1
00:00:00,240 --> 00:00:04,680
Speaker 1: From UFOs to psychic powers and government conspiracies. History is

2
00:00:04,760 --> 00:00:09,080
Speaker 1: riddled with unexplained events. You can turn back now or

3
00:00:09,200 --> 00:00:24,880
Speaker 1: learn the stuff they don't want you to know. M Hello,

4
00:00:25,079 --> 00:00:27,880
Speaker 1: welcome to the show. My name is Noel and there's

5
00:00:27,880 --> 00:00:30,920
Speaker 1: a match shaped void there. Yes, it is all true.

6
00:00:31,000 --> 00:00:34,920
Speaker 1: Our friend Matt, longtime compatriot, is pursuing adventures that you

7
00:00:34,920 --> 00:00:38,440
Speaker 1: will hear about very soon if all goes according to plan,

8
00:00:38,560 --> 00:00:40,800
Speaker 1: which I will if we know Matt right, he's a

9
00:00:40,840 --> 00:00:43,760
Speaker 1: solid guy. Whomi they called me Ben. We're here with

10
00:00:43,800 --> 00:00:47,120
Speaker 1: our super producer Paul Deckett. Most importantly, you are you,

11
00:00:47,240 --> 00:00:50,400
Speaker 1: and you are here, and that makes this stuff they

12
00:00:50,440 --> 00:00:55,080
Speaker 1: don't want you to know, an audio podcast that today

13
00:00:55,400 --> 00:00:57,920
Speaker 1: is going to be about YouTube. Yeah, that's kind of meta,

14
00:00:58,080 --> 00:01:00,680
Speaker 1: right is that? Okay? Is this too much? Now we're

15
00:01:00,680 --> 00:01:02,960
Speaker 1: gonna open up time space. It's gonna swallow us all

16
00:01:02,960 --> 00:01:08,040
Speaker 1: into our own Ben and Knoll shaped a digital do portal. Perhaps. Well,

17
00:01:08,600 --> 00:01:11,680
Speaker 1: it's interesting because a long time listeners, as you know,

18
00:01:12,080 --> 00:01:16,320
Speaker 1: we started this show uh years and years ago as

19
00:01:16,400 --> 00:01:21,520
Speaker 1: a YouTube show and slowly evolved into an audio podcast.

20
00:01:21,959 --> 00:01:24,160
Speaker 1: So we have a little bit of a behind the

21
00:01:24,280 --> 00:01:28,560
Speaker 1: scenes peek at the nuts and bolts, the strange arcane

22
00:01:28,600 --> 00:01:33,679
Speaker 1: mechanics that drive YouTube and this is nuts. We learned

23
00:01:33,720 --> 00:01:36,320
Speaker 1: some strange stuff as we were preparing for this episode.

24
00:01:37,319 --> 00:01:42,120
Speaker 1: What is America's largest social media platform? Most of us

25
00:01:42,160 --> 00:01:45,720
Speaker 1: would say, you know, Facebook, maybe Instagram or something. It's

26
00:01:45,760 --> 00:01:48,920
Speaker 1: not true. According to the Pew Research Center, seventy three

27
00:01:49,040 --> 00:01:53,600
Speaker 1: percent of US adults use YouTube, either via desktop or mobile.

28
00:01:53,800 --> 00:01:57,200
Speaker 1: And that's that compares to about sixty eight percent of

29
00:01:57,320 --> 00:02:02,040
Speaker 1: US adults who use Facebook. The numbers drop off precipitously

30
00:02:02,160 --> 00:02:06,720
Speaker 1: after that. The distant third is Instagram. Thirty five percent

31
00:02:06,800 --> 00:02:11,600
Speaker 1: of US adults use it. Adults aged eighteen to twenty

32
00:02:11,680 --> 00:02:16,720
Speaker 1: nine use YouTube, which is that's a massive number of anything.

33
00:02:16,840 --> 00:02:19,760
Speaker 1: Is pretty pretty impressive, right, Well, what's even crazier though,

34
00:02:19,840 --> 00:02:22,120
Speaker 1: is that, like that's eighteen to twenty nine and a

35
00:02:22,160 --> 00:02:25,799
Speaker 1: big segment of people on YouTube these days are kids. Yes,

36
00:02:26,120 --> 00:02:28,639
Speaker 1: didn't even yeah, burying the UIs. They're like, my kid

37
00:02:28,720 --> 00:02:31,840
Speaker 1: is obsessed. They call them YouTubers and all this stuff

38
00:02:31,840 --> 00:02:35,280
Speaker 1: and making me constantly feel like like an old man. Um.

