1
00:00:06,580 --> 00:00:09,139
Speaker 1: Hey, welcome to Stuff to Blow Your Mind. My name

2
00:00:09,200 --> 00:00:12,869
Speaker 1: is Robert Lamb. Today we have a Buffer Week Vault

3
00:00:12,930 --> 00:00:16,410
Speaker 1: episode for you. We are gathering our strength. We are

4
00:00:16,470 --> 00:00:20,370
Speaker 1: preparing for the October ahead, and we're going to try

5
00:00:20,390 --> 00:00:24,390
Speaker 1: and put out a whole lot of excellent Halloween content.

6
00:00:24,430 --> 00:00:26,550
Speaker 1: So get excited for that. And today we're going to

7
00:00:26,590 --> 00:00:29,050
Speaker 1: let you have it with The Devil You Know Part 1.

8
00:00:29,050 --> 00:00:31,300
Speaker 1: It originally published 8-26-2025.

9
00:00:32,479 --> 00:00:32,940
Speaker 2: Let's do it.

10
00:00:36,640 --> 00:00:47,010
Speaker 1: Welcome to Stuff to Blow Your Mind, a production of iHeartRadio. Hey,

11
00:00:47,050 --> 00:00:49,150
Speaker 1: welcome to Stuff to Blow Your Mind. My name is

12
00:00:49,190 --> 00:00:49,650
Speaker 1: Robert Lamb.

13
00:00:50,310 --> 00:00:52,910
Speaker 2: And I'm Joe McCormick. And in today's episode, we're going

14
00:00:52,930 --> 00:00:55,630
Speaker 2: to begin a series in which we look at a

15
00:00:55,710 --> 00:01:00,850
Speaker 2: phenomenon studied in psychology, economics, and decision theory, which is

16
00:01:00,910 --> 00:01:08,200
Speaker 2: called ambiguity aversion. Ambiguity aversion is when we prefer risks

17
00:01:08,380 --> 00:01:13,940
Speaker 2: with known probabilities over risks with unknown probabilities. Even if

18
00:01:13,980 --> 00:01:16,970
Speaker 2: you have no reason for thinking that the unknown risks

19
00:01:17,050 --> 00:01:20,110
Speaker 2: will be worse than the known risks. And we'll get

20
00:01:20,130 --> 00:01:23,590
Speaker 2: to some concrete examples and foundational research on this in

21
00:01:23,630 --> 00:01:26,470
Speaker 2: a bit. But to an approximation, a good way to

22
00:01:26,490 --> 00:01:31,500
Speaker 2: think about this is that ambiguity aversion is the sentiment

23
00:01:31,660 --> 00:01:35,320
Speaker 2: expressed in the saying, better the devil you know than

24
00:01:35,360 --> 00:01:38,440
Speaker 2: the devil you don't. This is an aphorism I feel

25
00:01:38,459 --> 00:01:41,880
Speaker 2: like people often quote when they're about to justify making

26
00:01:41,940 --> 00:01:46,699
Speaker 2: a really bad or indefensible decision. And the whole point

27
00:01:46,880 --> 00:01:49,640
Speaker 2: of this saying is that the devil you know is

28
00:01:49,700 --> 00:01:53,550
Speaker 2: already bad. That's usually implied. And there's no indication that

29
00:01:53,580 --> 00:01:56,250
Speaker 2: the devil you don't know will be worse. It could

30
00:01:56,290 --> 00:02:00,010
Speaker 2: well be a kinder, gentler devil with a less pointy pitchfork.

31
00:02:00,470 --> 00:02:05,190
Speaker 2: But still, we often find ourselves thinking, I'd just rather

32
00:02:05,270 --> 00:02:07,350
Speaker 2: stick with the one where I know how pointy it

33
00:02:07,410 --> 00:02:08,090
Speaker 2: is up front.

34
00:02:08,830 --> 00:02:11,390
Speaker 1: Yeah, it's often a form of indecision. Like, okay, I

35
00:02:11,430 --> 00:02:14,910
Speaker 1: refuse to do anything about the current devil situation. But,

36
00:02:14,930 --> 00:02:18,579
Speaker 1: you know, at least I know this devil. Next devil

37
00:02:18,620 --> 00:02:20,120
Speaker 1: could be worse. I don't want a worse devil.

38
00:02:20,600 --> 00:02:23,540
Speaker 2: That's right. This is another one of those aphorisms. We

39
00:02:23,560 --> 00:02:26,820
Speaker 2: were talking about this just recently in the St. Swithin episode.

40
00:02:26,840 --> 00:02:29,990
Speaker 2: It's one of these aphorisms that I think is only

41
00:02:30,160 --> 00:02:34,730
Speaker 2: wisdom given certain assumptions or conditions that are not really

42
00:02:34,790 --> 00:02:37,650
Speaker 2: part of the saying. Like, as stated, it could be

43
00:02:37,750 --> 00:02:41,049
Speaker 2: purely terrible advice. Like, what if the devil you know

44
00:02:41,130 --> 00:02:43,450
Speaker 2: is really bad? Right. And what if going with the

45
00:02:43,490 --> 00:02:45,570
Speaker 2: devil you don't know gives you a really good shot

46
00:02:45,620 --> 00:02:49,320
Speaker 2: at improving your stay in hell? But also, I think

47
00:02:50,000 --> 00:02:52,560
Speaker 2: it does have a kind of wisdom if you don't

48
00:02:52,700 --> 00:02:55,800
Speaker 2: think of it as advice, but rather as an ironic

49
00:02:55,900 --> 00:02:59,060
Speaker 2: observation about human nature, in which case I think it's

50
00:02:59,090 --> 00:03:01,870
Speaker 2: sort of dead on. This is often how we think.

51
00:03:02,270 --> 00:03:05,490
Speaker 2: And that is the core observation also of the ambiguity

52
00:03:05,630 --> 00:03:06,650
Speaker 2: aversion literature.

53
00:03:07,010 --> 00:03:07,290
Speaker 1: Yeah.

54
00:03:07,340 --> 00:03:07,470
Speaker 2: Yeah.

55
00:03:07,560 --> 00:03:09,739
Speaker 1: And I mean, because you can also also ask, well,

56
00:03:10,020 --> 00:03:11,739
Speaker 1: what if we just didn't have a devil? Is that

57
00:03:11,800 --> 00:03:13,600
Speaker 1: on the table? Can we just not have a devil?

58
00:03:13,660 --> 00:03:16,100
Speaker 1: Why do we have to worry about the devil we

59
00:03:16,180 --> 00:03:19,459
Speaker 1: have versus the devil we don't have yet? It's you know,

60
00:03:19,480 --> 00:03:22,160
Speaker 1: it can be. And I guess it's rather condemning of

61
00:03:22,220 --> 00:03:25,380
Speaker 1: human nature in that aspect as well, where the answer is, well,

62
00:03:25,400 --> 00:03:27,720
Speaker 1: you got to have a devil. We're humans. We make

63
00:03:27,760 --> 00:03:29,900
Speaker 1: our own and you've got to have one in play.

64
00:03:30,080 --> 00:03:32,040
Speaker 2: You should have thought of that before you double parked, Rob.

65
00:03:32,220 --> 00:03:32,419
Speaker 1: Yeah.

66
00:03:32,620 --> 00:03:34,030
Speaker 2: Now you're here in hell with the rest of us.

67
00:03:34,620 --> 00:03:37,500
Speaker 1: There's a rather keen Chinese saying that goes along these

68
00:03:37,540 --> 00:03:41,560
Speaker 1: lines as well. It's a well-known idiom that states that it's, quote,

69
00:03:41,740 --> 00:03:44,380
Speaker 1: easy to dodge the spear in the open, hard to

70
00:03:44,440 --> 00:03:45,880
Speaker 1: avoid a stab in the dark.

71
00:03:46,410 --> 00:03:50,290
Speaker 2: Oh, that's interesting because, yeah, that's the same idea. Like,

72
00:03:50,530 --> 00:03:54,950
Speaker 2: it isn't necessarily true in all circumstances. Like, in the dark,

73
00:03:55,010 --> 00:03:57,940
Speaker 2: the person stabbing at you also can't see you very well.

74
00:03:58,470 --> 00:04:00,300
Speaker 2: So it might be hard to dodge, but it's also

75
00:04:00,360 --> 00:04:02,720
Speaker 2: hard for them to hit you. But in the dark...

76
00:04:02,930 --> 00:04:05,570
Speaker 2: You just don't really know. You don't really know what's

77
00:04:05,620 --> 00:04:09,260
Speaker 2: going on. So it just seems scarier. Yeah, absolutely.

78
00:04:09,780 --> 00:04:12,100
Speaker 1: And it reminds me a bit of another Chinese saying,

79
00:04:12,580 --> 00:04:15,380
Speaker 1: better to be a dog in times of tranquility than

80
00:04:15,420 --> 00:04:18,320
Speaker 1: a human in times of chaos. And this has some

81
00:04:18,350 --> 00:04:21,790
Speaker 1: of the same sentiments as a 14th century Islamic saying

82
00:04:21,950 --> 00:04:26,450
Speaker 1: that is sometimes misattributed as being of Chinese origin itself.

83
00:04:26,770 --> 00:04:29,330
Speaker 1: And that is better a century of tyranny than one

84
00:04:29,390 --> 00:04:30,130
Speaker 1: day of chaos.

85
00:04:31,240 --> 00:04:33,140
Speaker 2: Also don't know if I actually agree with that.

86
00:04:33,660 --> 00:04:37,900
Speaker 1: Well, I don't know. I think about it a lot

87
00:04:39,240 --> 00:04:44,500
Speaker 1: for various reasons. And I do agree with the basic

88
00:04:44,700 --> 00:04:49,320
Speaker 1: sentiment of it when you find yourself dipping into chaos.

89
00:04:50,300 --> 00:04:53,159
Speaker 1: And I'm not talking about like capital C chaos, like

90
00:04:53,240 --> 00:04:56,110
Speaker 1: absolute chaos, but these little moments of chaos and disorder,

91
00:04:56,500 --> 00:04:59,510
Speaker 1: you can easily line yourself up with the idea of like, well, yeah,

92
00:04:59,550 --> 00:05:03,690
Speaker 1: this is why people you know, fall for tyranny. This

93
00:05:03,710 --> 00:05:06,859
Speaker 1: is because they have these moments of chaos and uncertainty,

94
00:05:06,920 --> 00:05:09,520
Speaker 1: and they're like, well, I would just rather have this

95
00:05:09,560 --> 00:05:13,100
Speaker 1: sort of known top-down oppression in place rather than some

96
00:05:13,140 --> 00:05:15,640
Speaker 1: sort of ambiguous chaos bubbling up around me.

97
00:05:16,000 --> 00:05:19,760
Speaker 2: Better something that's definitely bad than something I don't understand

98
00:05:19,820 --> 00:05:20,640
Speaker 2: and can't predict.

99
00:05:21,100 --> 00:05:25,040
Speaker 1: Yeah. Now, ambiguity and chaos, we should note here, are

100
00:05:25,160 --> 00:05:28,380
Speaker 1: not the same thing. Ambiguity is a lack of clarity,

101
00:05:28,760 --> 00:05:31,810
Speaker 1: while chaos is a state of disorder. And the former

102
00:05:31,870 --> 00:05:34,830
Speaker 1: may well lead into the other. And chaos may be

103
00:05:34,910 --> 00:05:38,750
Speaker 1: seen as a state of absolute ambiguity and uncertainty as well.

104
00:05:38,770 --> 00:05:41,440
Speaker 1: So in that respect, it becomes even more difficult when

105
00:05:41,450 --> 00:05:45,260
Speaker 1: we try and weigh chaos and ambiguity in our concerns

106
00:05:45,300 --> 00:05:46,000
Speaker 1: about the future.

107
00:05:46,339 --> 00:05:49,380
Speaker 2: I was trying to think of familiar everyday examples of

108
00:05:49,540 --> 00:05:54,700
Speaker 2: ambiguity aversion. And I think I came across one internally

109
00:05:54,960 --> 00:05:59,770
Speaker 2: in the now commonly acknowledged sentiment of not wanting to

110
00:05:59,810 --> 00:06:03,610
Speaker 2: go out and try anything, you know, that, so like, um,

111
00:06:03,990 --> 00:06:07,349
Speaker 2: do you ever find yourself trying to find a way

112
00:06:07,470 --> 00:06:11,190
Speaker 2: to avoid going through an experience that is new and

113
00:06:11,290 --> 00:06:15,190
Speaker 2: unfamiliar and thus unpredictable to you? So it seems like,

114
00:06:15,210 --> 00:06:17,060
Speaker 2: I don't know, it just seems kind of fraught like

115
00:06:17,120 --> 00:06:20,400
Speaker 2: a hassle. And instead of doing that, choosing to have

116
00:06:20,460 --> 00:06:23,080
Speaker 2: an experience that you pretty much know that you will

117
00:06:23,120 --> 00:06:24,359
Speaker 2: probably not enjoy.

