1 00:00:01,880 --> 00:00:08,982 The topic of this section is dishonesty, the ways we lie and cheat but still think 2 00:00:08,982 --> 00:00:13,900 of our self as good people. A guy goes into a church and goes into 3 00:00:13,900 --> 00:00:19,574 confession, and he says to the priest, forgive me Father for I have sinned. 4 00:00:19,575 --> 00:00:22,587 And the father said, what have you done my son? 5 00:00:22,588 --> 00:00:28,577 And the guy said, I'm 72 years old and I just had sex with two 25 year old twins. 6 00:00:28,578 --> 00:00:33,908 And the priest said, this is terrible. Please say 700 times Hail Mary and walk 7 00:00:33,908 --> 00:00:38,728 around the church seven times, and give a $1,000 to charity. 8 00:00:38,728 --> 00:00:42,687 And how long has it been since your last confession? 9 00:00:42,688 --> 00:00:46,428 I said, I've never been to confession. And the priest said, you're 72 years old 10 00:00:46,428 --> 00:00:49,857 and you've never been to confession? How can that be? 11 00:00:49,858 --> 00:00:54,708 The guy says, I'm Jewish. So then, the priest said, so what are you 12 00:00:54,708 --> 00:00:59,500 doing here and why you telling me? And the guy said, I'm telling everybody. 13 00:00:59,501 --> 00:01:05,232 What is it about dishonesty? When we think about dishonesty, we have a 14 00:01:05,232 --> 00:01:08,728 very simple model of dishonesty. We have a model that people do the cost 15 00:01:08,728 --> 00:01:10,860 benefit analysis. What do we tend to gain? 16 00:01:10,860 --> 00:01:13,757 What do we tend to lose? Is this worth it? 17 00:01:13,758 --> 00:01:17,800 But it turns out that throughout the studies you will see in this section, this 18 00:01:17,800 --> 00:01:22,612 is not actually a good model. Instead, lots of other things are driving 19 00:01:22,612 --> 00:01:27,861 us to both, be honest and be dishonest. And as you will see, in more nuance and 20 00:01:27,861 --> 00:01:32,689 complete understanding of this problem is both going to help us figure out what's 21 00:01:32,689 --> 00:01:36,146 going on. And also help us figure out how to fight 22 00:01:36,146 --> 00:01:56,304 dishonesty. So, first of all, try to be honest and ask 23 00:01:56,304 --> 00:02:13,855 yourself whether you have cheated in the last year. 24 00:02:13,856 --> 00:02:19,317 If you have, raise your hand. What about the last month? 25 00:02:19,318 --> 00:02:22,471 What about the last week? What about today? 26 00:02:22,471 --> 00:02:28,310 The reality's, as you probably realized, that we lie and cheat lots of the time. 27 00:02:28,310 --> 00:02:32,526 And at the same time, while we do those actions, we don't think of our self as 28 00:02:32,526 --> 00:02:36,286 being immoral. And that's kind of the dilemma I want to 29 00:02:36,286 --> 00:02:39,440 talk about. We often think about, crime, as a few 30 00:02:39,440 --> 00:02:44,315 individuals who steal a lot of money and cheat a lot, but I want us to think about 31 00:02:44,315 --> 00:02:49,190 the cheating that we can all do, that we can all be part of, and how does it really 32 00:02:49,190 --> 00:02:53,117 work. Okay, so I actually started thinking about 33 00:02:53,118 --> 00:02:57,717 dishonesty when Enron kind of collapsed all of a sudden. 34 00:02:57,718 --> 00:03:02,698 And my question, the question that puzzled me at the time was, what was the cause for 35 00:03:02,698 --> 00:03:07,442 the collapse of Enron? Were there three bad apples that planned 36 00:03:07,442 --> 00:03:11,250 and plotted and executed this incredible accounting scheme? 37 00:03:11,250 --> 00:03:14,320 Is this the best explanation for what's going on? 38 00:03:14,320 --> 00:03:19,934 Or, is the explanation to a lots of people at Enron and around Enron that were not 39 00:03:19,934 --> 00:03:25,884 active vicious people, but were turning a blind eye to something really bad that was 40 00:03:25,884 --> 00:03:31,217 going on and benefiting them at the same time, kind of wishful blindness? 41 00:03:31,218 --> 00:03:36,389 And I happened to meet John Perry Barlow, who used to write lyrics and manage a 42 00:03:36,389 --> 00:03:41,890 group called the, Grateful Dead. And at some point John Perry Barlow, who's 43 00:03:41,890 --> 00:03:45,516 a wonderful person, became a consultant for Enron. 