1
00:00:00,600 --> 00:00:02,700
Welcome to Keith's night. 
Don't tread on anyone and the 

2
00:00:02,700 --> 00:00:07,500
libertarian Institute here are 
14 logical fallacies by Michael 

3
00:00:07,500 --> 00:00:10,100
humor of the University of 
Boulder. 

4
00:00:10,200 --> 00:00:14,100
Colorado from his book, 
knowledge, reality and value, a 

5
00:00:14,100 --> 00:00:17,500
mostly Common Sense guide to 
philosophy. 

6
00:00:17,900 --> 00:00:21,900
Number one, anecdotal evidence 
often people try to support 

7
00:00:21,900 --> 00:00:26,400
generalizations by citing a 
single case or a few cases that 

8
00:00:26,400 --> 00:00:30,200
support the generalization 
scientists call this anecdotal 

9
00:00:30,400 --> 00:00:33,600
Evidence example. 
You try to show that immigrants 

10
00:00:33,600 --> 00:00:36,600
are dangerous by citing a few 
examples of immigrants. 

11
00:00:36,600 --> 00:00:40,000
Who committed crimes, anecdotal 
evidence has two problems, 

12
00:00:40,300 --> 00:00:42,100
first. 
Usually, when people do this, 

13
00:00:42,400 --> 00:00:45,500
they don't pick a case randomly.
They search for a case that 

14
00:00:45,500 --> 00:00:49,500
supports their conclusion, while
ignoring cases that don't see 

15
00:00:49,600 --> 00:00:54,200
cherry-picking, second random 
variation, even if you pick 

16
00:00:54,200 --> 00:00:57,700
these cases randomly, it can 
easily happen, just by chance 

17
00:00:57,900 --> 00:01:00,000
that you picked a few atypical 
cases. 

18
00:01:00,300 --> 00:01:03,200
In the immigration example, what
you should actually do is look 

19
00:01:03,200 --> 00:01:07,000
up these statistics on crime 
rates for immigrants compared 

20
00:01:07,000 --> 00:01:12,300
with native-born citizens. 
Number two assumptions. 

21
00:01:12,800 --> 00:01:17,400
One of the major ways we go 
wrong is that we simply assume 

22
00:01:17,400 --> 00:01:20,600
things that we don't know. 
Unfortunately, when you assume 

23
00:01:20,600 --> 00:01:24,700
things, you go wrong a lot more 
often than you expect. 

24
00:01:24,700 --> 00:01:28,500
You should assume that most of 
your assumptions are wrong. 

25
00:01:28,700 --> 00:01:32,000
It is hard to combat this 
because we often don't notice 

26
00:01:32,500 --> 00:01:35,400
what were assuming and it 
doesn't even occur to us. 

27
00:01:35,400 --> 00:01:37,900
To question it. 
Here are a couple examples 

28
00:01:38,200 --> 00:01:42,800
suppose you here statistic about
Out how common intimate partner 

29
00:01:42,800 --> 00:01:44,500
violence is in the United 
States. 

30
00:01:44,500 --> 00:01:46,900
This is where someone physically
abuses their girlfriend 

31
00:01:46,900 --> 00:01:49,900
boyfriend or spouse you 
naturally, assume the vast 

32
00:01:49,900 --> 00:01:54,000
majority of these cases are men 
beating up women and you might 

33
00:01:54,200 --> 00:01:58,300
just go on reasoning from that 
implicit assumption in reality. 

34
00:01:58,300 --> 00:02:01,400
Though, survey evidence suggests
that men and women suffer, this 

35
00:02:01,400 --> 00:02:05,700
kind of abuse about equally 
often or suppose. 

36
00:02:05,700 --> 00:02:08,900
You are a statistic stating that
most murder victims are killed 

37
00:02:08,900 --> 00:02:12,200
by a family member. 
Or someone they know you 

38
00:02:12,200 --> 00:02:16,300
naturally assumed that most 
murders result from domestic, 

39
00:02:16,300 --> 00:02:19,200
disagreements. 
And that the murders are 

40
00:02:19,200 --> 00:02:22,600
committed by ordinary people who
lost control during an argument 

41
00:02:22,600 --> 00:02:24,800
with a family member or 
something like that. 

42
00:02:24,800 --> 00:02:28,400
In fact, it turns out that 
almost everyone who commits a 

43
00:02:28,408 --> 00:02:31,300
murder has a prior criminal 
record. 

44
00:02:31,400 --> 00:02:34,900
Also the vast majority of these 
victims are also criminals the 

45
00:02:34,900 --> 00:02:38,400
category, a family member, or 
someone you knew include such 

46
00:02:38,400 --> 00:02:42,100
people as the His drug dealer, 
the victims criminal partner, 

47
00:02:42,400 --> 00:02:44,800
the victims fellow gang member 
and so on. 

48
00:02:45,200 --> 00:02:48,900
You just assumed that these were
Ordinary People but the original

49
00:02:48,900 --> 00:02:53,800
statistic didn't say that I 
can't really properly convey to 

50
00:02:53,800 --> 00:02:56,200
you. 
Just how often assuming things 

51
00:02:56,200 --> 00:02:58,700
leads, you astray. 
You need to experience being 

52
00:02:58,700 --> 00:03:03,100
wrong over and over again in 
order to appreciate the point. 

53
00:03:03,300 --> 00:03:06,900
Unfortunately, most people never
come to appreciate the point 

54
00:03:06,900 --> 00:03:10,400
because they never check on 
their assumptions to About how 

55
00:03:10,400 --> 00:03:15,200
many are wrong? 
Number three, base rate. 

56
00:03:15,200 --> 00:03:17,600
Neglect. 
A base rate is the frequency 

57
00:03:17,600 --> 00:03:20,400
with which some type of 
phenomenon happens in general. 

58
00:03:20,800 --> 00:03:23,000
EG. 
The base rate for heart disease,

59
00:03:23,100 --> 00:03:26,400
is the percentage of people in 
the general population who have 

60
00:03:26,400 --> 00:03:29,000
heart disease. 
The base rate for war is the 

61
00:03:29,008 --> 00:03:32,100
percentage of the time that a 
country is at War Etc. 