118
00:06:24,380 --> 00:06:29,930
Speaker 1: Uh, yeah. Yeah, I think I know what you're saying here. Yeah,

119
00:06:29,970 --> 00:06:35,190
Speaker 1: I mean, life is perpetually a situation where it's like,

120
00:06:35,250 --> 00:06:37,610
Speaker 1: am I going to stick with what's comfortable or am

121
00:06:37,730 --> 00:06:39,790
Speaker 1: I going to say yes to adventure? Am I going

122
00:06:39,810 --> 00:06:45,690
Speaker 1: to try something new? And I often find myself at

123
00:06:45,730 --> 00:06:48,810
Speaker 1: the crossroads of that moment because there's a huge part

124
00:06:48,870 --> 00:06:50,830
Speaker 1: of me that doesn't want to try new things. And

125
00:06:51,830 --> 00:06:56,120
Speaker 1: I get anxious about the various experiences aligned with those

126
00:06:56,180 --> 00:06:59,170
Speaker 1: kinds of quests. But on the other hand, I have

127
00:06:59,210 --> 00:07:03,020
Speaker 1: to recognize that most of the really fulfilling things in

128
00:07:03,060 --> 00:07:05,680
Speaker 1: my life have been because I said yes to adventure

129
00:07:05,700 --> 00:07:08,460
Speaker 1: and because I took a chance on something and I

130
00:07:08,860 --> 00:07:12,380
Speaker 1: went out of my comfort zone. But then again, we

131
00:07:12,420 --> 00:07:14,660
Speaker 1: call it our comfort zone because it is comfortable. It

132
00:07:14,720 --> 00:07:18,210
Speaker 1: is a nice place. It is a comforting place. So

133
00:07:18,250 --> 00:07:18,950
Speaker 1: it's give and take.

134
00:07:19,310 --> 00:07:22,030
Speaker 2: Yes, I think that's absolutely correct. But I think the

135
00:07:22,070 --> 00:07:24,910
Speaker 2: really weird thing is that sometimes your, quote, comfort zone

136
00:07:25,070 --> 00:07:28,870
Speaker 2: isn't comfortable. I mean, sometimes we prefer... an experience that's

137
00:07:29,010 --> 00:07:31,530
Speaker 2: not even very nice. It's just like something that we

138
00:07:31,570 --> 00:07:34,570
Speaker 2: know we won't like, but it's just, you know, it's familiar.

139
00:07:34,630 --> 00:07:36,920
Speaker 2: We at least know what we're getting with it over

140
00:07:36,980 --> 00:07:39,980
Speaker 2: something that is just question marks. Right, right.

141
00:07:40,300 --> 00:07:45,760
Speaker 1: We'll choose a known okayness over a potentially amazing experience.

142
00:07:46,290 --> 00:07:50,500
Speaker 2: So, One of the most important early writings on ambiguity

143
00:07:50,540 --> 00:07:54,480
Speaker 2: aversion is a classic paper in the Quarterly Journal of

144
00:07:54,560 --> 00:08:00,320
Speaker 2: Economics from 1961 by the American economist, whistleblower, and political

145
00:08:00,420 --> 00:08:05,180
Speaker 2: activist Daniel Ellsberg. The paper was called Risk, Ambiguity, and

146
00:08:05,200 --> 00:08:09,190
Speaker 2: the Savage Axioms. That's Savage with a capital S. That's

147
00:08:09,230 --> 00:08:12,290
Speaker 2: a person's name. It's not describing the axioms as savage.

148
00:08:13,680 --> 00:08:16,320
Speaker 2: And just to go ahead and mention it now, because

149
00:08:16,380 --> 00:08:19,380
Speaker 2: I also use this as a source in writing about

150
00:08:19,740 --> 00:08:24,060
Speaker 2: Ellsberg and this paper's impact. I was also reading a

151
00:08:24,180 --> 00:08:28,580
Speaker 2: paper called Ambiguity and Ambiguity Aversion in the Handbook of

152
00:08:28,620 --> 00:08:31,880
Speaker 2: the Economics of Risk and Uncertainty. This is by Mark J.

153
00:08:31,940 --> 00:08:37,920
Speaker 2: Machina and Marciano Siniscalchi, published by North Holland in 2014.

154
00:08:37,920 --> 00:08:40,790
Speaker 2: This is a sort of general overview of the concept

155
00:08:40,830 --> 00:08:43,630
Speaker 2: of ambiguity aversion and a survey of a bunch of

156
00:08:43,650 --> 00:08:46,630
Speaker 2: different models of it. Now, before we get into the

157
00:08:46,690 --> 00:08:49,990
Speaker 2: idea that Ellsberg explored in this very important paper, Rob,

158
00:08:50,090 --> 00:08:51,890
Speaker 2: you had a bit of bio on Ellsberg, didn't you?

159
00:08:52,550 --> 00:08:57,740
Speaker 1: Yeah, yeah. Daniel Ellsberg, quite an interesting fellow, lived 1931

160
00:08:57,740 --> 00:09:03,360
Speaker 1: through 2023. In addition to Ellsberg's paradox, which we're getting

161
00:09:03,400 --> 00:09:07,260
Speaker 1: into here, He is he's also famous, perhaps more so

162
00:09:07,320 --> 00:09:10,980
Speaker 1: in a general sense, as the Rand Corporation employee who

163
00:09:11,040 --> 00:09:13,580
Speaker 1: photocopied and released what would come to be known as

164
00:09:13,620 --> 00:09:17,400
Speaker 1: the Pentagon Papers. This is a United States Department of

165
00:09:17,420 --> 00:09:21,130
Speaker 1: Defense history of the U.S. 's political and military involvement

166
00:09:21,170 --> 00:09:24,030
Speaker 1: in Vietnam from 1945 through 1968.

167
00:09:24,740 --> 00:09:26,980
Speaker 2: The key point being that Ellsberg believed that getting these

168
00:09:27,020 --> 00:09:30,780
Speaker 2: papers out there was important in informing the public about

169
00:09:31,220 --> 00:09:33,580
Speaker 2: things that they didn't know about how the Vietnam War

170
00:09:33,620 --> 00:09:36,319
Speaker 2: was being prosecuted and represented. Right, right.

171
00:09:36,360 --> 00:09:40,230
Speaker 1: It was very much a whistleblowing act on Ellsberg's part.

172
00:09:41,020 --> 00:09:43,610
Speaker 1: And these papers were published by The New York Times

173
00:09:43,809 --> 00:09:47,489
Speaker 1: in 1971. This would be a decade after the publication

174
00:09:47,530 --> 00:09:51,860
Speaker 1: regarding the Ellsberg paradox. And yeah, they detailed secretive elements

175
00:09:51,940 --> 00:09:54,980
Speaker 1: about the scope of the U.S. military efforts in Vietnam,

176
00:09:55,340 --> 00:09:59,330
Speaker 1: the real sticking points being the deception involving the Gulf

177
00:09:59,370 --> 00:10:02,780
Speaker 1: of Tonkin and unreported expansion of the war, the true

178
00:10:02,820 --> 00:10:07,880
Speaker 1: objective of the war post-1965, which basically was more about

179
00:10:08,300 --> 00:10:11,590
Speaker 1: containing China and preventing a humiliating U.S. defeat rather than

180
00:10:11,809 --> 00:10:17,089
Speaker 1: any specific objectives to help the people of Vietnam, and

181
00:10:17,130 --> 00:10:19,470
Speaker 1: the lack of a clear plan for victory as well.

182
00:10:20,340 --> 00:10:21,720
Speaker 1: So I imagine a lot of you have at least

183
00:10:21,840 --> 00:10:24,100
Speaker 1: heard of the Pentagon Papers, even if you're not familiar

184
00:10:24,140 --> 00:10:28,050
Speaker 1: with the full story, because this incident, this led to

185
00:10:28,270 --> 00:10:33,630
Speaker 1: conspiracy and espionage charges against Ellsberg that were then later dismissed,

186
00:10:34,470 --> 00:10:37,910
Speaker 1: greatly fueled the anti-war movement, as well as distrust in

187
00:10:37,929 --> 00:10:40,740
Speaker 1: the government at the time, and led to the Nixon

188
00:10:40,770 --> 00:10:43,460
Speaker 1: White House's use of the infamous White House plumbers to

189
00:10:43,500 --> 00:10:46,040
Speaker 1: discredit Ellsberg. And of course, all of this ends up

190
00:10:46,080 --> 00:10:49,099
Speaker 1: feeding into the Watergate scandal, which is another one of

191
00:10:49,160 --> 00:10:53,010
Speaker 1: those major historical touchstones from the time period that, again,

192
00:10:53,070 --> 00:10:55,770
Speaker 1: if you're not super familiar with Watergate, you've at least

193
00:10:56,170 --> 00:10:59,250
Speaker 1: heard of Watergate and you've heard about the other, you know,

194
00:10:59,309 --> 00:11:03,209
Speaker 1: different gates out there that allude back to what a

195
00:11:03,250 --> 00:11:03,929
Speaker 1: scandal it was.

196
00:11:05,090 --> 00:11:06,860
Speaker 2: Now everything gets a gate suffix.

197
00:11:07,179 --> 00:11:11,860
Speaker 1: Right. No matter how deserving or undeserving. Now, Ellsberg remained

198
00:11:11,900 --> 00:11:16,900
Speaker 1: an outspoken activist and whistleblower throughout his life, particularly regarding

199
00:11:16,960 --> 00:11:20,960
Speaker 1: the U.S. 's military actions under multiple presidents, including the

200
00:11:21,300 --> 00:11:24,610
Speaker 1: current one, the Russian invasion of Ukraine, and also in

201
00:11:24,660 --> 00:11:28,350
Speaker 1: support of various other whistleblowers. He was a vocal opponent

202
00:11:28,730 --> 00:11:32,370
Speaker 1: of the use and the threat of nuclear weapons and

203
00:11:32,410 --> 00:11:36,090
Speaker 1: the rhetoric around their use. And he is sometimes described

204
00:11:36,230 --> 00:11:40,370
Speaker 1: as a nuclear war planner turned nuclear war activist. And

205
00:11:40,429 --> 00:11:44,150
Speaker 1: his work at the Rand Corporation, did involve such contemplations.

206
00:11:44,570 --> 00:11:46,910
Speaker 2: Yeah, I think a big part of his pitch in

207
00:11:46,980 --> 00:11:51,140
Speaker 2: his anti-war and anti-nuclear weapons activism was like, look, I've

208
00:11:51,260 --> 00:11:54,840
Speaker 2: been there where the power players are executing, you know,

209
00:11:54,880 --> 00:11:58,580
Speaker 2: global military and nuclear strategy. You should be worried. This is,

210
00:11:58,960 --> 00:12:00,080
Speaker 2: things are not okay.

211
00:12:01,000 --> 00:12:04,840
Speaker 1: Yeah, one of his major overarching criticisms of nuclear weapons,

212
00:12:04,920 --> 00:12:07,060
Speaker 1: and I feel like we're probably preaching to the choir

213
00:12:07,340 --> 00:12:08,800
Speaker 1: on this one. I can't imagine we have any big

214
00:12:09,179 --> 00:12:12,850
Speaker 1: nuclear war fans out there. was that their use as

215
00:12:12,890 --> 00:12:16,380
Speaker 1: a deterrent, one of the major positive spins that are

216
00:12:16,420 --> 00:12:21,020
Speaker 1: often given about having a vast nuclear arsenal, this idea

217
00:12:21,120 --> 00:12:24,740
Speaker 1: of the nuclear deterrent was itself highly unreliable and unstable.

218
00:12:25,120 --> 00:12:29,679
Speaker 1: There were simply too many variables, human, technological, situational, that

219
00:12:29,720 --> 00:12:35,250
Speaker 1: could lead to not only catastrophic, but existential unintended consequences.