44 00:03:45,516 --> 00:03:49,802 And this, of course, could be thought of as a commentary of the times. 45 00:03:49,802 --> 00:03:54,464 But beyond that, when I met John Perry Barlow, he told me that when he was a 46 00:03:54,464 --> 00:03:59,348 consultant for Enron, not only was he oblivious to what was going on, he created 47 00:03:59,348 --> 00:04:03,788 an ideology for, from himself that protected him from seeing all the bad 48 00:04:03,788 --> 00:04:08,350 things that were happening around. So, this was not just wish for blindness, 49 00:04:08,350 --> 00:04:12,077 this was active blindness. When he was trying to create an ideology 50 00:04:12,077 --> 00:04:16,633 to make himself not see all the bad things that he didn't want to see, because he was 51 00:04:16,633 --> 00:04:18,798 getting paid. So, this question of how do we want to 52 00:04:18,798 --> 00:04:22,017 think about dishonesty. Do we want to think about it as the 53 00:04:22,017 --> 00:04:27,372 property of a few bad apples, or do we want to think about it, about something 54 00:04:27,372 --> 00:04:30,833 that we can all do, is actually very important? 55 00:04:30,834 --> 00:04:35,667 Because, for example, what should we do about dishonesty as a company? 56 00:04:35,668 --> 00:04:40,027 If we think it's only bad apples, then we should improve our HR processes, we should 57 00:04:40,027 --> 00:04:43,917 just not hire bad apples. Or, if people become bad apples, we should 58 00:04:43,917 --> 00:04:47,296 fire them later on. But if everybody's capable of that, that's 59 00:04:47,296 --> 00:04:49,605 not very easy. The same thing we can ask about 60 00:04:49,605 --> 00:04:52,518 governemnt. What should the government do with bad 61 00:04:52,518 --> 00:04:54,956 apples? If it's just a few of them, we can think 62 00:04:54,956 --> 00:04:58,895 about how to kick them off. But there's something about the fabric of 63 00:04:58,895 --> 00:05:03,210 how the government governs and how people are influenced by each other. 64 00:05:03,210 --> 00:05:06,180 Then, we need to think very differently about these structures. 65 00:05:06,180 --> 00:05:09,097 And, of course, the same thing happens with individuals cheating. 66 00:05:09,097 --> 00:05:16,375 So, let's take a step back and let's say, what do we know about cheating and how do 67 00:05:16,375 --> 00:05:19,289 we measure it? So, let's start with the measurement 68 00:05:19,289 --> 00:05:21,311 issue. I do experiments, I try to create 69 00:05:21,311 --> 00:05:25,406 something that I can measure a bit more precisely than asking people for their 70 00:05:25,406 --> 00:05:27,930 opinions about something or what they would do. 71 00:05:27,930 --> 00:05:33,046 And here is the method I came up with. We gave people a sheet of paper with 20 72 00:05:33,046 --> 00:05:36,712 simple math problems. These were math problems that everybody 73 00:05:36,712 --> 00:05:40,294 could solve given enough time. You would just have to look for the 74 00:05:40,294 --> 00:05:44,477 different cells in the matrix and look for the two numbers that add to ten. 75 00:05:44,478 --> 00:05:48,093 There's lots of those combinations. And when you find the one that, the two 76 00:05:48,093 --> 00:05:50,850 numbers that add to ten, you know that you've found it. 77 00:05:50,850 --> 00:05:55,746 But, it takes a while to find this. So, we give people a sheet of these 78 00:05:55,746 --> 00:05:59,050 problems, we give them 20 problems in one sheet. 79 00:05:59,050 --> 00:06:03,259 And we say, why don't you go ahead and solve as many of those as you can in the 80 00:06:03,259 --> 00:06:07,744 next five minutes and we will pay you, let's say, a dollar per question for each 81 00:06:07,744 --> 00:06:11,726 one you get correct. We say to people, please flip the page and 82 00:06:11,726 --> 00:06:15,689 get go, start, and people start. They work as fast as they can. 83 00:06:15,690 --> 00:06:20,402 And the end of the five minutes we say, stop, put your pencil down and count how 84 00:06:20,402 --> 00:06:24,771 many questions you got correctly. Again, these are questions that are quite 85 00:06:24,771 --> 00:06:27,410 self-evident whether you got the right or not. 86 00:06:27,410 --> 00:06:32,402 People count how many ones they got correctly and then we say, now please take 87 