62
00:03:32,600 --> 00:03:35,300
When you want to know whether 
some kind of event is going to 

63
00:03:35,300 --> 00:03:39,600
happen or has happened Etc. 
The best place to start is with 

64
00:03:39,800 --> 00:03:42,100
the base rate. 
If you want to know whether you 

65
00:03:42,100 --> 00:03:45,300
have a certain disease, First 
find out how common the diseases

66
00:03:45,300 --> 00:03:47,200
in general. 
If one percent of the 

67
00:03:47,200 --> 00:03:50,400
population, has it been a good 
initial estimate? 

68
00:03:50,500 --> 00:03:54,100
Is that you have a one percent 
chance of having it from there. 

69
00:03:54,100 --> 00:03:56,200
You should adjust that estimate 
up or down. 

70
00:03:56,400 --> 00:04:00,100
According to any special risk 
factors or low risk. 

71
00:04:00,100 --> 00:04:03,200
Factors that you have. 
Most people don't do this. 

72
00:04:03,500 --> 00:04:07,000
People commonly ignore base 
rates example. 

73
00:04:07,000 --> 00:04:09,000
Suppose. 
There is a rare disease, that 

74
00:04:09,000 --> 00:04:11,500
afflicts one in a million 
people. 

75
00:04:11,700 --> 00:04:15,900
There's a test for this disease.
And that's 90% accurate. 

76
00:04:16,100 --> 00:04:19,100
Suppose you took the test and 
you test positive. 

77
00:04:19,100 --> 00:04:22,900
The test says, you have the 
disease disease question, given 

78
00:04:22,900 --> 00:04:25,600
all this information. 
What is the probability that you

79
00:04:25,600 --> 00:04:26,900
have? 
The disease? 

80
00:04:27,500 --> 00:04:31,300
Many people think it's 90%. 
Even doctors sometimes get this 

81
00:04:31,300 --> 00:04:35,600
wrong, which is disturbing. 
The correct answer is about zero

82
00:04:35,600 --> 00:04:40,900
point zero, zero zero nine 
percent less than 1 in 100,000 

83
00:04:41,700 --> 00:04:45,500
explanation. 
There are 300 million people in 

84
00:04:45,500 --> 00:04:52,000
the country of these 300 one, 
millionth have the disease and 

85
00:04:52,000 --> 00:04:54,800
299 million. 
Nine hundred ninety, nine 

86
00:04:54,800 --> 00:04:59,000
thousand seven hundred, don't 
the test is 90% accurate. 

87
00:04:59,100 --> 00:05:05,400
So 270 of the 300, people who 
have the disease wouldn't test 

88
00:05:05,400 --> 00:05:09,100
positive, that's 90 percent and 
twenty nine million, nine 

89
00:05:09,100 --> 00:05:11,100
hundred and ninety nine 
thousand, nine hundred and 

90
00:05:11,100 --> 00:05:13,900
seventy of the two. 
Hundred and ninety. 

91
00:05:13,900 --> 00:05:16,800
Nine million nine hundred and 
ninety-nine thousand seven 

92
00:05:16,800 --> 00:05:20,700
hundred who don't have the 
disease would also test positive

93
00:05:20,800 --> 00:05:24,700
that's 10%. 
So, out of all the people who 

94
00:05:24,700 --> 00:05:28,700
test positive, the proportion 
who actually have the disease is

95
00:05:28,700 --> 00:05:36,300
270 over 270 plus twenty nine 
million, nine hundred and ninety

96
00:05:36,300 --> 00:05:41,200
nine thousand, nine hundred and 
seventy approximately zero point

97
00:05:41,200 --> 00:05:48,100
zero zero zero. 09. 
Number four, cherry-picking, 

98
00:05:48,500 --> 00:05:51,600
cherry-picking, refers to the 
practice of sifting, through 

99
00:05:51,600 --> 00:05:54,800
evidence and selecting out, only
the bits, that support a 

100
00:05:54,800 --> 00:05:57,400
particular conclusion, ignoring 
the rest. 

101
00:05:57,700 --> 00:06:00,600
Simple example. 
I have a bag of marbles. 

102
00:06:01,000 --> 00:06:04,800
I want to convince you that most
of the marbles in the bag are 

103
00:06:04,800 --> 00:06:06,800
black. 
I look inside the bag which is 

104
00:06:06,800 --> 00:06:11,600
full of Many Colors of marbles. 
Black red teal chart truce and 

105
00:06:11,600 --> 00:06:13,800
so on. 
I pick out five black ones. 

106
00:06:14,100 --> 00:06:17,800
Show them to you and say see. 
These marbles came from the bag.

107
00:06:18,100 --> 00:06:22,000
I don't show you any of the 
other colored marbles that were 

108
00:06:22,100 --> 00:06:24,400
in the bag. 
You might be misled into 

109
00:06:24,400 --> 00:06:27,900
concluding that the bag is full 
of black marbles. 

110
00:06:28,600 --> 00:06:31,100
That's like what people do in 
political debate. 

111
00:06:31,100 --> 00:06:34,500
If I want to convince you say 
that affirmative action is bad. 

112
00:06:34,600 --> 00:06:37,800
I might search for cases where 
affirmative action was tried and

113
00:06:37,800 --> 00:06:40,100
it didn't work or it had harmful
effects. 

114
00:06:40,300 --> 00:06:43,400
If, I want to convince you, that
it's good, I search for cases, 

115
00:06:43,400 --> 00:06:45,700
where it really helped someone 
of course. 

116
00:06:45,900 --> 00:06:48,100
Course, both kinds of cases 
exist. 

117
00:06:48,100 --> 00:06:51,800
It's a big Society full of 
millions of people almost any 

118
00:06:51,800 --> 00:06:56,400
policy is going to benefit some 
people and harm others because 

119
00:06:56,400 --> 00:06:58,100
of this, you should be 
suspicious. 

120
00:06:58,300 --> 00:07:01,300
When someone tells you stories 
designed to support a 

121
00:07:01,300 --> 00:07:05,300
conclusion, always ask yourself 
whether they have a bias that 

122
00:07:05,300 --> 00:07:09,000
might have caused them to 
cherry-pick the data. 