220
00:12:35,970 --> 00:12:38,010
Speaker 2: Yeah, too many ways for it to go wrong. And

221
00:12:38,050 --> 00:12:40,510
Speaker 2: if it does go wrong, the consequences are too great.

222
00:12:41,059 --> 00:12:43,400
Speaker 1: By the way, Ellsberg has been portrayed, I believe, twice

223
00:12:43,700 --> 00:12:46,449
Speaker 1: in major motion pictures, once by James Spader and once

224
00:12:46,500 --> 00:12:47,370
Speaker 1: by Matthew Rhys.

225
00:12:57,730 --> 00:13:01,550
Speaker 2: So coming back to the Ellsberg paper from 1961, again,

226
00:13:01,570 --> 00:13:04,250
Speaker 2: that's Risk, Ambiguity, and the Savage Axioms. This is the

227
00:13:04,290 --> 00:13:06,630
Speaker 2: really important one. I looked it up. I think it

228
00:13:06,670 --> 00:13:10,550
Speaker 2: had like, I don't know... like tens of thousands of citations.

229
00:13:10,570 --> 00:13:14,890
Speaker 2: This is like a very, very famous and influential economics paper.

230
00:13:15,590 --> 00:13:19,250
Speaker 2: In this paper, he illustrates the idea of ambiguity aversion

231
00:13:19,330 --> 00:13:22,810
Speaker 2: with a number of thought experiments, and this is one

232
00:13:22,850 --> 00:13:28,130
Speaker 2: of them. So this is the three-color urn. Imagine that

233
00:13:28,210 --> 00:13:31,130
Speaker 2: there's an urn, like a jar, that is filled with

234
00:13:31,130 --> 00:13:34,930
Speaker 2: 90 little ping pong balls that are a mix of

235
00:13:34,970 --> 00:13:38,740
Speaker 2: different colors. You cannot see inside the urn. It's opaque.

236
00:13:38,860 --> 00:13:40,960
Speaker 2: But you're going to play a game where you reach

237
00:13:41,020 --> 00:13:43,490
Speaker 2: into the urn and you pull out one ball at

238
00:13:43,520 --> 00:13:46,590
Speaker 2: a time. And you can win prizes based on the

239
00:13:46,670 --> 00:13:49,729
Speaker 2: color of the ball that you pull out. You know

240
00:13:49,830 --> 00:13:53,270
Speaker 2: that 30 of the balls inside the urn are red.

241
00:13:53,690 --> 00:13:56,050
Speaker 2: So since 30 out of 90 are red, there's a

242
00:13:56,050 --> 00:13:58,750
Speaker 2: 1 in 3 chance that a random ball is going

243
00:13:58,770 --> 00:14:01,160
Speaker 2: to be red. The odds there are pretty clear. The

244
00:14:01,320 --> 00:14:05,380
Speaker 2: other 60 balls, however, are a mix of yellow and black.

245
00:14:05,740 --> 00:14:08,740
Speaker 2: And here's the really important part. You don't know the

246
00:14:08,800 --> 00:14:13,000
Speaker 2: composition of that group. So those 60 balls could be

247
00:14:13,000 --> 00:14:16,359
Speaker 2: 30 yellow and 30 black. It could be 60 yellow

248
00:14:16,420 --> 00:14:19,800
Speaker 2: and 0 black. It could be 60 black and 0 yellow,

249
00:14:20,290 --> 00:14:23,910
Speaker 2: or any mix in between. Since you don't know the mix,

250
00:14:24,290 --> 00:14:27,410
Speaker 2: you don't know the probability of drawing a yellow ball

251
00:14:27,630 --> 00:14:30,130
Speaker 2: or a black ball ahead of time. For red, it's

252
00:14:30,130 --> 00:14:34,120
Speaker 2: 1 in 3. For either yellow or black, it's question marks,

253
00:14:34,240 --> 00:14:37,800
Speaker 2: anywhere between 0 and 2 out of 3. Now imagine

254
00:14:37,820 --> 00:14:40,020
Speaker 2: it is your turn to draw a ball out of

255
00:14:40,060 --> 00:14:42,860
Speaker 2: the urn, and you can take one of two bets.

256
00:14:43,520 --> 00:14:47,620
Speaker 2: Bet A, you get $ 100 if your ball is red.

257
00:14:47,840 --> 00:14:50,110
Speaker 2: So if you pick this option, your chance of winning

258
00:14:50,110 --> 00:14:50,950
Speaker 2: $ 100 is 1 in 3. In bet B, you get

259
00:14:50,950 --> 00:14:53,390
Speaker 2: $ 100 if your ball is black. Since you don't know

260
00:14:53,450 --> 00:15:01,590
Speaker 2: the mix of yellow and black, You don't know your

261
00:15:01,670 --> 00:15:05,450
Speaker 2: odds of winning, but they are somewhere between 0 and

262
00:15:05,450 --> 00:15:08,290
Speaker 2: 2 out of 3, with the midpoint being 1 out

263
00:15:08,310 --> 00:15:10,480
Speaker 2: of 3, which is the same as the odds on red.

264
00:15:11,410 --> 00:15:12,250
Speaker 2: Which one do you pick?

265
00:15:13,060 --> 00:15:16,400
Speaker 1: Well, it sounds like bet A is the truer thing, right?

266
00:15:17,120 --> 00:15:19,440
Speaker 2: If somebody offered me this game, I would probably pick

267
00:15:19,480 --> 00:15:22,120
Speaker 2: bet A. Yeah, it just seems safer. It's like, OK,

268
00:15:22,240 --> 00:15:23,930
Speaker 2: I know what I'm dealing with there, 1 in 3.

269
00:15:23,920 --> 00:15:27,340
Speaker 2: The other option, your odds could be better than 1

270
00:15:27,340 --> 00:15:28,980
Speaker 2: in 3, but you just don't know.

271
00:15:29,660 --> 00:15:31,100
Speaker 1: They could be worse. It could be zero.

272
00:15:31,580 --> 00:15:33,840
Speaker 2: They could be worse, could be better. You have no

273
00:15:33,880 --> 00:15:38,540
Speaker 2: way of knowing either way. Ellsberg suggested, and later experiments

274
00:15:38,640 --> 00:15:43,040
Speaker 2: would empirically demonstrate, that most people would bet on red

275
00:15:43,180 --> 00:15:47,150
Speaker 2: instead of black, even though there is no objective reason

276
00:15:47,210 --> 00:15:50,550
Speaker 2: to think that red is a better bet. Following the

277
00:15:50,590 --> 00:15:54,010
Speaker 2: pick red strategy will not necessarily earn you more money

278
00:15:54,050 --> 00:15:56,970
Speaker 2: in this game, but people really do, in fact, just

279
00:15:57,150 --> 00:16:00,770
Speaker 2: like it better to pick red. It feels right. That's

280
00:16:00,810 --> 00:16:03,690
Speaker 2: just what we think we should do. That in itself

281
00:16:03,870 --> 00:16:06,410
Speaker 2: is kind of weird and interesting. Like, why do we

282
00:16:06,470 --> 00:16:10,700
Speaker 2: prefer to avoid ambiguity even when there is no clear

283
00:16:10,760 --> 00:16:14,400
Speaker 2: advantage in doing so? Then there's a really interesting wrinkle.

284
00:16:14,840 --> 00:16:17,300
Speaker 2: Imagine you're playing the same game. You are drawing a

285
00:16:17,480 --> 00:16:21,600
Speaker 2: ball out of an urn. And this time you're given

286
00:16:21,620 --> 00:16:24,440
Speaker 2: assurance that the mix of the balls is the same

287
00:16:24,500 --> 00:16:26,660
Speaker 2: as it was the first time. So whatever was the

288
00:16:26,700 --> 00:16:29,520
Speaker 2: case with With the mix of yellow and black in there,

289
00:16:29,920 --> 00:16:31,400
Speaker 2: it's going to be the same as it was with

290
00:16:31,420 --> 00:16:33,520
Speaker 2: your last draw. And you get a chance to make

291
00:16:33,560 --> 00:16:37,230
Speaker 2: a second bet. You can take either bet C, where

292
00:16:37,250 --> 00:16:40,650
Speaker 2: you get $ 100 if the ball drawn out is either

293
00:16:40,990 --> 00:16:45,810
Speaker 2: yellow or red. And this one has an unknown chance

294
00:16:45,850 --> 00:16:48,120
Speaker 2: of winning because, of course, you don't know how many

295
00:16:48,160 --> 00:16:50,730
Speaker 2: yellow or black balls there are. You know that your

296
00:16:50,770 --> 00:16:53,910
Speaker 2: chance of winning is somewhere between 1 and 3 if

297
00:16:54,010 --> 00:16:56,380
Speaker 2: there are, you know, the 30 red balls and 0

298
00:16:56,380 --> 00:16:59,900
Speaker 2: yellow balls, or 100% if there are 60 yellow balls.

299
00:17:00,580 --> 00:17:05,000
Speaker 2: Or you can take bet D, where you get $ 100

300
00:17:03,940 --> 00:17:07,060
Speaker 2: if the ball is either yellow or black. This one

301
00:17:07,119 --> 00:17:11,070
Speaker 2: has a guaranteed two-thirds chance to win, because we know

302
00:17:11,160 --> 00:17:14,210
Speaker 2: that together yellow and black make 60 of the 90 balls,

303
00:17:14,270 --> 00:17:17,050
Speaker 2: though we don't know what the mix is. In this

304
00:17:17,090 --> 00:17:21,869
Speaker 2: second trial... People will tend to pick bet D, once again,

305
00:17:21,950 --> 00:17:25,109
Speaker 2: because the probabilities are known. With D, it's a guaranteed

306
00:17:25,150 --> 00:17:28,910
Speaker 2: two-thirds chance to win. With bet C, it's somewhere between

307
00:17:28,910 --> 00:17:33,620
Speaker 2: 1 and 3 and 100%. Picking bet D allows them

308
00:17:33,660 --> 00:17:39,780
Speaker 2: to avoid placing bet C, which has ambiguous chances. Here's

309
00:17:39,820 --> 00:17:44,020
Speaker 2: the really interesting thing that Ellsberg pointed out. There's actually

310
00:17:44,220 --> 00:17:48,729
Speaker 2: a contradiction in people's behavior between between the first and

311
00:17:48,810 --> 00:17:52,350
Speaker 2: second round of this game. Assuming you are actually doing

312
00:17:52,369 --> 00:17:55,310
Speaker 2: your best to win the money, in round one, if

313
00:17:55,350 --> 00:17:58,570
Speaker 2: you bet on red instead of black, that implies you

314
00:17:58,630 --> 00:18:02,590
Speaker 2: believe that less than 30 of the unknown 60 balls

315
00:18:02,830 --> 00:18:06,280
Speaker 2: are black, right? That implies that red has a better

316
00:18:06,320 --> 00:18:09,080
Speaker 2: chance of winning. So less than half of them have

317
00:18:09,119 --> 00:18:12,040
Speaker 2: to be black. However, in the second round, if you

318
00:18:12,080 --> 00:18:15,760
Speaker 2: take the second option, picking yellow and black, it implies

319
00:18:15,820 --> 00:18:18,949
Speaker 2: that you think that more than 30 of the unknown

320
00:18:19,010 --> 00:18:22,669
Speaker 2: balls are black. Otherwise, the red and yellow combination option

321
00:18:23,030 --> 00:18:27,430
Speaker 2: would have the better chance of winning, right? These two

322
00:18:27,490 --> 00:18:33,810
Speaker 2: bets in combination make no sense because they imply self-contradictory beliefs.