39
00:02:35,320 --> 00:02:38,000
Speaker 1: But it's pretty wild how much stuff there is out

40
00:02:38,000 --> 00:02:41,240
Speaker 1: there that's specifically geared towards kids. And so I mean

41
00:02:41,280 --> 00:02:43,519
Speaker 1: it's almost when you take that into account, it's like

42
00:02:45,200 --> 00:02:49,600
Speaker 1: right right of adults eighteen nine and children. Yes, yeah,

43
00:02:49,639 --> 00:02:52,800
Speaker 1: that's a good point, because even the number is so big,

44
00:02:52,840 --> 00:02:55,440
Speaker 1: the proliferation is so great that even if you are

45
00:02:55,480 --> 00:02:59,720
Speaker 1: an anti YouTube person you're like, it's bananas, that's that's

46
00:02:59,760 --> 00:03:03,760
Speaker 1: not my thing, you will know someone who is very

47
00:03:03,800 --> 00:03:06,760
Speaker 1: into it, or at least casually watches it. And the

48
00:03:06,800 --> 00:03:10,480
Speaker 1: thing about YouTube is you cannot get that big without

49
00:03:10,560 --> 00:03:13,080
Speaker 1: making some waves. You know. It's sort of like the

50
00:03:13,120 --> 00:03:16,600
Speaker 1: old argument about behind at the root of every fortune

51
00:03:16,720 --> 00:03:19,399
Speaker 1: is some horrific crime, you know what I mean. It's

52
00:03:19,440 --> 00:03:22,840
Speaker 1: similar to that, and YouTube is no stranger to controversy.

53
00:03:22,960 --> 00:03:27,200
Speaker 1: So to find out how we got here and what

54
00:03:27,320 --> 00:03:30,840
Speaker 1: kind of crazy cover ups or conspiracies about YouTube exists

55
00:03:30,840 --> 00:03:33,640
Speaker 1: in the present day, we have to look at how

56
00:03:33,720 --> 00:03:38,000
Speaker 1: YouTube started here the facts. It's not that old. The

57
00:03:38,080 --> 00:03:43,080
Speaker 1: domain name YouTube dot com wasn't activate until February. I

58
00:03:43,120 --> 00:03:45,560
Speaker 1: have a very distinct memory of visiting l A for

59
00:03:45,600 --> 00:03:48,720
Speaker 1: the first time right right around that time and seeing

60
00:03:48,720 --> 00:03:51,280
Speaker 1: somebody work in the box office at the Upright Sits

61
00:03:51,320 --> 00:03:53,920
Speaker 1: in the Brigade Theater on a site called YouTube that

62
00:03:54,000 --> 00:03:58,800
Speaker 1: I was like, what the heck is that is this YouTube? Yeah,

63
00:03:58,960 --> 00:04:03,840
Speaker 1: and the first video wasn't even uploaded until April of

64
00:04:03,880 --> 00:04:06,800
Speaker 1: that year. A the video is something really but now

65
00:04:06,840 --> 00:04:09,160
Speaker 1: it was like dude playing with his dog or something.

66
00:04:09,200 --> 00:04:13,680
Speaker 1: He's it's one of the co founders zoo Tube. Yeah,

67
00:04:13,760 --> 00:04:16,160
Speaker 1: it's called me at the Zoo and the guys at

68
00:04:16,160 --> 00:04:19,640
Speaker 1: the San Diego Zoo. But just a year later, a

69
00:04:19,640 --> 00:04:21,840
Speaker 1: little more than a year later, on October nine, two

70
00:04:21,880 --> 00:04:26,000
Speaker 1: thousand six, YouTube was purchased by Google for one point

71
00:04:26,120 --> 00:04:29,640
Speaker 1: six five billion dollars. You know what I mean, that's

72
00:04:29,800 --> 00:04:32,520
Speaker 1: that's insane. A lot of us in the audience today

73
00:04:33,360 --> 00:04:35,440
Speaker 1: became familiar with stuff they don't want you to know

74
00:04:35,640 --> 00:04:40,200
Speaker 1: due to those original YouTube videos. And we have to

75
00:04:40,200 --> 00:04:42,680
Speaker 1: be very conscious of the fact that YouTube did a

76
00:04:42,680 --> 00:04:45,599
Speaker 1: lot for the show. It did a lot for many

77
00:04:45,680 --> 00:04:49,279
Speaker 1: How Stuff Work shows, And we could spend an entire

78
00:04:49,320 --> 00:04:53,679
Speaker 1: episode on YouTube's history, but we should probably skip ahead,

79
00:04:53,720 --> 00:04:58,120
Speaker 1: and for now, let's just note there's no argument that

80
00:04:58,200 --> 00:05:02,600
Speaker 1: YouTube is a powerful thing. Let's look at the numbers. Well,

81
00:05:02,720 --> 00:05:06,320
Speaker 1: we've got some statistics for you about YouTube today. Six

82
00:05:06,440 --> 00:05:10,520
Speaker 1: out of ten people prefer online video platforms to live TV.