123
00:07:10,500 --> 00:07:15,100
Number 5, confirmation bias, 
when asked to evaluate a theory,

124
00:07:15,300 --> 00:07:18,700
people have a systematic 
tendency to look for evidence 

125
00:07:18,700 --> 00:07:22,000
supporting the theory and not 
look for evidence against it 

126
00:07:22,100 --> 00:07:23,800
this happens, especially for 
theories. 

127
00:07:24,000 --> 00:07:27,000
We already believe but can also 
happen for theories. 

128
00:07:27,100 --> 00:07:31,300
We initially have no opinion. 
About example, if asked, whether

129
00:07:31,500 --> 00:07:34,100
liberal politicians are more 
corrupt than conservative 

130
00:07:34,100 --> 00:07:37,400
politicians, a conservative 
would search through his memory 

131
00:07:37,600 --> 00:07:39,300
for any cases of a liberal 
doing. 

132
00:07:39,400 --> 00:07:42,900
Something corrupt and he would 
not search through his memory. 

133
00:07:42,900 --> 00:07:46,700
For cases of conservatives being
corrupt, a liberal on the other,

134
00:07:46,700 --> 00:07:50,500
hand would do the reverse. 
Each just looks for cases that 

135
00:07:50,500 --> 00:07:53,700
support his existing belief and 
does not look for evidence 

136
00:07:54,000 --> 00:07:57,300
against it. 
This is called confirmation bias

137
00:07:57,600 --> 00:08:00,100
to combat. 
This, it is necessary to make a 

138
00:08:00,100 --> 00:08:02,500
conscious effort to think of 
exceptions to the 

139
00:08:02,500 --> 00:08:06,600
generalizations that you accept 
an to look for evidence against 

140
00:08:06,600 --> 00:08:11,000
your existing beliefs, whenever 
you feel Inclined to cite some 

141
00:08:11,000 --> 00:08:15,000
examples of supporting belief a 
stop and ask yourself. 

142
00:08:15,100 --> 00:08:18,400
Whether you can also think of 
similar examples supporting 

143
00:08:18,500 --> 00:08:23,300
approximately a Number six, for 
Julie. 

144
00:08:23,900 --> 00:08:27,700
Humans are born credulous. 
We instinctively believe what 

145
00:08:27,700 --> 00:08:30,700
people tell us. 
Even with no, cooperation. 

146
00:08:30,900 --> 00:08:34,200
We are especially crunch Allis 
about statistics or other 

147
00:08:34,200 --> 00:08:36,500
information. 
That sounds like objective 

148
00:08:36,500 --> 00:08:39,000
facts. 
Unfortunately, we are not so 

149
00:08:39,000 --> 00:08:42,900
scrupulous when it comes to 
accurately and non misleadingly 

150
00:08:43,000 --> 00:08:46,100
reporting the facts. 
There was an enormous amount of 

151
00:08:46,100 --> 00:08:50,800
disinformation in the world 
particularly about politics and 

152
00:08:50,800 --> 00:08:52,600
other Matters of public 
interest. 

153
00:08:52,700 --> 00:08:56,700
If the public is interested in 
it, there is bullshit about it. 

154
00:08:56,900 --> 00:08:59,800
I have noticed that this 
bullshit tends to fall into 

155
00:08:59,800 --> 00:09:04,000
three main categories. 
First ideological propaganda. 

156
00:09:04,400 --> 00:09:08,300
If you learn about an issue from
a partisan source, for instance,

157
00:09:08,400 --> 00:09:10,700
you've read about gun control on
a gun control. 

158
00:09:10,700 --> 00:09:13,900
Advocacy website, or you hear 
the day's news from a 

159
00:09:13,900 --> 00:09:18,800
conservative radio show, you 
will get pretty much 100% 

160
00:09:18,800 --> 00:09:21,000
propaganda. 
Facts will be exaggerated. 

161
00:09:21,100 --> 00:09:24,900
It cherry-pick deceptively 
phrased or otherwise misleading.

162
00:09:25,300 --> 00:09:28,700
Normally you will have no way of
guessing the specific way in 

163
00:09:28,700 --> 00:09:31,800
which the information is 
deceptive making the information

164
00:09:31,800 --> 00:09:34,700
essentially worthless for 
drawing inferences. 

165
00:09:35,600 --> 00:09:39,200
Second sensationalism mainstream
news sources. 

166
00:09:39,200 --> 00:09:42,200
Make money by getting as many 
people as possible to watch 

167
00:09:42,200 --> 00:09:46,400
their shows, read their articles
and so on to do that, they try 

168
00:09:46,400 --> 00:09:50,100
to make everything sound as 
scary, exciting, outrageous or 

169
00:09:50,100 --> 00:09:53,000
otherwise. 
As possible. 

170
00:09:53,500 --> 00:09:58,200
Third laziness, most people who 
write for public consumption are

171
00:09:58,200 --> 00:10:01,400
lazy and alack. 
Expertise about the things they 

172
00:10:01,400 --> 00:10:03,900
write about. 
If a story has some technical 

173
00:10:03,900 --> 00:10:08,100
aspect example, Science News 
journalist probably won't 

174
00:10:08,100 --> 00:10:11,700
understand it and they may get 
basic facts backwards. 

175
00:10:11,700 --> 00:10:15,700
Also, they often just talked to 
one or a few sources and print 

176
00:10:15,700 --> 00:10:18,900
whatever those sources. 
Say, even if the sources have 

177
00:10:18,900 --> 00:10:23,100
obvious biases. 
I can't give you adequate 

178
00:10:23,100 --> 00:10:26,100
evidence for all that right now.
But here's an anecdote that 

179
00:10:26,100 --> 00:10:29,700
illustrates what I mean. 
I once heard a story on NPR 

180
00:10:29,900 --> 00:10:33,100
National Public Radio, a 
left-leaning radio news source, 

181
00:10:33,300 --> 00:10:36,700
it was about a man on death row.
Who was about to be executed 

182
00:10:36,700 --> 00:10:39,900
from the story. 
It appeared that the man was 

183
00:10:39,900 --> 00:10:42,600
innocent new evidence had 
emerged after the trial. 