323
00:18:35,250 --> 00:18:37,970
Speaker 2: It says that I believe that more balls are black

324
00:18:38,310 --> 00:18:41,890
Speaker 2: and less balls are black at the same time. Obviously,

325
00:18:41,990 --> 00:18:45,270
Speaker 2: that can't really be the case. But there could be

326
00:18:45,410 --> 00:18:49,929
Speaker 2: another issue driving this behavior, which is simply that people

327
00:18:50,310 --> 00:18:55,030
Speaker 2: don't like ambiguity and they will pay up for known odds,

328
00:18:55,270 --> 00:19:00,780
Speaker 2: even if doing so implies mutually exclusive assumptions. I was

329
00:19:00,820 --> 00:19:02,740
Speaker 2: trying to think, you know, as I said a minute ago,

330
00:19:02,780 --> 00:19:05,419
Speaker 2: I think I would probably pick red if given the

331
00:19:05,520 --> 00:19:09,260
Speaker 2: option to play this game. Obviously, I sort of already

332
00:19:09,300 --> 00:19:10,920
Speaker 2: knew the deal here, so I can't know what I

333
00:19:10,960 --> 00:19:14,389
Speaker 2: would do going into it blind like most people. But

334
00:19:14,430 --> 00:19:18,080
Speaker 2: I think I would, for no rational reason, prefer the

335
00:19:18,359 --> 00:19:23,260
Speaker 2: uncertainty-reducing options. I think the most likely reason for that is,

336
00:19:23,960 --> 00:19:26,460
Speaker 2: even if I were given assurances that this game was

337
00:19:26,540 --> 00:19:30,320
Speaker 2: on the up-and-up, I would suspect some kind of trick. Like,

338
00:19:30,380 --> 00:19:33,800
Speaker 2: if I'm being asked to bet on uncertainty, there's some

339
00:19:33,920 --> 00:19:37,260
Speaker 2: part of me that just kind of rises up and says, uh-uh,

340
00:19:37,380 --> 00:19:39,459
Speaker 2: there's a scam here that you are not seeing.

341
00:19:39,700 --> 00:19:42,240
Speaker 1: Yeah, yeah. I mean, you'd kind of expect that within the...

342
00:19:43,080 --> 00:19:46,680
Speaker 1: constraints of a testing environment, and you'd definitely expect that

343
00:19:46,859 --> 00:19:50,920
Speaker 1: out of the world. Whatever form this was taking, you'd

344
00:19:50,940 --> 00:19:54,900
Speaker 1: assume that somebody had a vested interest in manipulating how

345
00:19:54,960 --> 00:19:56,959
Speaker 1: you were going to behave to the, how you're going

346
00:19:56,980 --> 00:19:58,470
Speaker 1: to respond to the two possibilities.

347
00:19:58,810 --> 00:20:01,409
Speaker 2: Yeah. Ellsberg also in this paper, by the way, he

348
00:20:01,450 --> 00:20:03,669
Speaker 2: came up with a second illustration of the same principle.

349
00:20:03,690 --> 00:20:06,280
Speaker 2: So if you go reading about this You'll read about

350
00:20:06,300 --> 00:20:09,460
Speaker 2: the three-color urn and then the two-urn experiment. The second

351
00:20:09,480 --> 00:20:12,620
Speaker 2: illustration used two urns instead of one because it made

352
00:20:12,640 --> 00:20:17,930
Speaker 2: the contradiction between the different bets even clearer. And this

353
00:20:18,170 --> 00:20:24,250
Speaker 2: self-contradictory behavior is now called the Ellsberg paradox. Ellsberg himself

354
00:20:24,290 --> 00:20:26,560
Speaker 2: didn't call it that. That's what other people, they put

355
00:20:26,600 --> 00:20:31,960
Speaker 2: his name on it. Anyway, Ellsberg's paper about ambiguity aversion says,

356
00:20:33,020 --> 00:20:36,440
Speaker 2: was not just about observing a weird little quirk of

357
00:20:36,500 --> 00:20:40,840
Speaker 2: human behavior. It was especially of interest to the fields

358
00:20:40,880 --> 00:20:45,719
Speaker 2: of economics and decision theory, modeling how people make decisions,

359
00:20:46,359 --> 00:20:50,040
Speaker 2: because it sought to undermine some of the major underpinnings

360
00:20:50,420 --> 00:20:55,369
Speaker 2: of the fields, specifically providing evidence against some of the

361
00:20:55,440 --> 00:21:00,720
Speaker 2: tenets of a framework called Subjective Expected Utility Theory, or

362
00:21:00,920 --> 00:21:05,660
Speaker 2: Savage's Axioms, named after the American mathematician Leonard Savage, who

363
00:21:05,760 --> 00:21:11,080
Speaker 2: formulated them. Subjective Expected Utility Theory, or SCU theory, is

364
00:21:11,100 --> 00:21:15,960
Speaker 2: a way of modeling how people make decisions when we

365
00:21:16,020 --> 00:21:20,000
Speaker 2: don't have perfect or complete information, which of course in

366
00:21:20,060 --> 00:21:22,899
Speaker 2: life we rarely do, right? So, you know, this is

367
00:21:22,920 --> 00:21:25,300
Speaker 2: going to be relevant in understanding how people behave in

368
00:21:25,400 --> 00:21:29,700
Speaker 2: most real-world situations, most situations in life are not a

369
00:21:29,800 --> 00:21:32,740
Speaker 2: casino table game where you have known odds of winning.

370
00:21:33,320 --> 00:21:35,459
Speaker 1: That's right. Or, you know, or, you know, you can

371
00:21:35,480 --> 00:21:38,300
Speaker 1: easily imagine to a situation where you're, you're playing some

372
00:21:38,320 --> 00:21:40,300
Speaker 1: sort of a game, be it a tabletop role playing

373
00:21:40,340 --> 00:21:42,879
Speaker 1: game or some sort of like a, you know, video game.

374
00:21:43,200 --> 00:21:46,560
Speaker 1: And you're given a choice between two items for your character, right? You,

375
00:21:46,859 --> 00:21:49,280
Speaker 1: you have the stats of those items right there. You know,

376
00:21:49,320 --> 00:21:51,379
Speaker 1: what kind of enemies you go up against in the game.

377
00:21:51,400 --> 00:21:54,090
Speaker 1: Maybe you've played through the game before you have something

378
00:21:54,130 --> 00:21:59,190
Speaker 1: at least approaching perfect knowledge. of a simplified world, and

379
00:21:59,359 --> 00:22:01,939
Speaker 1: our real world is not so simplified, and there are

380
00:22:01,960 --> 00:22:03,040
Speaker 1: so many variables to it.

381
00:22:03,600 --> 00:22:07,399
Speaker 2: Exactly right. So the real world, we're constantly dealing with uncertainty,

382
00:22:07,619 --> 00:22:10,500
Speaker 2: ambiguous situations where we don't have all the information to

383
00:22:10,560 --> 00:22:13,580
Speaker 2: judge what outcomes are most likely, and yet somehow we

384
00:22:13,640 --> 00:22:17,810
Speaker 2: navigate this world making decisions all the time. The question

385
00:22:18,390 --> 00:22:21,669
Speaker 2: being addressed here is, how do people make those decisions

386
00:22:21,750 --> 00:22:24,830
Speaker 2: when they don't actually know what the relative likelihood of

387
00:22:24,869 --> 00:22:30,080
Speaker 2: different outcomes are? Subjective expected utility theory has several rules,

388
00:22:30,119 --> 00:22:34,160
Speaker 2: but a simplified version is that it says even when

389
00:22:34,260 --> 00:22:38,740
Speaker 2: people don't know the objective likelihood of an outcome, they

390
00:22:38,800 --> 00:22:43,240
Speaker 2: form beliefs about how likely that outcome is. And then

391
00:22:43,260 --> 00:22:47,140
Speaker 2: they act consistently with those beliefs in order to maximize

392
00:22:47,200 --> 00:22:51,010
Speaker 2: their personal benefit. I came across a passage by the

393
00:22:51,030 --> 00:22:54,870
Speaker 2: American economist Frank Knight, which I think summarizes this idea well.

394
00:22:55,930 --> 00:22:59,149
Speaker 2: Knight writes, quote, We must observe at the outset that

395
00:22:59,210 --> 00:23:03,949
Speaker 2: when an individual instance, i.e. a one-time event, only is

396
00:23:04,050 --> 00:23:06,930
Speaker 2: at issue, there is no difference for conduct between a

397
00:23:07,010 --> 00:23:12,520
Speaker 2: measurable risk and an unmeasurable uncertainty. The individual, as already observed,

398
00:23:12,660 --> 00:23:15,560
Speaker 2: throws his estimate of the value of an opinion into

399
00:23:15,619 --> 00:23:19,780
Speaker 2: the probability form of A successes in B trials, A

400
00:23:19,859 --> 00:23:24,080
Speaker 2: slash B being a proper fraction, and feels toward it

401
00:23:24,460 --> 00:23:28,870
Speaker 2: as toward any other probability situation. So we essentially, we

402
00:23:29,970 --> 00:23:33,630
Speaker 2: form a belief or a feeling about how likely something is,

403
00:23:34,030 --> 00:23:36,430
Speaker 2: and then we act as if that was like, you know,

404
00:23:36,690 --> 00:23:39,590
Speaker 2: that we know the odds of a coin flip coming

405
00:23:39,650 --> 00:23:41,970
Speaker 2: up heads are one in two, whether or not we're

406
00:23:42,010 --> 00:23:46,550
Speaker 2: actually right about those feelings or beliefs. Ellsberg used the

407
00:23:46,570 --> 00:23:51,950
Speaker 2: principle of ambiguity aversion to say, actually, no, people do

408
00:23:52,090 --> 00:23:55,850
Speaker 2: not always behave as if they have consistent beliefs about

409
00:23:55,930 --> 00:23:58,889
Speaker 2: what is more or less likely to happen. And you

410
00:23:58,910 --> 00:24:02,450
Speaker 2: can prove this because they make bets that would imply

411
00:24:02,690 --> 00:24:07,990
Speaker 2: self-contradictory beliefs in the same situation, meaning they either don't

412
00:24:08,170 --> 00:24:11,930
Speaker 2: actually have or act on consistent beliefs about a situation,

413
00:24:12,600 --> 00:24:16,820
Speaker 2: or they don't always act to maximize their own monetary benefit.

414
00:24:17,460 --> 00:24:22,280
Speaker 2: Ellsberg's point here was that we can have obscure previously

415
00:24:22,460 --> 00:24:27,970
Speaker 2: unacknowledged motivations like the desire to avoid dealing with ambiguity.

416
00:24:28,530 --> 00:24:32,410
Speaker 2: And that particular desire is so strong in some cases

417
00:24:32,510 --> 00:24:36,449
Speaker 2: that it overrides our consistent thinking and a simple dollar

418
00:24:36,510 --> 00:24:40,960
Speaker 2: value understanding of rational self-interest. Uh, and by the way,

419
00:24:41,130 --> 00:24:44,129
Speaker 2: the technical name of the savage axiom that Ellsberg was

420
00:24:44,170 --> 00:24:46,940
Speaker 2: attacking here was it's called the sure thing principle. Uh,

421
00:24:48,720 --> 00:24:51,959
Speaker 2: Another thing to understand about Ellsberg's paper was that he

422
00:24:52,010 --> 00:24:56,030
Speaker 2: was arguing that we need to make a stronger distinction

423
00:24:56,070 --> 00:25:02,640
Speaker 2: between two different concepts. Those are risk and ambiguity. Ellsberg

424
00:25:02,740 --> 00:25:05,800
Speaker 2: used the definition that risk is when we have a

425
00:25:05,840 --> 00:25:09,440
Speaker 2: stake in an outcome and we know the theoretical likelihood

426
00:25:09,480 --> 00:25:12,379
Speaker 2: of that outcome. So this is like the casino games

427
00:25:12,440 --> 00:25:15,209
Speaker 2: or a coin toss. A fair coin toss is 50-50.

428
00:25:15,270 --> 00:25:18,929
Speaker 2: We know what the outcome likelihood is and we can

429
00:25:18,950 --> 00:25:21,810
Speaker 2: take a risk. A fair die roll is one in

430
00:25:21,869 --> 00:25:25,629
Speaker 2: six to get a particular number. Ambiguity is when we

431
00:25:25,800 --> 00:25:29,300
Speaker 2: don't have an objective way of knowing what the probability

432
00:25:29,340 --> 00:25:33,300
Speaker 2: of our desired outcome is. To Ellsberg, ambiguity was, quote,

433
00:25:33,520 --> 00:25:37,439
Speaker 2: the nature of one's information concerning the relative likelihood of events,

434
00:25:37,940 --> 00:25:49,930
Speaker 2: a quality depending on the amount, And so it's worth

435
00:25:50,010 --> 00:25:53,010
Speaker 2: noting that you can have confidence or even what feels

436
00:25:53,050 --> 00:25:56,290
Speaker 2: like certainty or act as if you have confidence or

437
00:25:56,330 --> 00:25:59,650
Speaker 2: certainty without actually being correct. You can go through life

438
00:25:59,690 --> 00:26:04,470
Speaker 2: having high confidence in objectively low probability things. One of

439
00:26:04,510 --> 00:26:09,090
Speaker 2: the actually really useful things about SCU theories, the subjective

440
00:26:09,130 --> 00:26:13,260
Speaker 2: expected utility theory, was that it made things simple and

441
00:26:13,410 --> 00:26:18,480
Speaker 2: easier to understand because it treated risk and ambiguity the same.

