12.092 the first thing you said was goals. 00:57:13.332 --> 00:57:19.052 So I see a bit of a gap between these two positions that to me also show that 00:57:19.052 --> 00:57:22.912 there's still a lot of work to be done in this field of understanding collaboration. 00:57:23.892 --> 00:57:28.392 So I would say, what would you see as the most critical question that we have 00:57:28.392 --> 00:57:31.092 to answer to make progress on this topic? 00:57:42.052 --> 00:57:45.412 Well, that's a big question, and what's the most critical question to answer 00:57:45.412 --> 00:57:46.932 to make progress on this topic? 00:57:47.212 --> 00:57:50.932 That's what you said, and I'd say to deal with that, we should lighten up, 00:57:50.972 --> 00:57:55.172 as to say not try to give a 30-second answer right off the top of my head, 00:57:55.292 --> 00:57:57.932 but thinking about what's the most critical question. 00:57:57.932 --> 00:58:02.972 Well, but you know, Bob, there's another thing about creativity, 00:58:03.012 --> 00:58:05.332 about a historical prepared mind, right? 00:58:05.432 --> 00:58:09.812 So I had hoped that it would just come out like that because you cannot sleep 00:58:09.812 --> 00:58:12.152 at night because of that one question you need to answer. 00:58:16.163 --> 00:58:25.023 That's sometimes true. Sometimes you have a question that you're pondering and it just bothers you. 00:58:28.843 --> 00:58:37.423 I had such a thing in graduate school when I was wondering, 00:58:37.603 --> 00:58:45.203 based on game theory, How would you measure whether a situation is very much like a zero-sum game, 00:58:45.343 --> 00:58:50.743 or is it very much like a partnership game where the interests are completely 00:58:50.743 --> 00:58:54.463 aligned and whatever helps one helps the other, where zero-sum, 00:58:54.463 --> 00:58:56.243 of course, whatever helps one hurts the other? 00:58:56.383 --> 00:58:58.803 And how would you measure that? 00:59:00.683 --> 00:59:09.943 And I couldn't do it. uh and um i but i put it aside and several years later 00:59:09.943 --> 00:59:17.163 uh i came back to potter to get it when i woke up one morning uh i had what 00:59:17.163 --> 00:59:20.663 i regarded as the answer and the mathematical proof that it was, 00:59:22.043 --> 00:59:25.743 um not all the conditions and it was the only measure that didn't meet all the 00:59:25.743 --> 00:59:30.083 conditions and then i said well why did it take me so long and the answer it 00:59:30.083 --> 00:59:31.723 turned out is i had to forget get something. 00:59:33.283 --> 00:59:35.303 One of the factors I thought that 00:59:35.303 --> 00:59:42.283 you could put into this turned out to be irrelevant and you don't need it. 00:59:44.123 --> 00:59:48.403 After several years, when I came back to the problem, I didn't bother. 00:59:48.663 --> 00:59:52.523 I didn't remind myself of that and that was good. 00:59:53.923 --> 01:00:01.603 Even when you have a question that's really bothering you, lightening up can be very helpful. 01:00:01.943 --> 01:00:06.083 Right. So my last two questions to bring us to the finish line. 01:00:06.443 --> 01:00:11.743 Do you believe that humans will ever manage to maintain sustainable collaboration? 01:00:15.322 --> 01:00:21.202 Well, yes, in some regards. I mean, to take an example, we have a sustainable 01:00:21.202 --> 01:00:24.062 collaboration about what the U.S. dollar means. 01:00:24.482 --> 01:00:28.262 Any place in the world, you can hand a person a U.S. dollar, 01:00:28.462 --> 01:00:30.302 and they'll give you an apple or whatever. 01:00:30.702 --> 01:00:33.722 So there's understanding. It's just a piece of paper. 01:00:35.682 --> 01:00:38.622 Why should anybody take a green piece of paper and give you an apple? 01:00:39.002 --> 01:00:41.682 They're not going to eat the paper. And the reason, of course, 01:00:41.762 --> 01:00:45.442 is that we all have a common understanding that somebody else will take, 01:00:45.522 --> 01:00:48.242 they can use that piece of paper to buy shoes or whatever. 01:00:49.162 --> 01:00:53.762 So there's an example, and that's sustainable. It's been around for a long time, 01:00:53.982 --> 01:00:58.522 certainly a dominant world currency for almost 100 years, and it's likely to 01:00:58.522 --> 01:00:59.642 be sustained for quite a while. 01:00:59.722 --> 01:01:03.082 Maybe not the only major currency that's universally recognized. 01:01:03.082 --> 01:01:06.042 Nice but uh so that that's an example of sustainable 01:01:06.042 --> 01:01:09.422 understanding of a situation that promotes collaboration 01:01:09.422 --> 01:01:16.222 at a very high level now uh we need to have a sustained collaboration on climate 01:01:16.222 --> 01:01:23.302 change and that's a toffee because everybody wants to burn coal because it's 01:01:23.302 --> 01:01:27.762 the cheapest source of energy today but of course it's terrible and so we can 01:01:27.762 --> 01:01:29.402 talk to each other They didn't say, 01:01:29.462 --> 01:01:32.142 oh, we'll try to reduce our coal if you try to reduce your coal. 01:01:32.262 --> 01:01:42.182 But there's no formal agreement or mutual monitoring or punishments for a violation of the trust. 01:01:42.902 --> 01:01:47.102 And so there is a case where sustainable cooperation is absolutely essential. 01:01:47.362 --> 01:01:51.782 But it's really, really hard, partly because we're organized into the nation states. 01:01:53.402 --> 01:01:58.562 So if you now would be able to change one feature of humans, 01:01:58.822 --> 01:02:02.962 anything, through magic, what would that one feature be? 01:02:04.742 --> 01:02:09.402 Do nothing. And the reason I say that is if you change something, 01:02:09.622 --> 01:02:16.202 I don't know that that would actually be an improvement considering all the 01:02:16.202 --> 01:02:19.782 very difficult to understand side effects. 01:02:21.082 --> 01:02:26.302 Right so I don't know you might think well I'd like to eliminate disease well 01:02:26.302 --> 01:02:30.842 but then if people live much much longer you know there's a lot of things to 01:02:30.842 --> 01:02:32.102 think about how good that is, 01:02:34.542 --> 01:02:40.102 so that's why I say if I was to change one thing I'd do nothing great well thanks 01:02:40.102 --> 01:02:43.922 a lot thank you very much for this conversation hi you listened to one of our 01:02:43.922 --> 01:02:45.942 podcasts in the series on collaboration, 01:02:46.982 --> 01:02:50.322 produced by the Ernst Trumman Forum and the Convergent Science Network. 01:02:51.042 --> 01:02:54.002 You can find more episodes on our website.