184
00:10:42,700 --> 00:10:45,900
Several of the witnesses have 
recanted their testimony yet. 

185
00:10:45,900 --> 00:10:49,000
The courts had refused to Grant 
a new trial. 

186
00:10:49,000 --> 00:10:52,700
The only remaining Hope was for 
the governor to Grant a stay of 

187
00:10:52,700 --> 00:10:55,000
execution. 
There was an online petition 

188
00:10:55,000 --> 00:10:57,800
that listeners could sign. 
Usually I just accept new 

189
00:10:57,800 --> 00:11:01,100
stories and then go on with my 
day, but on that occasion, I 

190
00:11:01,108 --> 00:11:04,200
decided to look into the story 
before signing the petition with

191
00:11:04,200 --> 00:11:07,300
a little Googling. 
I found the court decision from 

192
00:11:07,300 --> 00:11:10,500
the convicts most recent appeal,
which had been denied. 

193
00:11:10,500 --> 00:11:13,400
I read the decision which 
contained a summary of The Facts

194
00:11:13,400 --> 00:11:17,100
of the case and an explanation 
of the judge's decision. 

195
00:11:17,600 --> 00:11:21,000
What it revealed was that the 
NPR story was. 

196
00:11:21,200 --> 00:11:24,800
Bullshit. 
What NPR said was basically just

197
00:11:24,800 --> 00:11:28,100
what the defendants lawyer had 
claimed. 

198
00:11:28,100 --> 00:11:29,600
The court carefully explained 
why? 

199
00:11:29,600 --> 00:11:32,500
Each of those claims was bogus 
and provided? 

200
00:11:32,500 --> 00:11:36,500
No basis for an appeal, the most
striking claim which had 

201
00:11:36,500 --> 00:11:38,600
initially made me. 
Think the defendant was probably

202
00:11:38,600 --> 00:11:43,400
innocent, was the multiple 
Witnesses had recanted their 

203
00:11:43,700 --> 00:11:46,000
testimony. 
What had actually happened was 

204
00:11:46,000 --> 00:11:49,600
this the defense lawyer went 
back to the witnesses many years

205
00:11:49,600 --> 00:11:52,800
after the original trial. 
And question them on details of 

206
00:11:52,800 --> 00:11:55,100
the case. 
Several of them either couldn't 

207
00:11:55,100 --> 00:11:57,500
remember the details or reporter
details. 

208
00:11:57,500 --> 00:12:01,400
Slightly differently example. 
What color shirt someone was 

209
00:12:01,400 --> 00:12:05,200
wearing the lawyer described 
this as recanting their 

210
00:12:05,200 --> 00:12:08,900
testimony, but none of them had 
changed their mind about the 

211
00:12:08,900 --> 00:12:13,300
defendant being guilty. 
The NPR journalists had 

212
00:12:13,300 --> 00:12:17,700
apparently, just credulous Lee 
reported want the lawyer told 

213
00:12:17,700 --> 00:12:21,500
them without bothering to look 
up the court documents from The 

214
00:12:21,500 --> 00:12:26,000
case, why would they do that? 
Three reasons, one ideological 

215
00:12:26,000 --> 00:12:27,800
bias. 
The story, painted the death 

216
00:12:27,800 --> 00:12:31,400
penalty in a bad light which a 
left-leaning news Outlet would 

217
00:12:31,400 --> 00:12:36,000
like to sensationalism the story
of an innocent man about to be 

218
00:12:36,000 --> 00:12:39,900
executed, grab the audience's 
attention and inflamed their 

219
00:12:40,000 --> 00:12:44,300
passions three laziness. 
Checking on the story, would 

220
00:12:44,300 --> 00:12:46,900
have required work. 
Why put in the work? 

221
00:12:46,900 --> 00:12:50,400
When you know, almost all of 
your audience will just accept 

222
00:12:50,400 --> 00:12:53,300
whatever you say. 
Long experience has led me to 

223
00:12:53,300 --> 00:12:57,800
think that that case was not 
unusual. 

224
00:12:58,000 --> 00:13:01,500
This is the way the news media 
Works lesson. 

225
00:13:01,800 --> 00:13:04,400
Popular media stories are 
untrustworthy. 

226
00:13:04,700 --> 00:13:05,600
By the way. 
It's no good. 

227
00:13:05,600 --> 00:13:09,200
Checking them against other 
popular news sources because 

228
00:13:09,200 --> 00:13:10,900
they basically all copy from 
each other. 

229
00:13:11,100 --> 00:13:15,700
That also goes for example, most
bloggers your next-door neighbor

230
00:13:15,900 --> 00:13:20,400
and other casual information 
sources for relatively reliable 

231
00:13:20,400 --> 00:13:22,800
information. 
Look at academic books and 

232
00:13:22,800 --> 00:13:25,200
articles. 
And government reports. 

233
00:13:25,400 --> 00:13:29,800
For example, Census Bureau 
reports, FBI crime reports. 

234
00:13:31,300 --> 00:13:35,800
Number seven, dogmatism, and 
overconfidence people who study 

235
00:13:35,800 --> 00:13:38,900
rationality have a notion called
calibration. 

236
00:13:39,100 --> 00:13:42,400
Your beliefs are said to be 
well, calibrated when your level

237
00:13:42,400 --> 00:13:46,300
of confidence matches, the 
probability of your being 

238
00:13:46,300 --> 00:13:50,700
correct, for example, for all 
the beliefs that you hold with 

239
00:13:50,700 --> 00:13:55,800
90% confidence, about 90% of 
them should be true. 

240
00:13:55,800 --> 00:13:58,100
When you're 100% confident of 
things. 

241
00:13:58,300 --> 00:14:01,900
They should be true. 100% of the
Time Etc. 

242
00:14:02,100 --> 00:14:04,300
Most people are badly 
calibrated. 

243
00:14:04,300 --> 00:14:08,500
In fact, almost everyone airs in
a particular direction almost 

244
00:14:08,500 --> 00:14:11,700
everyone's beliefs are too 
confident. 

245
00:14:11,900 --> 00:14:16,800
People say they are 100% certain
of a bunch of things but it 

246
00:14:16,800 --> 00:14:20,400
turns out that only say 85% of 
those things are actually true. 