442
00:26:19,080 --> 00:26:22,260
Speaker 2: So a person making bets on a die roll will

443
00:26:22,380 --> 00:26:25,139
Speaker 2: know that the probability of rolling a specific number is

444
00:26:25,260 --> 00:26:28,290
Speaker 2: one in six and they will behave accordingly. And a

445
00:26:28,310 --> 00:26:30,590
Speaker 2: person who has a stake in an outcome with an

446
00:26:30,910 --> 00:26:37,940
Speaker 2: unknown probability will, according to Savage's axioms, mentally, even subconsciously,

447
00:26:38,480 --> 00:26:42,520
Speaker 2: form an internal belief about the probability of that outcome.

448
00:26:43,280 --> 00:26:46,719
Speaker 2: So example is you imagine there is a one in

449
00:26:46,800 --> 00:26:50,219
Speaker 2: four chance it's going to rain today. That might not

450
00:26:50,280 --> 00:26:52,600
Speaker 2: have anything to do with reality, but you just take

451
00:26:52,660 --> 00:26:56,350
Speaker 2: that belief on. And you behave as if that is

452
00:26:56,470 --> 00:26:59,300
Speaker 2: that you make all of your decisions based on that belief.

453
00:26:59,420 --> 00:27:02,340
Speaker 2: They will act according to that belief the same way

454
00:27:02,380 --> 00:27:05,340
Speaker 2: they would act to knowing that a coin flip is 50-50.

455
00:27:06,040 --> 00:27:09,960
Speaker 2: And Ellsberg argued, no, that is not how we make decisions.

456
00:27:10,240 --> 00:27:14,660
Speaker 2: SCU theory with Savage's axioms failed to account for other

457
00:27:14,720 --> 00:27:18,990
Speaker 2: complexities in human behavior. For example, our deep desire to

458
00:27:19,109 --> 00:27:23,470
Speaker 2: get away from ambiguity. We don't like ambiguity so strongly

459
00:27:23,490 --> 00:27:27,650
Speaker 2: that We will make decisions that imply self-contradictory beliefs just

460
00:27:27,710 --> 00:27:29,669
Speaker 2: to avoid it. Now, there have been a lot of

461
00:27:29,750 --> 00:27:34,690
Speaker 2: arguments over the years about how to interpret ambiguity aversion.

462
00:27:35,330 --> 00:27:39,810
Speaker 2: Is it actually rational in ways that were not previously recognized?

463
00:27:40,390 --> 00:27:43,730
Speaker 2: Or is it more like some kind of systematic error

464
00:27:43,930 --> 00:27:47,550
Speaker 2: or cognitive bias? We might come back and revisit that

465
00:27:47,590 --> 00:27:51,179
Speaker 2: debate in the subsequent part of the series. But we've

466
00:27:51,200 --> 00:27:54,980
Speaker 2: got a couple other things we wanted to talk about today. First, Rob,

467
00:27:55,020 --> 00:27:57,119
Speaker 2: I know you have a really interesting place you want

468
00:27:57,140 --> 00:27:58,939
Speaker 2: to take this. But first, I wanted to mention the

469
00:27:58,980 --> 00:28:04,020
Speaker 2: idea just of experimental confirmation, because one important thing to

470
00:28:04,040 --> 00:28:07,619
Speaker 2: understand is that Ellsberg's original paper did not have an

471
00:28:07,680 --> 00:28:11,120
Speaker 2: experimental component. It was a thought experiment. And he realized

472
00:28:11,160 --> 00:28:14,429
Speaker 2: the phenomenon needed to be tested with real human subjects.

473
00:28:15,070 --> 00:28:17,770
Speaker 2: In the decades since his paper, it has been tested many,

474
00:28:17,810 --> 00:28:22,440
Speaker 2: many times, many different ways. And generally, it has proved robust.

475
00:28:23,220 --> 00:28:28,370
Speaker 2: There are some nuances and exceptions, but generally people behave

476
00:28:28,410 --> 00:28:31,710
Speaker 2: the way Ellsberg predicted they would. A majority of people

477
00:28:32,130 --> 00:28:36,070
Speaker 2: structure their choices to avoid ambiguity, even when there's no

478
00:28:36,210 --> 00:28:40,030
Speaker 2: clear objective benefit to doing so. And they even make

479
00:28:40,110 --> 00:28:44,740
Speaker 2: bets that would imply against self-contradictory beliefs, according to SCU theory.

480
00:28:45,210 --> 00:28:47,730
Speaker 2: if those bets can get them out of dealing with

481
00:28:47,810 --> 00:28:50,690
Speaker 2: ambiguous probabilities. So I don't have to mess with that.

482
00:28:52,190 --> 00:28:56,730
Speaker 2: Regarding those experiments, my main source here is, again, that

483
00:28:56,770 --> 00:29:02,890
Speaker 2: survey article by Machina and Siniscalchi. Again, that's Ambiguity and

484
00:29:03,010 --> 00:29:07,150
Speaker 2: Ambiguity Aversion from 2014. The authors here list a whole

485
00:29:07,190 --> 00:29:10,540
Speaker 2: bunch of experiments where people... carried out versions of like

486
00:29:10,560 --> 00:29:14,260
Speaker 2: the three color urn or this other experiment that Ellsberg described,

487
00:29:14,300 --> 00:29:18,820
Speaker 2: the two urn experiment. They list Fellner in 1961, which

488
00:29:18,880 --> 00:29:23,040
Speaker 2: found that people would rather take 50-50 odds than unknown odds,

489
00:29:23,100 --> 00:29:26,680
Speaker 2: which could be better or worse. They also cite Becker

490
00:29:26,720 --> 00:29:28,900
Speaker 2: and Brownson, 64, McCrimmon, 68, Slovik and Tversky from 1974,

491
00:29:28,900 --> 00:29:36,760
Speaker 2: Curley and Yates from 89, among others. all generally finding

492
00:29:36,800 --> 00:29:42,790
Speaker 2: confirmation of Ellsberg's predictions about ambiguity. aversion. Quote, Although most

493
00:29:42,850 --> 00:29:46,990
Speaker 2: of these experiments use students as subjects, researchers such as McCrimmon, 65,

494
00:29:46,970 --> 00:29:51,580
Speaker 2: Hogarth and Kuhnreuther, 89, Einhorn and Hogarth, 86, Viscousi and Chesson, 99, Ho, Keller,

495
00:29:51,640 --> 00:30:01,850
Speaker 2: and Kelticke, 2002 and Mafioletti and Centauri 2005 have examined

496
00:30:01,890 --> 00:30:07,110
Speaker 2: the ambiguity preferences of business owners, trade union leaders, actuaries, managers,

497
00:30:07,170 --> 00:30:10,830
Speaker 2: and executives with the same overall findings. So it appears

498
00:30:10,950 --> 00:30:15,940
Speaker 2: quite robust. The authors here also report some interesting findings

499
00:30:16,180 --> 00:30:20,540
Speaker 2: of McCrimmon and Larson from 79, which they looked into

500
00:30:20,620 --> 00:30:23,040
Speaker 2: it and they found that while the majority of people

501
00:30:23,360 --> 00:30:27,780
Speaker 2: did try to avoid ambiguity, some minority of people did

502
00:30:27,840 --> 00:30:31,800
Speaker 2: not and even just chose to embrace it. They also found,

503
00:30:32,260 --> 00:30:35,790
Speaker 2: as you might expect, that our relative tolerance for ambiguity

504
00:30:35,850 --> 00:30:39,410
Speaker 2: went up or down depending on how good the known

505
00:30:39,690 --> 00:30:44,090
Speaker 2: odds were in the known odds bet condition. So, for example,

506
00:30:44,130 --> 00:30:46,610
Speaker 2: if you do this three-color urn experiment, if you take

507
00:30:46,630 --> 00:30:48,880
Speaker 2: the number of red balls, that's the ones where you

508
00:30:48,920 --> 00:30:50,780
Speaker 2: know how many there are, if you take that down

509
00:30:50,840 --> 00:30:53,380
Speaker 2: to zero or even just down to five or ten,

510
00:30:53,960 --> 00:30:57,400
Speaker 2: Most people will take the gamble with the unknown odds

511
00:30:57,480 --> 00:30:59,880
Speaker 2: on yellow and black because the odds on red are

512
00:30:59,960 --> 00:31:03,160
Speaker 2: clearly very bad. If you make the odds on red

513
00:31:03,260 --> 00:31:06,440
Speaker 2: really good, even more people will stick with the known

514
00:31:06,560 --> 00:31:09,640
Speaker 2: odds and ambiguity aversion becomes even stronger. So that shouldn't

515
00:31:09,660 --> 00:31:13,300
Speaker 2: be all that surprising. But yeah, in the middle category

516
00:31:13,360 --> 00:31:16,280
Speaker 2: where the odds on red are somewhere in between good

517
00:31:16,340 --> 00:31:19,260
Speaker 2: and bad. you're just kind of like, I don't know,

518
00:31:19,320 --> 00:31:22,860
Speaker 2: but still people tend to go for the known odds

519
00:31:22,940 --> 00:31:26,600
Speaker 2: rather than the unknown odds. However, the authors also found

520
00:31:26,700 --> 00:31:32,850
Speaker 2: that even when the idea of subjective expected utility and

521
00:31:32,920 --> 00:31:36,790
Speaker 2: choice consistency was explained to the people doing this experiment,

522
00:31:36,830 --> 00:31:39,830
Speaker 2: explained to the subjects, or even in cases where the

523
00:31:39,890 --> 00:31:45,310
Speaker 2: known odds were bad, some subjects just stubbornly stuck to

524
00:31:45,390 --> 00:31:50,140
Speaker 2: red and avoided betting on any ambiguous conditions so most

525
00:31:50,200 --> 00:31:54,580
Speaker 2: of us seem to dislike ambiguity in in the majority

526
00:31:54,600 --> 00:31:56,640
Speaker 2: of cases and some of us just don't like it

527
00:31:56,880 --> 00:32:09,600
Speaker 2: at all and will not tolerate it now as is

528
00:32:09,660 --> 00:32:12,840
Speaker 2: often the case with uh stuff in decision theory and,

529
00:32:13,060 --> 00:32:16,200
Speaker 2: and economics and stuff, you start off playing these little

530
00:32:16,240 --> 00:32:18,620
Speaker 2: like gambling games with like balls in a jar or

531
00:32:18,700 --> 00:32:20,650
Speaker 2: something like that. And so it seems like, well, how

532
00:32:20,730 --> 00:32:23,870
Speaker 2: consequential could this really be? But actually I think you

533
00:32:23,910 --> 00:32:27,370
Speaker 2: can take the idea of ambiguity aversion to, uh, to

534
00:32:27,790 --> 00:32:31,330
Speaker 2: number one, it has been applied to very consequential situations

535
00:32:31,490 --> 00:32:36,480
Speaker 2: in real life and thus understanding it can have major consequences,

536
00:32:36,560 --> 00:32:40,060
Speaker 2: not only for people's individual lives, but for world events. Yeah.

537
00:32:40,840 --> 00:32:44,420
Speaker 1: That's right. And that brings us back to Ellsberg, who, again,

538
00:32:44,740 --> 00:32:49,990
Speaker 1: in addition to crafting this highly influential paper, was also

539
00:32:51,150 --> 00:32:54,730
Speaker 1: a highly vocal critic of nuclear weapons and, again, the

540
00:32:54,790 --> 00:33:00,090
Speaker 1: rhetoric surrounding nuclear weapons. And I was reading a good

541
00:33:00,130 --> 00:33:02,690
Speaker 1: bit from his book, The Doomsday Machine, which came out

542
00:33:02,730 --> 00:33:05,280
Speaker 1: in 2012, which is quite a good read if you

543
00:33:05,300 --> 00:33:07,440
Speaker 1: want a much deeper dive into his thoughts on all

544
00:33:07,460 --> 00:33:12,900
Speaker 1: of this. But In that book, the subject of ambiguous

545
00:33:12,960 --> 00:33:17,520
Speaker 1: data comes up numerous times. How will a nuclear power

546
00:33:17,600 --> 00:33:22,740
Speaker 1: react in response to ambiguous data about a potential incoming attack?

547
00:33:23,540 --> 00:33:27,600
Speaker 1: And additionally, it talks about the purported advantage of ambiguity

548
00:33:27,740 --> 00:33:33,270
Speaker 1: in intent slash response, namely in the case of madman theory.