00:06:32,402 --> 00:06:37,316 the sheet of paper and go to the back of the room, and please shred the sheet of 88 00:06:37,316 --> 00:06:39,502 paper. And then, come back to the front of the 89 00:06:39,502 --> 00:06:42,570 room and tell me how many questions you got correctly, and I will pay you 90 00:06:42,570 --> 00:06:45,531 accordingly. People do this, they count, they shred, 91 00:06:45,531 --> 00:06:49,953 the come back to the front of the room, and they announce how many questions they 92 00:06:49,953 --> 00:06:54,080 got correctly. People on average announced that they 93 00:06:54,080 --> 00:06:56,480 solved six problems and they get paid accordingly. 94 00:06:56,480 --> 00:07:01,840 What the people in this experiment do not know is that we fixed the shredder so the 95 00:07:01,840 --> 00:07:04,758 shredder only shreds the sides of the page. 96 00:07:04,758 --> 00:07:09,046 So, when you put the page in, it gives you the satisfaction of the noise of the 97 00:07:09,046 --> 00:07:13,194 shredder and the slowness of the progression but the main part of the page 98 00:07:13,194 --> 00:07:16,132 remains intact. And we can jump in later on and find out 99 00:07:16,132 --> 00:07:18,777 how many problems people really solve correctly. 100 00:07:18,778 --> 00:07:23,146 And what do we find when we do it? We find that people solve 4 problems, but 101 00:07:23,146 --> 00:07:27,108 they report to be solving 6. Now you can ask the question of, how is 102 00:07:27,108 --> 00:07:31,755 this distribution of cheating looks like? Is it most people are honest, reporting 103 00:07:31,755 --> 00:07:34,828 four and solving really four? And a few people are cheating a lot, and 104 00:07:34,828 --> 00:07:38,702 they are shifting the average? No, it turns out that lots and lots of 105 00:07:38,702 --> 00:07:42,140 people cheat. In fact, we find very few people that 106 00:07:42,140 --> 00:07:47,132 cheat in a egregious way, and instead we find lots and lots of people who cheat 107 00:07:47,132 --> 00:07:50,222 just a little. Now, you can say, okay. 108 00:07:50,222 --> 00:07:55,036 So, let's find out why are people and why are people cheating a little bit. 109 00:07:55,036 --> 00:08:00,204 So, to examine this, let's first think about what is the economic rational theory 110 00:08:00,204 --> 00:08:03,066 of cheating. An economic rational theory of cheating 111 00:08:03,066 --> 00:08:05,880 goes back to the famous Chicago economist, Gary Becker. 112 00:08:05,880 --> 00:08:10,096 And the story goes that Gary Becker is one day driving to the University of Chicago 113 00:08:10,096 --> 00:08:14,412 looking for a place to park his car. He can't find a legal parking spot and 114 00:08:14,412 --> 00:08:18,680 he's about to be late for a meeting. And Gary Becker does a quick calculation. 115 00:08:18,680 --> 00:08:22,740 He said, what is the benefit of getting to the meeting on time? 116 00:08:22,740 --> 00:08:28,250 What is the cost of parking illegally? And he makes, he compares those things. 117 00:08:28,250 --> 00:08:31,539 Decide to park illegally and gets to the meeting on time. 118 00:08:31,540 --> 00:08:36,205 And as he walks around the University of Chicago to the meeting, he thinks that 119 00:08:36,205 --> 00:08:40,091 what he has done is basically a cost/benefit analysis about a crime, a 120 00:08:40,091 --> 00:08:43,480 minor crime, but a crime. He was thinking about the benefit of 121 00:08:43,480 --> 00:08:47,512 getting to the meeting on time, he was thinking about the risk of being caught 122 00:08:47,512 --> 00:08:51,733 and getting a ticket and the cost of such parking ticket, and he was comparing the 123 00:08:51,733 --> 00:08:54,775 two of them. And in comparing the benefit and the cost, 124 00:08:54,775 --> 00:08:58,525 he decided, well, it's worthwhile for him to park illegally, or not. 125 00:08:58,526 --> 00:09:03,630 And from that point on, Gary Becker said that all decisions about moral issues and 126 00:09:03,630 --> 00:09:08,227 dishonesty are like that. They're about the cost and benefits. 127 00:09:08,228 --> 00:09:12,252 So, for example, every time you pass by a grocery store or a gas station, you should 128 00:09:12,252 --> 00:09:14,884 ask yourself, how much money is in the till? 129 00:09:14,885 --> 00:09:18,970 What's the chance of catch me if I rob this grocery store and gas station? 