247
00:14:20,800 --> 00:14:22,600
There are psychological studies 
of this. 

248
00:14:22,900 --> 00:14:27,100
This is the problem of 
overconfidence, almost everyone 

249
00:14:27,100 --> 00:14:30,600
has it and almost no one has the
opposite problem. 

250
00:14:30,700 --> 00:14:33,700
L'm under confidence. 
So you should assume that you 

251
00:14:33,700 --> 00:14:37,600
are probably overconfident to 
you should therefore try to 

252
00:14:37,600 --> 00:14:40,900
reduce your confidence in your 
beliefs, particularly about 

253
00:14:40,900 --> 00:14:45,400
controversial things, and 
particularly for speculative and

254
00:14:45,500 --> 00:14:47,400
subjective. 
Claims. 

255
00:14:48,400 --> 00:14:54,300
Number eight ideological cause 
judgments back in 2008 to 2009, 

256
00:14:54,500 --> 00:14:57,300
America suffered a severe 
economic recession. 

257
00:14:57,400 --> 00:15:00,000
A lot of people lost money and 
lost their jobs and were 

258
00:15:00,000 --> 00:15:02,600
generally. 
Unhappy would set it off with 

259
00:15:02,600 --> 00:15:06,100
problems in real estate. 
Home prices had gotten very 

260
00:15:06,100 --> 00:15:09,600
high, then they dropped a lot of
people started defaulting on, 

261
00:15:09,700 --> 00:15:13,200
not repaying their home loans. 
Banks were in a lot of trouble 

262
00:15:13,400 --> 00:15:16,400
and other investors and 
financial companies were in 

263
00:15:16,400 --> 00:15:18,500
trouble. 
Because they'd made Investments 

264
00:15:18,700 --> 00:15:22,500
That depended on home prices, 
Staying High, and Home Loans. 

265
00:15:22,500 --> 00:15:25,600
Getting repaid. 
In the wake of the crisis, many 

266
00:15:25,600 --> 00:15:29,400
people tried to explain why it 
had all happened. 

267
00:15:29,500 --> 00:15:32,600
This included people. 
With opposing ideologies, 

268
00:15:32,800 --> 00:15:36,800
roughly, there were people with 
pro-government and people with 

269
00:15:36,800 --> 00:15:40,100
anti-government ideologies and 
both tried to explain the 

270
00:15:40,100 --> 00:15:42,300
crisis. 
Can you guess what? 

271
00:15:42,300 --> 00:15:45,400
The two sides said, the 
pro-government, people said the 

272
00:15:45,400 --> 00:15:49,100
recession happened because there
wasn't enough regulation, and 

273
00:15:49,100 --> 00:15:52,800
they had listed regulations. 
That if they had been in place, 

274
00:15:53,100 --> 00:15:55,300
would probably have presented 
the crisis. 

275
00:15:55,500 --> 00:15:58,800
The anti-government people said,
the recession happened because 

276
00:15:58,800 --> 00:16:00,500
there was too much government 
intervention. 

277
00:16:00,700 --> 00:16:03,400
Engine and they listed existing 
government policies. 

278
00:16:03,600 --> 00:16:06,500
That if they hadn't been in 
place, the crisis probably 

279
00:16:06,500 --> 00:16:10,600
wouldn't have happened. 
Notice that the basic factual 

280
00:16:10,600 --> 00:16:13,500
claims of both sides are 
perfectly consistent. 

281
00:16:13,500 --> 00:16:16,300
It's perfectly possible that 
they were some actions. 

282
00:16:16,300 --> 00:16:19,300
The government took such that if
the government hadn't taken 

283
00:16:19,300 --> 00:16:21,500
them, the crisis wouldn't have 
happened. 

284
00:16:21,600 --> 00:16:24,900
And also, there were some 
actions, the government failed 

285
00:16:24,900 --> 00:16:28,000
to take such that. 
If it had taken them, the crisis

286
00:16:28,000 --> 00:16:30,500
wouldn't have happened. 
It's perfectly plausible. 

287
00:16:30,600 --> 00:16:34,400
Oil, that the crisis could have 
been averted in more than one 

288
00:16:34,400 --> 00:16:37,900
way, either by adding certain 
government interventions or by 

289
00:16:37,900 --> 00:16:41,400
removing some other government 
interventions, which alternative

290
00:16:41,400 --> 00:16:44,800
you focus on depends on your 
initial ideology. 

291
00:16:44,800 --> 00:16:50,000
Both sides took the episode to 
further support their ideology. 

292
00:16:50,000 --> 00:16:53,100
We have too much government, or 
we need more government. 

293
00:16:53,100 --> 00:16:57,500
These conclusions were supported
by their respective, causal 

294
00:16:57,500 --> 00:17:00,200
interpretations. 
The recession was caused by 

295
00:17:00,200 --> 00:17:02,800
government. 
Interventions, or the recession 

296
00:17:02,800 --> 00:17:06,099
was caused by failure of 
government to intervene. 

297
00:17:07,300 --> 00:17:11,300
Who was right, assume the facts 
are as stated that some 

298
00:17:11,300 --> 00:17:14,500
additional interventions would 
have prevented the recession and

299
00:17:14,500 --> 00:17:17,099
the repeal of some other 
interventions would have 

300
00:17:17,099 --> 00:17:19,800
prevented the recession in that 
case. 

301
00:17:20,099 --> 00:17:24,900
We should either accept both 
causal claims or reject both 

302
00:17:24,900 --> 00:17:30,000
causal claims, depending on what
we mean by cause if we mean Soul

303
00:17:30,000 --> 00:17:31,900
cause then we should reject 
both. 

304
00:17:31,900 --> 00:17:35,300
Causal claims IE. 
We should say, the recession was

305
00:17:35,300 --> 00:17:39,900
not caused either by Government 
intervention or by failure to 

306
00:17:39,900 --> 00:17:44,800
intervene if we just mean Factor
such that if it were changed, 

307
00:17:44,900 --> 00:17:46,400
the effect wouldn't have 
happened. 

308
00:17:46,700 --> 00:17:49,700
Then we should accept both 
causal claims. 