549
00:33:34,090 --> 00:33:38,510
Speaker 1: This is very closely associated with Richard Nixon, but also

550
00:33:38,590 --> 00:33:42,970
Speaker 1: tied to various other figures, domestic and international, including the

551
00:33:43,010 --> 00:33:48,210
Speaker 1: current U.S. president, a kind of supposed perceived madness or

552
00:33:48,270 --> 00:33:52,840
Speaker 1: volatility that would lead adversaries to second guess any plans

553
00:33:52,900 --> 00:33:53,800
Speaker 1: to move against them.

554
00:33:54,520 --> 00:33:57,020
Speaker 2: Right. And we're going to get to some major criticisms

555
00:33:57,140 --> 00:33:59,959
Speaker 2: of the madman theory strategy in a bit. But the

556
00:34:00,120 --> 00:34:04,460
Speaker 2: idea of it is that you can leverage a reputation

557
00:34:04,610 --> 00:34:08,950
Speaker 2: for unpredictability to your advantage in negotiations.

558
00:34:09,600 --> 00:34:11,680
Speaker 1: That's right. You know, this is very much the nobody

559
00:34:11,719 --> 00:34:15,250
Speaker 1: knows what I'm going to do approach. That's the message

560
00:34:15,290 --> 00:34:18,090
Speaker 1: you put out there. And the idea, part of the

561
00:34:18,170 --> 00:34:21,410
Speaker 1: idea here is that it makes logically empty or extreme

562
00:34:21,510 --> 00:34:23,530
Speaker 1: threats seem more possible.

563
00:34:23,820 --> 00:34:23,990
Speaker 2: Yeah.

564
00:34:24,360 --> 00:34:25,899
Speaker 1: Now, madman theory, we should point out, is a term

565
00:34:25,940 --> 00:34:29,040
Speaker 1: that Nixon even used himself when talking about this approach.

566
00:34:29,560 --> 00:34:32,040
Speaker 1: It was outlined by Daniel Ellsberg and Thomas C. Schelling

567
00:34:32,100 --> 00:34:36,560
Speaker 1: as well. But the basic concept Pretending to be erratic, irrational,

568
00:34:36,620 --> 00:34:39,560
Speaker 1: or unhinged to influence the other side of a bargaining table,

569
00:34:40,020 --> 00:34:42,989
Speaker 1: this goes back a long ways. Machiavelli wrote about it

570
00:34:43,219 --> 00:34:46,870
Speaker 1: in 1517, and it's probably as old as the first

571
00:34:46,910 --> 00:34:50,210
Speaker 1: prehistoric human to, like, go around hitting trees randomly with

572
00:34:50,250 --> 00:34:52,750
Speaker 1: a club, you know, and say, who knows what Thrag

573
00:34:52,810 --> 00:34:55,270
Speaker 1: is going to do? Why would you mess with Thrag

574
00:34:55,330 --> 00:34:55,609
Speaker 1: on this?

575
00:34:56,390 --> 00:34:58,420
Speaker 2: Yeah, we better just give Thrag what he wants.

576
00:34:58,820 --> 00:35:02,930
Speaker 1: Yeah. So, again, the Ellsberg paradox highlights a core flaw

577
00:35:03,010 --> 00:35:07,070
Speaker 1: in human decision making, the aversion to ambiguity. The madman

578
00:35:07,110 --> 00:35:10,370
Speaker 1: theory as a political strategy seeks to exploit this by

579
00:35:10,469 --> 00:35:15,509
Speaker 1: cultivating a reputation of irrationality, creating an ambiguous threat. The

580
00:35:15,610 --> 00:35:18,730
Speaker 1: adversary and their aversion to this ambiguity and the unknown

581
00:35:18,850 --> 00:35:23,850
Speaker 1: risk of an unpredictable response may be coerced into backing down.

582
00:35:23,890 --> 00:35:26,190
Speaker 1: Or at least this is the idea. This is the

583
00:35:26,290 --> 00:35:27,810
Speaker 1: logic behind the approach.

584
00:35:28,130 --> 00:35:31,489
Speaker 2: Yeah. The classic example of madman theory in practice in

585
00:35:31,550 --> 00:35:36,170
Speaker 2: real life is that the Nixon administration could, for example,

586
00:35:36,190 --> 00:35:40,649
Speaker 2: have Kissinger call up his counterparts in the Kremlin and say, look,

587
00:35:40,890 --> 00:35:42,870
Speaker 2: you know, we're all trying to calm Nixon down, but

588
00:35:42,910 --> 00:35:46,549
Speaker 2: he's boiling mad at you. He's borderline crazy. You've got

589
00:35:46,570 --> 00:35:48,390
Speaker 2: to do X, Y and Z. You know, you've got

590
00:35:48,410 --> 00:35:50,469
Speaker 2: to give him these concessions he wants or we don't

591
00:35:50,530 --> 00:35:52,129
Speaker 2: know if we'll be able to control him.

592
00:35:52,710 --> 00:35:54,669
Speaker 1: Yeah, kind of a good cop, bad cop thing, too,

593
00:35:54,710 --> 00:35:55,760
Speaker 1: I imagine. Yeah.

594
00:35:56,180 --> 00:35:59,540
Speaker 2: And this was a way of making I think a

595
00:35:59,580 --> 00:36:01,140
Speaker 2: big part of it was it was a way of

596
00:36:01,219 --> 00:36:04,989
Speaker 2: making the nuclear deterrent feel once again like it had

597
00:36:05,150 --> 00:36:09,110
Speaker 2: force in matters other than just deterring a first strike.

598
00:36:10,110 --> 00:36:13,570
Speaker 2: Because the problem with using nuclear weapons as leverage in

599
00:36:13,610 --> 00:36:18,410
Speaker 2: negotiation between two nuclear armed powers is obviously if the

600
00:36:18,450 --> 00:36:22,200
Speaker 2: weapons get used at all, everyone loses. There is not

601
00:36:22,239 --> 00:36:25,859
Speaker 2: a winner in mutually assured destruction. So if the White

602
00:36:25,900 --> 00:36:28,590
Speaker 2: House was trying to get the Kremlin to do something

603
00:36:28,650 --> 00:36:31,629
Speaker 2: like I'll nuke you if you don't do it, that

604
00:36:31,670 --> 00:36:35,210
Speaker 2: was not a credible threat because both parties knew that

605
00:36:35,250 --> 00:36:38,450
Speaker 2: the other party knows that war would be the destruction

606
00:36:38,469 --> 00:36:42,590
Speaker 2: of them both. So no rational actor would ever strike first.

607
00:36:43,070 --> 00:36:46,640
Speaker 2: Between rational actors, I'll nuke you if you don't do X, Y,

608
00:36:46,660 --> 00:36:50,220
Speaker 2: Z is an empty threat. The goal of madman theory

609
00:36:50,239 --> 00:36:54,319
Speaker 2: is is to gain leverage in negotiations by introducing some

610
00:36:54,360 --> 00:36:58,120
Speaker 2: amount of fear that your counterparty may not be a

611
00:36:58,160 --> 00:37:01,060
Speaker 2: rational actor. He might be crazy enough to do it.

612
00:37:01,160 --> 00:37:05,779
Speaker 2: We don't know. Therefore, the madman strategy makes direct use

613
00:37:05,980 --> 00:37:09,560
Speaker 2: of the opponent's ambiguity aversion as a tactic to get

614
00:37:09,660 --> 00:37:13,080
Speaker 2: leverage over them. The madman player in this game is

615
00:37:13,239 --> 00:37:17,120
Speaker 2: betting that the fear and uncertainty about how to handle

616
00:37:17,180 --> 00:37:21,920
Speaker 2: their unpredictability will cause their counterparty to make concessions they

617
00:37:21,960 --> 00:37:24,900
Speaker 2: might not make otherwise. Like, you end up thinking, well,

618
00:37:25,360 --> 00:37:27,839
Speaker 2: I'll just take a bad deal rather than wander into

619
00:37:27,900 --> 00:37:31,860
Speaker 2: territory with unclear levels of risk. Right, right.

620
00:37:31,969 --> 00:37:34,390
Speaker 1: And not to derail the point here, but I want

621
00:37:34,410 --> 00:37:38,450
Speaker 1: to bring it back to Ellsberg's other criticisms of nuclear

622
00:37:38,469 --> 00:37:41,430
Speaker 1: weapons and the rhetoric surrounding nuclear weapons, that being there's

623
00:37:41,450 --> 00:37:45,569
Speaker 1: all these other uncertainty factors and ambiguous data that would

624
00:37:45,610 --> 00:37:47,719
Speaker 1: be coming in about what the enemy is or is

625
00:37:47,780 --> 00:37:51,719
Speaker 1: not doing. And therefore, any given situation like this is

626
00:37:51,760 --> 00:37:55,580
Speaker 1: going to be just wrought with with potential ambiguity, potential

627
00:37:55,660 --> 00:37:58,820
Speaker 1: missteps and potential mistakes. So, like, if it were only

628
00:37:58,960 --> 00:38:01,339
Speaker 1: as easy as we're laying it out here, it would

629
00:38:01,360 --> 00:38:02,580
Speaker 1: almost be a different situation.

630
00:38:03,060 --> 00:38:06,860
Speaker 2: Totally. I mean, Ellsberg's point about the all of the

631
00:38:07,000 --> 00:38:11,419
Speaker 2: uncertainty in the actual command and control of, you know,

632
00:38:11,480 --> 00:38:16,790
Speaker 2: nuclear strategy like that's true. That introduces real levels of

633
00:38:16,950 --> 00:38:21,270
Speaker 2: existential risk into playing games like this. There are a

634
00:38:21,330 --> 00:38:24,330
Speaker 2: lot of reasonable criticisms of madman theory, but I think

635
00:38:24,370 --> 00:38:26,430
Speaker 2: a major one, at least from my point of view,

636
00:38:26,489 --> 00:38:30,650
Speaker 2: is that at best, it is a strategy for short

637
00:38:30,830 --> 00:38:34,710
Speaker 2: term gain at the expense of long term stability and

638
00:38:34,750 --> 00:38:39,339
Speaker 2: negotiating power, because one way or another. It undercuts your

639
00:38:39,400 --> 00:38:42,730
Speaker 2: credibility and your ability to be seen as an honest

640
00:38:42,790 --> 00:38:47,250
Speaker 2: and reliable broker in future negotiations, either because you come

641
00:38:47,350 --> 00:38:50,810
Speaker 2: to be seen as actually irrational, in which case people

642
00:38:50,850 --> 00:38:53,340
Speaker 2: don't want to deal with you and they will find

643
00:38:53,380 --> 00:38:57,400
Speaker 2: what they may be motivated to do something bad to

644
00:38:57,460 --> 00:38:59,920
Speaker 2: strike first to you because you're too dangerous to be

645
00:38:59,960 --> 00:39:03,680
Speaker 2: let loose, or they may try to find ways to

646
00:39:03,739 --> 00:39:07,020
Speaker 2: cut you out of negotiations. So that could be one consequence,

647
00:39:07,200 --> 00:39:09,840
Speaker 2: or just because you are revealed to have been bluffing,

648
00:39:09,880 --> 00:39:11,569
Speaker 2: in which case you lose your credibility.

649
00:39:12,120 --> 00:39:15,370
Speaker 1: Right, right. And yeah, this is all valid. I was

650
00:39:15,390 --> 00:39:18,190
Speaker 1: looking at a couple of papers on madman theory. And

651
00:39:18,510 --> 00:39:22,170
Speaker 1: madman theory, like ambiguity aversion itself, is something that has

652
00:39:22,510 --> 00:39:25,250
Speaker 1: been written about a lot. There are no shortage of

653
00:39:25,290 --> 00:39:27,890
Speaker 1: papers out there. But I looked at just a couple.

654
00:39:28,590 --> 00:39:31,290
Speaker 1: I looked at one titled Cravings. Like a Fox, Are

655
00:39:31,330 --> 00:39:34,930
Speaker 1: Leaders with Reputations for Madness More Successful at International Coercion?