130 00:09:18,970 --> 00:09:21,930 How much time will I get in prison minus the time for good behavior? 131 00:09:21,930 --> 00:09:27,166 And you should do a cost benefit analysis and decide whether to hit the gas station 132 00:09:27,166 --> 00:09:30,069 or not. Realize that in this conception of this 133 00:09:30,069 --> 00:09:35,103 problem, there's no morality. It's just about cost and benefit. 134 00:09:35,104 --> 00:09:40,078 So, that's with this understanding about constant benefit and the rational approach 135 00:09:40,078 --> 00:09:43,397 for dishonesty. Let's take a step back and let's go back 136 00:09:43,397 --> 00:09:48,215 to the method, the little experiment with the matrices, and see, can we test the 137 00:09:48,215 --> 00:09:50,924 Becker theory? So, the Becker theory has three 138 00:09:50,924 --> 00:09:53,324 components. How much money you intend to gain? 139 00:09:53,325 --> 00:09:56,806 What's the probability you will get caught, and how much time will you get in 140 00:09:56,806 --> 00:09:58,870 prison? What's the side of the punishment? 141 00:09:58,870 --> 00:10:02,530 We started with the first one, how much money do you stand to gain? 142 00:10:02,530 --> 00:10:06,670 We gave this sheet to some people and we said, if you solve these problems 143 00:10:06,670 --> 00:10:09,953 correctly, for each on you solve, you will get 10 cents. 144 00:10:09,954 --> 00:10:13,132 To another group you said, for each problem you will solve, you will get 25 145 00:10:13,132 --> 00:10:17,314 cents. Another group we offered 50 cents, another 146 00:10:17,314 --> 00:10:20,826 $1, $2, $5. The highest group was promised $10 per 147 00:10:20,826 --> 00:10:23,926 correct question. We gave them, we gave them five minutes, 148 00:10:23,926 --> 00:10:26,877 they did the task, they shredded the piece of paper, they reported. 149 00:10:26,878 --> 00:10:31,108 How much cheating did we see? And what happened to cheating as a 150 00:10:31,108 --> 00:10:36,060 function of how much money they were promised per correct question? 151 00:10:36,060 --> 00:10:39,714 Before I go to the answer for this, we created another experiment in which we 152 00:10:39,714 --> 00:10:42,710 just describe to people in the same way I'm describing to you. 153 00:10:42,710 --> 00:10:46,225 We describe to people the experiment. And we say, hey, if this was the 154 00:10:46,225 --> 00:10:51,189 experiment, how much do you think people would cheat at 25 cents, 50 cents, $1, $2, 155 00:10:51,189 --> 00:10:55,003 $5, $10? And what people predicted, what they 156 00:10:55,003 --> 00:10:59,287 thought would happen was that cheating would increase as the amount of the reward 157 00:10:59,287 --> 00:11:01,864 goes up. In essence, people had the same theory as 158 00:11:01,864 --> 00:11:05,726 Becker, that cheating is influenced by the amount of money we stand to gain. 159 00:11:05,726 --> 00:11:09,600 And therefore, as the amount of money goes up, so will the level of cheating. 160 00:11:09,600 --> 00:11:13,479 But when we did the experiment for real, what did we find? 161 00:11:13,480 --> 00:11:17,020 We find lots of people cheating just by a little bit. 162 00:11:17,020 --> 00:11:21,268 And this amount of cheating was independent on the amount of money that 163 00:11:21,268 --> 00:11:24,696 they got. In fact, for $10 per correct question, 164 00:11:24,696 --> 00:11:28,870 cheating was slightly lower than for all other amounts. 165 00:11:28,870 --> 00:11:31,975 So, first of all, it doesn't go up. And if anything, it goes down. 166 00:11:31,976 --> 00:11:36,307 That was the first part of the Becker element. 167 00:11:36,308 --> 00:11:38,840 What about the next one? What about the probability of being 168 00:11:38,840 --> 00:11:40,549 caught? We played with that as well. 169 00:11:40,549 --> 00:11:45,352 Some people, we asked them to tear half the sheet and only shred half the sheet 170 00:11:45,352 --> 00:11:50,077 and submit to us the other half so we would have some evidence of what they did 171 00:11:50,077 --> 00:11:54,458 and some evidence were eliminated. Other people we asked to eliminate and 172 00:11:54,458 --> 00:11:58,442 shred the whole sheet of paper. Another group we asked to shred the piece 173 00:11:58,442 --> 00:12:02,522 of paper, leave the room, and pay themselves from a bowl of money that had 174 00:12:02,522 --> 00:12:06,487 over $100 in it. Now, if you think about it, shredding half 175 00:12:06,487 --> 00:12:09,957 the page has some probability of being caught. 