309
00:17:49,900 --> 00:17:53,200
The recession was caused by 
intervention and by failure to 

310
00:17:53,400 --> 00:17:57,100
intervene. 
It's okay to say that X was 

311
00:17:57,100 --> 00:18:01,600
caused by y provided. 
That you also recognize all the 

312
00:18:01,600 --> 00:18:06,700
other things that cause X in the
same sense, if there are many 

313
00:18:07,000 --> 00:18:11,100
Different causes then you need 
additional evidence, or 

314
00:18:11,100 --> 00:18:15,900
arguments to establish, which of
the causes is the best one to 

315
00:18:15,900 --> 00:18:19,200
change. 
In the recession case, we would 

316
00:18:19,200 --> 00:18:23,600
need independent arguments to 
establish, which, cause of the 

317
00:18:23,600 --> 00:18:25,900
recession intervention or 
failure to intervene. 

318
00:18:26,200 --> 00:18:28,300
It would have been better to 
change. 

319
00:18:29,500 --> 00:18:34,900
Number nine over simplification 
people, very often oversimplify 

320
00:18:35,000 --> 00:18:38,100
philosophical issues. 
Say you're thinking about the 

321
00:18:38,100 --> 00:18:42,600
morality of abortion attempting.
Simplification would be to say 

322
00:18:42,700 --> 00:18:45,900
that there are two positions 
pro-choice and pro-life or 

323
00:18:45,900 --> 00:18:50,200
pro-and anti-abortion. 
Either fetuses are people in 

324
00:18:50,200 --> 00:18:53,600
killing them as murder or fetus 
as aren't people and killing 

325
00:18:53,600 --> 00:18:56,300
them is perfectly fine. 
But this overlooks the 

326
00:18:56,300 --> 00:18:58,800
possibility that late-term 
fetus. 

327
00:18:58,900 --> 00:19:02,600
Our people but early term 
fetuses are not or maybe 

328
00:19:02,600 --> 00:19:07,000
personhood comes in degrees and 
fetus has become progressively, 

329
00:19:07,000 --> 00:19:10,800
more person like as they 
develop, or maybe, fetuses are 

330
00:19:10,800 --> 00:19:15,200
persons in some senses, but 
non-persons in other senses. 

331
00:19:15,200 --> 00:19:19,100
So, there was a range of 
possible positions not just to 

332
00:19:19,100 --> 00:19:23,000
viewing things in black and 
white terms is a common over 

333
00:19:23,000 --> 00:19:25,800
simplification. 
We look at two simple positions,

334
00:19:25,800 --> 00:19:29,600
rather than, considering a 
spectrum of Possibilities. 

335
00:19:29,700 --> 00:19:34,000
The problem is that often the 
truth is a more subtle position 

336
00:19:34,300 --> 00:19:39,100
that doesn't clearly fall under 
either of the two simplest 

337
00:19:39,100 --> 00:19:45,500
categories of view. 
Number 10 P hacking similar to 

338
00:19:45,500 --> 00:19:48,300
cherry-picking P hacking or data
mining. 

339
00:19:48,400 --> 00:19:52,500
Sometimes happens in science. 
A scientist has a large amount 

340
00:19:52,500 --> 00:19:55,500
of statistical data with 
different variables. 

341
00:19:56,100 --> 00:19:58,700
Even if all the data is 
completely random. 

342
00:19:58,900 --> 00:20:03,800
Any complex set of data is going
to show some pattern that looked

343
00:20:03,800 --> 00:20:08,300
significant, essentially one can
take the data and use it to test

344
00:20:08,400 --> 00:20:10,800
many different possible 
hypotheses. 

345
00:20:11,200 --> 00:20:14,900
Even if all the hypotheses are 
false, eventually just by chance

346
00:20:14,900 --> 00:20:16,800
due to random variations in the 
data. 

347
00:20:17,100 --> 00:20:21,500
One of the hypotheses will pass 
a test for statistical 

348
00:20:21,500 --> 00:20:24,000
significance. 
This is one reason why many 

349
00:20:24,000 --> 00:20:26,900
published research results, 
especially in medicine 

350
00:20:26,900 --> 00:20:31,800
psychology and social science 
are false example, a study will 

351
00:20:31,800 --> 00:20:34,600
find that some food increases 
the risk of cancer for 

352
00:20:34,600 --> 00:20:38,200
non-smoking middle-aged men, but
then when someone tries to 

353
00:20:38,200 --> 00:20:41,700
replicate it and they don't get 
the same result, it's because 

354
00:20:41,700 --> 00:20:45,600
the original result was just due
to chance. 

355
00:20:46,500 --> 00:20:50,300
Number 11, speculation 
speculative claims are 

356
00:20:50,300 --> 00:20:54,500
essentially guesses about things
that we lack the evidence to 

357
00:20:54,500 --> 00:20:59,200
establish as yet claims about 
the future or claims about What 

358
00:20:59,200 --> 00:21:02,800
would have happened in 
hypothetical alternative 

359
00:21:02,800 --> 00:21:07,800
possibilities are good examples 
of speculative claims example, 

360
00:21:08,000 --> 00:21:11,100
you're arguing about whether 
it's good for government to try 

361
00:21:11,100 --> 00:21:14,400
to stimulate the economy by 
spending money. 

362
00:21:14,400 --> 00:21:17,900
You say this is good. 
Because example, if the 

363
00:21:17,900 --> 00:21:21,800
government hadn't stimulated, 
the economy back in 2009, the 

364
00:21:21,800 --> 00:21:24,100
recession would have continued 
much longer. 

365
00:21:24,300 --> 00:21:27,500
This is speculative. 
We don't know what would have 

366
00:21:27,500 --> 00:21:31,500
happened because Was in fact, 
the government did pass a 

367
00:21:31,500 --> 00:21:36,900
stimulus plan and we can't go 
back in time and change that to 

368
00:21:36,900 --> 00:21:39,400
see what would have happened. 
If they didn't. 

369
00:21:39,400 --> 00:21:42,400
The problem with speculative 
claims is that people with 

370
00:21:42,400 --> 00:21:46,300
different philosophical or 
political or religious beliefs 

371
00:21:46,500 --> 00:21:50,800
tend to find very different. 
Speculations plausible example, 

372
00:21:51,100 --> 00:21:53,600
people who are suspicious of 
government will find it more 

373
00:21:53,600 --> 00:21:56,400
plausible that without 
government stimulus. 