83
00:05:10,760 --> 00:05:15,440
Speaker 1: So take that Comcast let let them take something they've

84
00:05:15,480 --> 00:05:18,480
Speaker 1: been holding sway over the world of entertainment for too long,

85
00:05:19,200 --> 00:05:22,640
Speaker 1: at least here in our neck of uh. In an

86
00:05:22,680 --> 00:05:25,800
Speaker 1: average month, eight out of ten eighteen to forty nine

87
00:05:25,880 --> 00:05:29,120
Speaker 1: year olds watch YouTube. That's just a rephrasing of the

88
00:05:29,200 --> 00:05:33,880
Speaker 1: other the similar statistic from earlier in two thousand fifteen,

89
00:05:34,200 --> 00:05:36,960
Speaker 1: eighteen to forty nine year old spent four percent less

90
00:05:36,960 --> 00:05:40,240
Speaker 1: time watching TV and time spent on YouTube went up

91
00:05:40,320 --> 00:05:46,040
Speaker 1: seventy four percent, which means not only is YouTube taking

92
00:05:46,080 --> 00:05:51,520
Speaker 1: away existing viewing time from traditional television, it's also adding

93
00:05:51,760 --> 00:05:55,440
Speaker 1: viewing time. Well, it's just like because it's like crowdsourced content.

94
00:05:55,520 --> 00:05:57,720
Speaker 1: You know, anything you can think of, somebody is probably

95
00:05:57,960 --> 00:06:01,160
Speaker 1: found it and uploaded it. Even if it gets wiped

96
00:06:01,200 --> 00:06:04,839
Speaker 1: by the web shaff or whatever relatively quickly, it'll just

97
00:06:04,839 --> 00:06:07,040
Speaker 1: pop up again somewhere else. So it's like, I mean,

98
00:06:07,279 --> 00:06:11,080
Speaker 1: I've always found it pretty amazing. What like obscure music

99
00:06:11,200 --> 00:06:13,760
Speaker 1: videos or concert videos you can find, you know, I mean,

100
00:06:13,800 --> 00:06:15,520
Speaker 1: it makes sense. It's just kind of like a rabbit

101
00:06:15,560 --> 00:06:20,200
Speaker 1: hole you can go down and at least on the surface.

102
00:06:20,279 --> 00:06:23,719
Speaker 1: The appeal of YouTube is that you are in charge

103
00:06:23,800 --> 00:06:27,080
Speaker 1: of the tube. You don't have to wait until eight

104
00:06:27,120 --> 00:06:30,840
Speaker 1: pm every Wednesday to watch your favorite show. You can

105
00:06:30,880 --> 00:06:33,480
Speaker 1: just pull up episodes and watch them whenever you want,

106
00:06:34,000 --> 00:06:37,000
Speaker 1: or at least segments of episodes. You can see some weird,

107
00:06:37,040 --> 00:06:40,280
Speaker 1: interesting stuff that you would not normally see on cable television.

108
00:06:40,960 --> 00:06:45,080
Speaker 1: And also you can navigate YouTube in around seventies six

109
00:06:45,160 --> 00:06:49,599
Speaker 1: different languages. The platforms launched in over eighty eight countries.

110
00:06:50,320 --> 00:06:54,440
Speaker 1: There are actual YouTube millionaires. The highest earning one is

111
00:06:54,480 --> 00:06:59,359
Speaker 1: a guy named Dan Middleton in He made about sixteen

112
00:06:59,440 --> 00:07:02,960
Speaker 1: and a half million dollars from YouTube alogne and that's

113
00:07:03,080 --> 00:07:07,280
Speaker 1: I believe. That's uh after YouTube or Google or their

114
00:07:07,279 --> 00:07:11,400
Speaker 1: parent company Alphabet took their their own toll because they

115
00:07:11,440 --> 00:07:14,320
Speaker 1: have a pretty hefty have a pretty hefty rate that

116
00:07:14,360 --> 00:07:18,640
Speaker 1: they impose on every creator. And on average there are

117
00:07:19,000 --> 00:07:23,440
Speaker 1: a billion mobile video views per day, one billion. This

118
00:07:23,600 --> 00:07:26,200
Speaker 1: is insane. And no, let me ask you a question.