176 00:12:09,958 --> 00:12:12,490 Shredding the whole page has less probability of being caught. 177 00:12:12,490 --> 00:12:15,974 Shredding the whole page and leaving and paying yourself with a bowl of money had 178 00:12:15,974 --> 00:12:19,068 even lower probability of being caught. What happened? 179 00:12:19,068 --> 00:12:23,350 Again, before I tell you the real result, we ask people to predict. 180 00:12:23,350 --> 00:12:27,705 What do you think would happen? And people predicted, as the probability 181 00:12:27,705 --> 00:12:30,920 of being caught will go down, the amount of cheating will go up. 182 00:12:30,920 --> 00:12:35,543 Again, suggesting that peoples' naive theories about driving, what's driving 183 00:12:35,543 --> 00:12:39,167 this dishonesty, is very much in line with the Becker theory. 184 00:12:39,168 --> 00:12:41,420 But what did we find when we did the experiments for real? 185 00:12:41,420 --> 00:12:46,070 We found that lots of people cheated just by a little bit independent of the 186 00:12:46,070 --> 00:12:51,020 probability of being caught. Now, you might say something else. 187 00:12:51,020 --> 00:12:55,180 You might say, you know, maybe what's going on here is the people think they can 188 00:12:55,180 --> 00:12:59,342 cheat a little bit and not being caught. If you say, oh, you know, if I really saw 189 00:12:59,342 --> 00:13:02,120 four and I only report six, nobody will be suspicious. 190 00:13:02,120 --> 00:13:05,979 But if I report seven, all of a sudden people would become suspicious of that. 191 00:13:05,980 --> 00:13:09,980 So, we tried that as well. We told some people, we say, hey, the 192 00:13:09,980 --> 00:13:14,127 average person in this experiment solves four problems. 193 00:13:14,128 --> 00:13:17,788 We told another group that the average person in the experiment solves eight 194 00:13:17,788 --> 00:13:20,429 problems. Now, if people just don't want to be 195 00:13:20,429 --> 00:13:25,257 detected when they're told other people have solved four, maybe they would claim 196 00:13:25,257 --> 00:13:27,440 six. And when they were told other people would 197 00:13:27,440 --> 00:13:30,819 solve eight, they could report ten. But what happened? 198 00:13:30,820 --> 00:13:33,180 No difference between those two conditions. 199 00:13:33,180 --> 00:13:37,810 In both of them, people really solved four and reported six. 200 00:13:37,810 --> 00:13:42,162 So again, it doesn't look like it's about the probability of being caught or about 201 00:13:42,162 --> 00:13:45,763 wanting to go above some threshold, and not even about the reward. 202 00:13:45,764 --> 00:13:50,690 So, what is going on here? You could say, well, people don't care 203 00:13:50,690 --> 00:13:56,140 about the rewards and punishment. How much they stand to gain, probability 204 00:13:56,140 --> 00:13:59,736 of being caught. People are just moral, wonderful, 205 00:13:59,736 --> 00:14:03,006 creatures. Maybe in Freudian terms, it would be the 206 00:14:03,006 --> 00:14:05,618 super ego. We have this notion of society about what 207 00:14:05,618 --> 00:14:08,904 is appropriate and inappropriate and we've internalized that. 208 00:14:08,905 --> 00:14:11,190 And that's basically what's ruling our behavior. 209 00:14:11,190 --> 00:14:14,378 What kind of angels? But, you know, we can't say that as well 210 00:14:14,378 --> 00:14:16,540 because we see lots of people are cheating. 211 00:14:16,540 --> 00:14:19,268 Just a little bit but lots of people are cheating. 212 00:14:19,268 --> 00:14:23,560 So, we have this tension that people are dishonest but at the same time people 213 00:14:23,560 --> 00:14:29,800 don't react to their economic incentives. So, what we need is a new model of what 214 00:14:29,800 --> 00:14:37,144 causes people to be dishonest that is not relying solely on economic incentive of 215 00:14:37,144 --> 00:14:43,732 the amount of money to be gained, probability of being caught, and time in 216 00:14:43,732 --> 00:14:44,642 prison.