374
00:21:56,400 --> 00:21:58,400
The recession would have been 
shorter. 

375
00:21:58,400 --> 00:22:01,700
So Arguments that start from 
speculative. 

376
00:22:01,800 --> 00:22:07,000
Premises are typically not 
rationally persuasive advice. 

377
00:22:07,300 --> 00:22:10,200
If you want to rationally 
persuade, people of something 

378
00:22:10,400 --> 00:22:16,800
try to avoid speculation number 
12 subjective claims roughly. 

379
00:22:16,800 --> 00:22:20,800
A subjective claim is one that 
requires a judgment call. 

380
00:22:20,900 --> 00:22:23,800
So it can't just be 
straightforwardly and decisively

381
00:22:24,000 --> 00:22:26,800
established. 
For example, the Judgment that 

382
00:22:26,900 --> 00:22:30,600
political candidate a Is 
unqualified for office. 

383
00:22:30,700 --> 00:22:33,500
The Judgment that it's worse to 
be unjustly prison for five 

384
00:22:33,500 --> 00:22:37,000
years, then to be prevented from
migrating to the country. 

385
00:22:37,000 --> 00:22:40,100
One wants to live in the 
Judgment that Louis CK's jokes 

386
00:22:40,100 --> 00:22:43,300
are offensive Etc. 
This differs from speculative 

387
00:22:43,300 --> 00:22:47,400
claims because in the case of 
speculation there might be ways 

388
00:22:47,400 --> 00:22:50,800
that the claim could in 
principle be decisively 

389
00:22:50,800 --> 00:22:52,600
verified. 
It just hasn't. 

390
00:22:52,600 --> 00:22:57,000
In fact been verified note. 
I am not saying that there was 

391
00:22:57,100 --> 00:23:00,300
no fact or no. 
Answer as to, whether these 

392
00:23:00,300 --> 00:23:03,700
things are the case, or that 
they are dependent on people's 

393
00:23:03,700 --> 00:23:06,100
opinions. 
What I am saying is that there 

394
00:23:06,100 --> 00:23:10,700
are not clear established 
criteria for these claims. 

395
00:23:10,800 --> 00:23:13,100
So it is difficult to verify 
them. 

396
00:23:13,400 --> 00:23:17,400
Maybe it's true that Louise 
jokes are offensive. 

397
00:23:17,500 --> 00:23:22,000
But if someone doesn't find him 
offencive there is no decisive 

398
00:23:22,000 --> 00:23:26,200
way of proving that he is people
often rely on subjective 

399
00:23:26,200 --> 00:23:28,100
premises. 
When arguing about 

400
00:23:28,100 --> 00:23:31,700
controversial, All issues. 
The problem with this is that 

401
00:23:31,700 --> 00:23:36,400
subjective claims are more open 
to bias than relatively 

402
00:23:36,400 --> 00:23:38,800
objective. 
That's the opposite of 

403
00:23:38,800 --> 00:23:42,100
subjective claims. 
So, people with different 

404
00:23:42,100 --> 00:23:45,500
philosophical or political or 
religious views will tend to 

405
00:23:45,500 --> 00:23:48,400
disagree a lot about subjective 
claims. 

406
00:23:48,700 --> 00:23:52,300
And for that reason, they are 
ill-suited to serve as premises 

407
00:23:52,500 --> 00:23:56,300
in philosophical political or 
religious arguments advice. 

408
00:23:56,500 --> 00:24:02,400
Try to base your arguments as As
possible on relatively objective

409
00:24:02,400 --> 00:24:06,600
claims. 
Number 13, treatment effects, 

410
00:24:06,600 --> 00:24:11,400
adverse selection effects. 
Let's say you have created a new

411
00:24:11,400 --> 00:24:14,000
educational program for 
preschool children. 

412
00:24:14,100 --> 00:24:17,500
You want to know whether it 
improves learning or not. 

413
00:24:17,500 --> 00:24:21,000
What you would do is look at 
kids after they've had your 

414
00:24:21,000 --> 00:24:25,000
program and compare them to kids
of the same age who didn't have 

415
00:24:25,000 --> 00:24:29,700
your program and see if the 
first group perform better on 

416
00:24:29,800 --> 00:24:31,900
tests. 
Let's say, kids who had your 

417
00:24:31,900 --> 00:24:36,900
special program perform 10% Her 
on later tests on average, then 

418
00:24:36,900 --> 00:24:40,100
you'd probably conclude that 
your program works. 

419
00:24:40,400 --> 00:24:42,900
But wait here is another 
possibility. 

420
00:24:43,000 --> 00:24:47,200
Suppose as would usually be the 
case that the kids who entered 

421
00:24:47,200 --> 00:24:50,500
your special educational 
program, where kids whose 

422
00:24:50,500 --> 00:24:54,400
parents chose to enroll them in 
that program. 

423
00:24:54,500 --> 00:24:57,800
The rest were kids whose parents
did not decide to enroll them. 

424
00:24:58,100 --> 00:25:01,000
Furthermore, maybe the parents 
who enrolled their kids in 

425
00:25:01,000 --> 00:25:06,400
special programs are on average.
Smarter and value education more

426
00:25:06,700 --> 00:25:09,200
than the parents who didn't do 
that. 

427
00:25:09,400 --> 00:25:12,900
Furthermore, maybe intelligence 
and value placed on learning are

428
00:25:12,900 --> 00:25:15,800
partly genetic. 
And so these parents pass those 

429
00:25:15,800 --> 00:25:20,600
traits on to their kids. 
So the children who went into 

430
00:25:20,600 --> 00:25:24,900
your program, we're already on 
average smarter and more 

431
00:25:24,900 --> 00:25:29,000
interested in learning than the 
children who didn't go into the 

432
00:25:29,000 --> 00:25:33,300
program and maybe that explains 
why they did ten percent better 

433
00:25:33,500 --> 00:25:36,000
on. 
Asks, after the program, maybe 

434
00:25:36,000 --> 00:25:37,800
your program had no effect at 
all. 