656
00:39:35,370 --> 00:39:39,370
Speaker 1: by Roseanne W. McManus. This came out in 2019 from

657
00:39:39,469 --> 00:39:42,670
Speaker 1: Cambridge University Press. And in this, the author points out

658
00:39:43,050 --> 00:39:45,520
Speaker 1: that a reputation for madness is would seem to be

659
00:39:46,520 --> 00:39:50,910
Speaker 1: more often harmful than helpful in international coercion. So it

660
00:39:51,030 --> 00:39:56,730
Speaker 1: undercuts the leader's ability to make believable peace commitments, treaties,

661
00:39:56,850 --> 00:39:59,890
Speaker 1: and so forth, which, and it really almost feels absurd

662
00:39:59,910 --> 00:40:02,450
Speaker 1: that we need to stress this, these are all vital

663
00:40:02,630 --> 00:40:06,190
Speaker 1: for peaceful relations between nations. No one knows what I'll

664
00:40:06,230 --> 00:40:09,180
Speaker 1: do easily slides into no one knows what I'll honor.

665
00:40:09,360 --> 00:40:13,440
Speaker 1: No one knows what treaties I will actually stand by

666
00:40:13,500 --> 00:40:16,259
Speaker 1: and so forth. You know, to bring this back to

667
00:40:16,300 --> 00:40:18,160
Speaker 1: Dungeons and Dragons, I think any D & D player

668
00:40:18,180 --> 00:40:20,239
Speaker 1: worth their salt knows that while you might well enter

669
00:40:20,280 --> 00:40:23,760
Speaker 1: into a pact with a lawful evil devil, you never

670
00:40:23,820 --> 00:40:27,090
Speaker 1: enter into a pact with a chaotic evil demon because

671
00:40:27,120 --> 00:40:28,730
Speaker 1: the lords of the nine hell are going to stick

672
00:40:28,750 --> 00:40:30,810
Speaker 1: to their letter of the contract, if not the spirit

673
00:40:30,830 --> 00:40:34,250
Speaker 1: of the contract. But demon lords such as the Demogorgon

674
00:40:34,469 --> 00:40:38,469
Speaker 1: will honor nothing. And there is no coherence even within themselves.

675
00:40:38,510 --> 00:40:41,500
Speaker 1: They're just pure ambiguity. And you can't strike any sort

676
00:40:41,540 --> 00:40:43,920
Speaker 1: of deal because they won't they won't stand by it

677
00:40:43,960 --> 00:40:47,489
Speaker 1: no matter what. Whereas the devils being lawful evil, they

678
00:40:47,550 --> 00:40:50,709
Speaker 1: may look for that wiggle room. They may find ways to,

679
00:40:51,590 --> 00:40:54,029
Speaker 1: you know, to avoid the spirit of the deal, but

680
00:40:54,050 --> 00:40:55,300
Speaker 1: they're still bound to the letter.

681
00:40:55,700 --> 00:40:58,620
Speaker 2: Right. So if you think you discover that your counterparty

682
00:40:58,660 --> 00:41:01,239
Speaker 2: is actually just chaotic evil, all you can do is

683
00:41:01,260 --> 00:41:03,820
Speaker 2: roll for initiative. Like there's no making a deal.

684
00:41:04,040 --> 00:41:08,740
Speaker 1: Yeah. And then you slide into chaos, pure chaos. McManus,

685
00:41:09,160 --> 00:41:10,960
Speaker 1: back to her paper, though, she found that Mad Men

686
00:41:11,000 --> 00:41:14,919
Speaker 1: theory may be helpful in crisis bargaining, but only under

687
00:41:15,040 --> 00:41:20,400
Speaker 1: certain conditions, namely when employed by military weak leaders. So

688
00:41:20,420 --> 00:41:23,319
Speaker 1: we're not talking about a true superpower here that can

689
00:41:23,360 --> 00:41:27,640
Speaker 1: engage in like true mutually assured destruction. Rather, we would

690
00:41:27,700 --> 00:41:30,270
Speaker 1: be dealing with a state that could do a lot

691
00:41:30,310 --> 00:41:35,170
Speaker 1: of damage, but in striking out would just thoroughly destroy themselves.

692
00:41:35,739 --> 00:41:37,939
Speaker 1: So the idea here is that no one would take

693
00:41:38,739 --> 00:41:42,130
Speaker 1: this hypothetical nation or the leader of this hypothetical nation

694
00:41:42,310 --> 00:41:47,910
Speaker 1: seriously unless they presented an air of madness. And then perhaps, hey,

695
00:41:47,969 --> 00:41:51,969
Speaker 1: they might do it anyway if provoked. They're unhinged. And thus,

696
00:41:52,050 --> 00:41:56,350
Speaker 1: it can be a form of asymmetric leverage. So we're

697
00:41:56,370 --> 00:42:01,189
Speaker 1: talking low probability but high consequence. Yeah. And McManus also

698
00:42:01,230 --> 00:42:02,950
Speaker 1: points out that there would also seem to be more

699
00:42:02,969 --> 00:42:06,060
Speaker 1: of an advantage if you had a mild reputation for

700
00:42:06,120 --> 00:42:08,500
Speaker 1: madness rather than an extreme one. You don't want to

701
00:42:08,560 --> 00:42:12,719
Speaker 1: come off like, you know, a complete Dada, you know,

702
00:42:12,840 --> 00:42:16,529
Speaker 1: eye rolling maniac in one of these cases. The idea

703
00:42:16,550 --> 00:42:18,450
Speaker 1: would be like, well, you know, we can we can

704
00:42:18,489 --> 00:42:21,989
Speaker 1: still get this individual to the negotiation table, but we

705
00:42:22,010 --> 00:42:25,700
Speaker 1: just have to be hypersensitive. And I guess, again, kind

706
00:42:25,739 --> 00:42:30,060
Speaker 1: of avoiding that complete fall off into chaotic evil demon

707
00:42:30,100 --> 00:42:32,860
Speaker 1: hood and instead dealing with devils that can still be

708
00:42:32,900 --> 00:42:35,930
Speaker 1: bound by some sort of law. All right. I have

709
00:42:35,950 --> 00:42:40,650
Speaker 1: a hypothetical example here. So we have demigodic hero Hercules

710
00:42:41,710 --> 00:42:46,010
Speaker 1: and we have his mortal cousin Eurystheus. And they're grabbing

711
00:42:46,050 --> 00:42:47,510
Speaker 1: lunch at the local Euro place.

712
00:42:47,810 --> 00:42:48,130
Speaker 2: Great.

713
00:42:49,010 --> 00:42:52,270
Speaker 1: So. So Eurystheus warns his cousin that if he tries

714
00:42:52,310 --> 00:42:54,469
Speaker 1: to steal any of his fries, he's going to flip

715
00:42:54,489 --> 00:42:56,029
Speaker 1: the table and all their food's going to wind up

716
00:42:56,070 --> 00:42:58,680
Speaker 1: on the floor. And he acts just really sensitive and

717
00:42:58,739 --> 00:43:01,180
Speaker 1: unhinged about the whole thing. Yes.

718
00:43:01,380 --> 00:43:01,700
Speaker 2: Okay.

719
00:43:02,080 --> 00:43:06,320
Speaker 1: Hercules obviously knows that Eurystheus can't take him in a

720
00:43:06,360 --> 00:43:08,460
Speaker 1: fight and might not even be able to flip that

721
00:43:08,540 --> 00:43:14,879
Speaker 1: table over before Hercules stops him. And rationally, Eurystheus should

722
00:43:14,920 --> 00:43:17,430
Speaker 1: realize this and just let Big Herc have a few

723
00:43:17,489 --> 00:43:19,390
Speaker 1: extra fries if he wants. Like, what does it matter?

724
00:43:20,170 --> 00:43:23,930
Speaker 1: But from Hercules' standpoint, does he really want to risk

725
00:43:24,050 --> 00:43:26,859
Speaker 1: his lunch winding up on the floor? Maybe he should

726
00:43:26,880 --> 00:43:29,560
Speaker 1: just let Eurystheus keep all of his fries.

727
00:43:29,880 --> 00:43:30,260
Speaker 2: Okay.

728
00:43:31,180 --> 00:43:35,859
Speaker 1: The downside, of course, here is that maybe Hercules just

729
00:43:35,940 --> 00:43:39,279
Speaker 1: won't invite Eurystheus out for lunch next time. And he's

730
00:43:39,320 --> 00:43:41,990
Speaker 1: certainly not going to pick up the check. And on

731
00:43:42,030 --> 00:43:44,950
Speaker 1: top of all that, how are they going to work

732
00:43:44,989 --> 00:43:48,009
Speaker 1: together to defeat the Hydra? Because as you'll remember from

733
00:43:48,070 --> 00:43:51,029
Speaker 1: past episodes and from Greek mythology in general, Hercules can't

734
00:43:51,090 --> 00:43:52,890
Speaker 1: do that on his own. He has to have his

735
00:43:52,930 --> 00:43:55,610
Speaker 1: cousin's help to burn the stumps after each head of

736
00:43:55,630 --> 00:43:56,509
Speaker 1: the Hydra is cut off.

737
00:43:56,590 --> 00:44:01,410
Speaker 2: Very good points. I mean, apart from any otherwise moral

738
00:44:01,469 --> 00:44:05,760
Speaker 2: considerations or honesty considerations about deploying something like madman theory,

739
00:44:05,800 --> 00:44:09,960
Speaker 2: just from a strategic point of view, it seems like

740
00:44:10,160 --> 00:44:14,090
Speaker 2: it is probably best for situations where Like your counterparty

741
00:44:14,410 --> 00:44:17,589
Speaker 2: has to deal with you. It is not optional for them.

742
00:44:18,370 --> 00:44:21,850
Speaker 2: And you don't have to like or respect each other

743
00:44:22,170 --> 00:44:26,589
Speaker 2: or ever work together on anything. And you don't care

744
00:44:26,630 --> 00:44:30,660
Speaker 2: about long-term goals. You are only interested in this situation

745
00:44:30,780 --> 00:44:33,560
Speaker 2: right now in extracting a short-term advantage.

746
00:44:34,440 --> 00:44:37,839
Speaker 1: That's right. That's right. But some papers out there that

747
00:44:37,860 --> 00:44:41,560
Speaker 1: have crunched the numbers on all this and done some experimentation,

748
00:44:41,620 --> 00:44:44,640
Speaker 1: questionnaires and so forth, Do acknowledge that, you know, there

749
00:44:44,680 --> 00:44:48,779
Speaker 1: are multiple dimensions to any of these situations, namely, and

750
00:44:48,800 --> 00:44:50,640
Speaker 1: this is something I think everyone can relate to, like

751
00:44:50,660 --> 00:44:54,080
Speaker 1: you can have a leader that is engaging in an

752
00:44:54,180 --> 00:44:59,759
Speaker 1: international situation, but there is still the domestic view of

753
00:44:59,800 --> 00:45:03,380
Speaker 1: that situation. There's still the domestic response, the domestic relationship.

754
00:45:04,700 --> 00:45:06,640
Speaker 1: And so one of the papers that I was looking at,

755
00:45:06,800 --> 00:45:09,880
Speaker 1: this one is titled Madman or Mad Genius, the International

756
00:45:09,940 --> 00:45:13,800
Speaker 1: Benefits and Domestic Costs of Madman Strategy by Joshua A. Schwartz,

757
00:45:13,860 --> 00:45:17,900
Speaker 1: published in 2023 in Security Studies. And in this, the

758
00:45:18,020 --> 00:45:22,160
Speaker 1: author found that the madman approach can work in negotiations

759
00:45:22,300 --> 00:45:26,509
Speaker 1: with foreign adversaries, but, quote, entails significant domestic costs that

760
00:45:26,530 --> 00:45:30,510
Speaker 1: potentially erode its efficacy. He also points out that madman

761
00:45:30,530 --> 00:45:33,850
Speaker 1: theory may simply not work against major powers because while

762
00:45:33,910 --> 00:45:37,540
Speaker 1: it might, make a threat more credible that doesn't necessarily

763
00:45:37,580 --> 00:45:41,320
Speaker 1: make the threat more effective and make the adversary willing

764
00:45:41,360 --> 00:45:44,379
Speaker 1: to cave to those demands. So he points out that

765
00:45:44,560 --> 00:45:47,080
Speaker 1: this is perhaps why Nixon's use of madman theory didn't

766
00:45:47,120 --> 00:45:50,280
Speaker 1: work against the Soviets. Like, okay, it made Nixon's threat

767
00:45:50,440 --> 00:45:55,380
Speaker 1: more credible, but did it actually make it more effective?