435
00:25:38,000 --> 00:25:42,400
It's just that you got the smart
kids in it and that made the 

436
00:25:42,400 --> 00:25:46,900
program look good. 
That is an example of a 

437
00:25:46,900 --> 00:25:52,200
selection effect a case where it
looks like a causes B, but it's 

438
00:25:52,200 --> 00:25:56,100
actually just that the instances
of a that you tested were 

439
00:25:56,100 --> 00:26:00,300
already more likely to be bees, 
for other reasons, selection, 

440
00:26:00,300 --> 00:26:03,200
effects are contrasted with 
treatment effects. 

441
00:26:03,800 --> 00:26:07,300
Cases, where the thing you're 
testing, really causes the 

442
00:26:07,300 --> 00:26:10,400
effect that it's thought to 
cause any. 

443
00:26:10,400 --> 00:26:14,900
Education example, academic 
success is correlated with 

444
00:26:14,900 --> 00:26:18,200
taking the special program. 
This could be due to treatment 

445
00:26:18,200 --> 00:26:20,100
effects. 
Meaning, the program causes kids

446
00:26:20,100 --> 00:26:23,000
to learn more or due to 
selection effects. 

447
00:26:23,200 --> 00:26:26,900
Meaning the program, selects 
students who are already good at

448
00:26:26,900 --> 00:26:30,600
Learning. 
Selection effects are very often

449
00:26:30,600 --> 00:26:34,900
mistaken for treatment effects. 
Another example, you want to 

450
00:26:34,900 --> 00:26:38,100
know if some vitamin improves 
People's Health. 

451
00:26:38,100 --> 00:26:42,200
So you look at people who take 
the supplements of that vitamin 

452
00:26:42,200 --> 00:26:45,500
regularly and you find that they
are healthier than the people 

453
00:26:45,500 --> 00:26:47,800
who don't take it. 
You think that shows that 

454
00:26:47,800 --> 00:26:51,600
vitamin supplements are good for
people, but actually it's more 

455
00:26:51,600 --> 00:26:54,600
likely, a selection effect. 
People who take vitamins are 

456
00:26:54,600 --> 00:26:57,400
more likely to also be 
exercising eating. 

457
00:26:57,700 --> 00:27:01,200
Seafoods and so on which is why 
they would be healthier on 

458
00:27:01,200 --> 00:27:05,300
average, even if the vitamins 
did absolutely nothing. 

459
00:27:06,100 --> 00:27:11,100
Number 14. 
What about ISM similar to to 

460
00:27:11,200 --> 00:27:14,000
quo? 
Quay C-section 4.1 above? 

461
00:27:14,000 --> 00:27:17,800
What about ISM occurs, when 
someone criticizes something bad

462
00:27:18,000 --> 00:27:22,500
and you respond with what about 
X where X is some other bad 

463
00:27:22,500 --> 00:27:25,600
thing example. 
Someone complains that the 

464
00:27:25,600 --> 00:27:27,700
current president's budget 
proposal. 

465
00:27:27,800 --> 00:27:31,800
All has a very high deficit. 
You say, what about the previous

466
00:27:31,800 --> 00:27:34,300
president? 
He had deficits to or someone 

467
00:27:34,300 --> 00:27:37,500
complains that the president 
just murdered a child. 

468
00:27:37,600 --> 00:27:40,900
You respond that some other 
political figure from an 

469
00:27:40,900 --> 00:27:43,700
opposing party also murdered 
children. 

470
00:27:43,900 --> 00:27:47,500
What about that? 
You demand the reason people 

471
00:27:47,500 --> 00:27:51,300
engage in what about ism is that
rather than being interested in 

472
00:27:51,300 --> 00:27:54,900
Practical issues about what 
should be done in our current 

473
00:27:54,900 --> 00:27:57,500
situation? 
They instead see political. 

474
00:27:57,700 --> 00:28:02,300
Discussion as a kind of tribal 
contest, a competition between 

475
00:28:02,300 --> 00:28:07,300
their side and the other side 
where whoever makes their side. 

476
00:28:07,300 --> 00:28:10,900
Look better wins. 
So they don't want attention 

477
00:28:10,900 --> 00:28:14,100
focused on any flaws in one of 
their sides. 

478
00:28:14,100 --> 00:28:17,800
People example, a politician 
from their own political party. 

479
00:28:18,200 --> 00:28:21,900
So they try to divert attention 
to something. 

480
00:28:21,900 --> 00:28:25,300
That's bad about someone on the 
other side. 

481
00:28:25,600 --> 00:28:28,400
The problem. 
Is that this Practice 

482
00:28:28,700 --> 00:28:32,500
systematically prevents evil 
from being addressed for any 

483
00:28:32,500 --> 00:28:35,800
evil in the world unless it's 
literally the worst thing in the

484
00:28:35,800 --> 00:28:38,100
world. 
One can always Identify some 

485
00:28:38,100 --> 00:28:43,100
other even worse evil and say, 
what about that for any evil 

486
00:28:43,400 --> 00:28:47,100
done by any political leader? 
It will virtually always be 

487
00:28:47,100 --> 00:28:49,500
true. 
That some other leader from 

488
00:28:49,500 --> 00:28:54,800
another party has at some time, 
committed a similar evil and 

489
00:28:54,800 --> 00:28:58,500
also that members of that 
person's Party didn't do 

490
00:28:58,500 --> 00:29:01,600
anything about it. 
If your response when you hear 

491
00:29:01,600 --> 00:29:05,400
about any evil currently 
happening is to deflect 

492
00:29:05,400 --> 00:29:08,000
attention to some past evil 
committed. 

493
00:29:08,000 --> 00:29:12,600
By another person or group. 
That means that evil never gets 

494
00:29:12,600 --> 00:29:16,300
addressed attention. 
Always gets deflected Away by. 

495
00:29:16,500 --> 00:29:19,200
What about his mm. 
The next time someone else is 

496
00:29:19,200 --> 00:29:21,900
doing something evil. 
That won't be addressed either. 

497
00:29:22,000 --> 00:29:26,500
Because people will say what 
about the present evil that 

498
00:29:26,500 --> 00:29:28,900
wasn't properly. 
Pearly addressed.