768
00:45:55,400 --> 00:45:57,360
Speaker 1: Did he actually get what he wanted to out of

769
00:45:57,660 --> 00:46:01,300
Speaker 1: these bluffs, these threats? And then you have the domestic

770
00:46:01,340 --> 00:46:06,320
Speaker 1: side of things. With the hypothetical leader's own citizens not

771
00:46:06,510 --> 00:46:11,469
Speaker 1: loving the heightened stakes because, you know, obviously, I mean,

772
00:46:11,489 --> 00:46:12,969
Speaker 1: I say obviously, but this is one of those things

773
00:46:12,989 --> 00:46:14,750
Speaker 1: that we often have to be reminded of in a

774
00:46:14,770 --> 00:46:17,770
Speaker 1: case of mutually assured destruction or even a case of

775
00:46:17,930 --> 00:46:22,100
Speaker 1: asymmetric exchange involving nuclear weapons or some other kind of just,

776
00:46:22,340 --> 00:46:25,520
Speaker 1: you know, horrible weapon of mass destruction. A leader is

777
00:46:25,739 --> 00:46:28,980
Speaker 1: always bargaining with the lives of their own citizens. You

778
00:46:29,060 --> 00:46:31,000
Speaker 1: are the chips on the board in a no one

779
00:46:31,020 --> 00:46:34,229
Speaker 1: knows what I'll do wager. Yeah. And the author also

780
00:46:34,310 --> 00:46:36,550
Speaker 1: notes that, OK, you know, this is going to also

781
00:46:36,610 --> 00:46:38,810
Speaker 1: differ depending on how much of a voice the people

782
00:46:38,870 --> 00:46:42,080
Speaker 1: have in a given nation versus how much power the

783
00:46:42,120 --> 00:46:45,940
Speaker 1: ruler has. There's going to be less backlash in places

784
00:46:46,020 --> 00:46:47,819
Speaker 1: where the people have less of a voice and the

785
00:46:47,860 --> 00:46:49,380
Speaker 1: ruler has more absolute power.

786
00:46:49,640 --> 00:46:51,990
Speaker 2: That's right. But one thing I did want to clarify

787
00:46:52,010 --> 00:46:55,229
Speaker 2: is that the sense in which you are the chips

788
00:46:55,310 --> 00:46:57,410
Speaker 2: on the board and no one knows what I'll do

789
00:46:57,489 --> 00:47:01,759
Speaker 2: wager is. is true is not just limited to like

790
00:47:02,380 --> 00:47:05,299
Speaker 2: the worst possible scenario, like nuclear warfare. I mean, it's

791
00:47:05,420 --> 00:47:08,890
Speaker 2: also the case in say like trade negotiations or something

792
00:47:08,930 --> 00:47:11,850
Speaker 2: like that. Like your economic prospects are in a way,

793
00:47:12,330 --> 00:47:14,850
Speaker 2: those are the chips on the board. Like the, the,

794
00:47:14,930 --> 00:47:18,150
Speaker 2: the citizens fates and futures are the things that are

795
00:47:18,190 --> 00:47:20,190
Speaker 2: being negotiated with. That's right.

796
00:47:20,210 --> 00:47:24,810
Speaker 1: Yeah. In, in, in contemplating potential nuclear exchanges, everything is

797
00:47:24,850 --> 00:47:27,600
Speaker 1: a bit more stark and a bit more black and white,

798
00:47:27,700 --> 00:47:30,600
Speaker 1: but obviously, you know, any, You can think of any

799
00:47:30,660 --> 00:47:33,879
Speaker 1: number of scenarios where the stakes are still quite high

800
00:47:34,400 --> 00:47:35,160
Speaker 1: for the individual.

801
00:47:35,440 --> 00:47:37,359
Speaker 2: So, yes, it's not hard to see at all why

802
00:47:37,380 --> 00:47:41,950
Speaker 2: a citizenry can easily become upset and annoyed if their leader,

803
00:47:41,989 --> 00:47:44,830
Speaker 2: who's supposed to be representing their interests in negotiations, is

804
00:47:45,030 --> 00:47:49,850
Speaker 2: acting unreliable and unpredictable. The leader may think that they

805
00:47:49,870 --> 00:47:52,250
Speaker 2: can get good gains out of that in the short term,

806
00:47:52,290 --> 00:47:55,160
Speaker 2: but the citizens are probably thinking about a lot of

807
00:47:55,200 --> 00:47:58,239
Speaker 2: these Even if not, you know, thinking specifically about them,

808
00:47:58,280 --> 00:48:00,859
Speaker 2: just thinking intuitively about there seems like there's a lot

809
00:48:00,900 --> 00:48:02,980
Speaker 2: of downside to this. Yeah.

810
00:48:03,120 --> 00:48:06,880
Speaker 1: And again, coming back to what we mentioned earlier, the idea, too,

811
00:48:07,020 --> 00:48:10,700
Speaker 1: that by constant, if you swing around this, the sword

812
00:48:10,719 --> 00:48:13,239
Speaker 1: of madman theory too much, then it makes it more

813
00:48:13,300 --> 00:48:19,410
Speaker 1: difficult to engage in the other highly important tools of statecraft,

814
00:48:19,790 --> 00:48:23,190
Speaker 1: peace treaties and agreements and so forth. So it's like

815
00:48:23,210 --> 00:48:25,210
Speaker 1: you can only swing that sword around for so long.

816
00:48:25,810 --> 00:48:28,130
Speaker 1: And then you have to be able to engage in

817
00:48:28,170 --> 00:48:33,140
Speaker 1: these other acts as well. Again, short-term gain with madman theory,

818
00:48:34,260 --> 00:48:38,380
Speaker 1: not so in the long term. But, you know, unfortunately,

819
00:48:38,420 --> 00:48:41,399
Speaker 1: this ends up rolling out into so many things about

820
00:48:42,320 --> 00:48:45,460
Speaker 1: human experience and human perception is that we are so

821
00:48:45,520 --> 00:48:47,640
Speaker 1: focused on the short term and we don't think about

822
00:48:47,680 --> 00:48:48,219
Speaker 1: the long term.

823
00:48:48,739 --> 00:48:50,420
Speaker 2: So we've just been talking about a lot of reasons

824
00:48:50,500 --> 00:48:53,700
Speaker 2: that it might actually be bad and not as clever

825
00:48:53,760 --> 00:48:57,130
Speaker 2: as it first seems. to try to leverage knowledge of

826
00:48:57,230 --> 00:49:02,450
Speaker 2: ambiguity aversion offensively in negotiations. But I think one way

827
00:49:02,510 --> 00:49:05,810
Speaker 2: that it can definitely be useful is to have knowledge

828
00:49:05,950 --> 00:49:09,509
Speaker 2: of ambiguity aversion to use defensively, to think about in

829
00:49:09,690 --> 00:49:15,390
Speaker 2: analyzing your own behavior and being aware of your bias

830
00:49:15,469 --> 00:49:19,129
Speaker 2: to avoid ambiguity and sort of check yourself and think like,

831
00:49:19,190 --> 00:49:22,110
Speaker 2: wait a minute, am I actually making the right decision here?

832
00:49:22,630 --> 00:49:25,700
Speaker 2: Is this actually what's rational or am I just having

833
00:49:25,719 --> 00:49:30,259
Speaker 2: a an irrational bias against situations with unknown probabilities that

834
00:49:30,360 --> 00:49:34,300
Speaker 2: are seeming having an outsized scariness in my mind because

835
00:49:34,320 --> 00:49:37,660
Speaker 2: of the ambiguity involved? Like the example we talked about

836
00:49:37,700 --> 00:49:40,580
Speaker 2: earlier with people being afraid to try new experiences because

837
00:49:40,620 --> 00:49:41,980
Speaker 2: there's some amount of ambiguity.

838
00:49:42,000 --> 00:49:45,819
Speaker 1: Yeah, absolutely. Again, there's always ambiguity in any new experience.

839
00:49:46,360 --> 00:49:48,080
Speaker 1: There's so many ways it could go wrong, but there

840
00:49:48,100 --> 00:49:50,719
Speaker 1: are so many ways it could go right. A lot

841
00:49:50,760 --> 00:49:52,839
Speaker 1: of us do tend to focus more on the ways

842
00:49:52,930 --> 00:49:53,510
Speaker 1: it could go wrong.

843
00:49:53,860 --> 00:49:56,090
Speaker 2: The devil you know is not necessarily better.

844
00:49:56,110 --> 00:49:59,770
Speaker 1: Yeah, the devil you don't know could be really fun.

845
00:50:00,330 --> 00:50:01,490
Speaker 1: Give him a shot sometimes.

846
00:50:02,750 --> 00:50:05,250
Speaker 2: The devil you know could be like a chaotic good devil.

847
00:50:05,570 --> 00:50:07,770
Speaker 2: Those are those exist, right? Are there some?

848
00:50:09,570 --> 00:50:12,090
Speaker 1: You know, maybe under the new, it's possible under the

849
00:50:12,130 --> 00:50:12,870
Speaker 1: new rules, but.

850
00:50:12,890 --> 00:50:16,430
Speaker 2: Maybe it's actually just a friendly tiefling that you give.

851
00:50:16,570 --> 00:50:17,050
Speaker 1: There you go.

852
00:50:17,100 --> 00:50:17,620
Speaker 2: There you go.

853
00:50:17,700 --> 00:50:19,020
Speaker 1: You could have a chaotic good tiefling.

854
00:50:19,040 --> 00:50:19,960
Speaker 2: There you go.

855
00:50:20,040 --> 00:50:21,620
Speaker 1: It's entirely possible. All right.

856
00:50:21,660 --> 00:50:23,759
Speaker 2: Well, that's going to do it for part one of

857
00:50:23,960 --> 00:50:27,140
Speaker 2: our look at ambiguity aversion, but we're going to be

858
00:50:27,180 --> 00:50:30,300
Speaker 2: back again next time to talk about how this applies

859
00:50:30,340 --> 00:50:34,060
Speaker 2: to some other domains of life and maybe some subsequent research.

860
00:50:34,080 --> 00:50:37,480
Speaker 2: I might get some into the different ways that the

861
00:50:37,600 --> 00:50:42,700
Speaker 2: ambiguity aversion observations have been interpreted in terms of different

862
00:50:42,739 --> 00:50:45,480
Speaker 2: types of decision theory, like is it an error or

863
00:50:45,540 --> 00:50:48,299
Speaker 2: is it actually a type of rationality that needs to

864
00:50:48,340 --> 00:50:51,350
Speaker 2: be better understood? So yeah, we'll talk about that kind

865
00:50:51,390 --> 00:50:52,069
Speaker 2: of stuff next time.

866
00:50:52,430 --> 00:50:55,820
Speaker 1: Who knows what angles we'll actually discuss in the next episode?

867
00:50:55,920 --> 00:50:58,280
Speaker 1: You can tune in to find out. In the meantime,

868
00:50:58,739 --> 00:51:00,859
Speaker 1: certainly write in. We'd love to hear from you. And

869
00:51:00,920 --> 00:51:02,980
Speaker 1: we'd like to remind you that Stuff to Blow Your

870
00:51:03,030 --> 00:51:05,430
Speaker 1: Mind is primarily a science and culture podcast with core

871
00:51:05,450 --> 00:51:09,370
Speaker 1: episodes on Tuesdays and Thursdays, short-form episode on Wednesdays, and

872
00:51:09,410 --> 00:51:11,489
Speaker 1: on Fridays we set aside most serious concerns to just

873
00:51:11,510 --> 00:51:14,090
Speaker 1: talk about a weird film on Weird House Cinema.

874
00:51:14,640 --> 00:51:18,000
Speaker 2: Huge thanks, as always, to our excellent audio producer, J.J. Posway.

875
00:51:18,480 --> 00:51:19,980
Speaker 2: If you would like to get in touch with us

876
00:51:20,020 --> 00:51:22,460
Speaker 2: with feedback on this episode or any other, to suggest

877
00:51:22,520 --> 00:51:24,400
Speaker 2: a topic for the future, or just to say hello,

878
00:51:24,480 --> 00:51:36,029
Speaker 2: you can email us at contact at stufftoblowyourmind.com. Stuff to

879
00:51:36,050 --> 00:51:38,790
Speaker 2: Blow Your Mind is a production of iHeartRadio. For more

880
00:51:38,810 --> 00:51:42,620
Speaker 2: podcasts from iHeartRadio, visit the iHeartRadio app. Apple Podcasts are

881
00:51:42,640 --> 00:51:44,280
Speaker 2: wherever you listen to your favorite shows.

882
00:51:51,320 --> 00:51:51,520
Speaker 1: Thank you.