119
00:07:26,480 --> 00:07:29,400
Speaker 1: Have you ever just had auto play on when you

120
00:07:29,400 --> 00:07:31,640
Speaker 1: have YouTube on in the background or something? Yeah? I

121
00:07:31,880 --> 00:07:34,280
Speaker 1: actually I don't watch YouTube like that as much as

122
00:07:34,360 --> 00:07:36,560
Speaker 1: as some, but my my roommate Frank does a lot,

123
00:07:36,840 --> 00:07:39,520
Speaker 1: and it really does a good job of tracking his

124
00:07:39,600 --> 00:07:43,000
Speaker 1: behavior and feeding him and stuff. That's read up his alley. So,

125
00:07:43,160 --> 00:07:48,200
Speaker 1: I mean, obviously algorithm is been finally tuned over some time, right, Yeah,

126
00:07:48,200 --> 00:07:51,720
Speaker 1: that's what we're getting too. Yes, we are going to

127
00:07:51,880 --> 00:07:59,160
Speaker 1: explore the strange consequences of trusting your taste and preferences

128
00:07:59,480 --> 00:08:06,160
Speaker 1: to an incredibly complicated, smart piece of software and algorithm that,

129
00:08:07,400 --> 00:08:11,680
Speaker 1: largely without human intervention, attempts to figure out what you like,

130
00:08:12,280 --> 00:08:15,240
Speaker 1: why you like it, and how to give you more

131
00:08:15,280 --> 00:08:25,320
Speaker 1: of that. We'll be back after a word from our sponsor. Alright,

132
00:08:25,360 --> 00:08:29,280
Speaker 1: so nol hit me with it the story of the algorithm.

133
00:08:29,320 --> 00:08:33,120
Speaker 1: It's not there's not some cadre of people just hiding

134
00:08:33,240 --> 00:08:36,640
Speaker 1: in the next room purposely sending you videos when you

135
00:08:36,640 --> 00:08:39,679
Speaker 1: watch stuff. No, there are cadres of coders that can

136
00:08:39,800 --> 00:08:45,079
Speaker 1: fine tune this thing over time to a very sharp edge. Um.

137
00:08:45,160 --> 00:08:48,840
Speaker 1: So YouTube obsessively tracks everything that you watch, where you

138
00:08:48,840 --> 00:08:51,520
Speaker 1: watch it, exactly how far into the video you make it,

139
00:08:51,559 --> 00:08:54,040
Speaker 1: when you drop off, when you come back, and so

140
00:08:54,120 --> 00:08:56,360
Speaker 1: on and so on and so on. Um. And it's

141
00:08:56,400 --> 00:08:59,160
Speaker 1: this continuing effort to figure out exactly what you like.

142
00:08:59,320 --> 00:09:02,040
Speaker 1: And honestly, that's sort of the name of the game

143
00:09:02,200 --> 00:09:05,079
Speaker 1: these days with any of these streaming services. It's like um,

144
00:09:05,120 --> 00:09:10,280
Speaker 1: I think Netflix, Um actually crowdsourced a are offered a

145
00:09:10,360 --> 00:09:13,520
Speaker 1: prize to any outside coders that could figure out how

146
00:09:13,520 --> 00:09:17,160
Speaker 1: to help them get their algorithm for movie recommendations ten

147
00:09:17,600 --> 00:09:21,079
Speaker 1: more accurate, and they paid out to some outside coders

148
00:09:21,080 --> 00:09:22,559
Speaker 1: who did figure out how to do that thing. So

149
00:09:22,800 --> 00:09:26,600
Speaker 1: it's the more accurate your recommendations are, the more likely

150
00:09:26,679 --> 00:09:29,560
Speaker 1: people are to stay on the platform and not you know,

151
00:09:29,760 --> 00:09:34,160
Speaker 1: go outside, right. Yeah, absolutely, you can sandbox the much

152
00:09:34,200 --> 00:09:39,200
Speaker 1: more effectively, and ultimately, you know, the goal is to

153
00:09:39,240 --> 00:09:42,199
Speaker 1: get you to buy something. It's important to note that

154
00:09:43,040 --> 00:09:49,480
Speaker 1: almost every digital enterprise does something like this these days, Hulu, Netflix,

155
00:09:49,520 --> 00:09:53,520
Speaker 1: as you said, Amazon, eBay, Facebook, when you watch a

156
00:09:53,559 --> 00:09:57,559
Speaker 1: Facebook video, of course they do that. That's that data

157
00:09:57,720 --> 00:10:00,840
Speaker 1: is gold. And as we know, Reese years have proven

158
00:10:01,080 --> 00:10:03,720
Speaker 1: if you are not paying for a service, that means

159
00:10:03,960 --> 00:10:08,000
Speaker 1: you are the products right, absolutely, and even in this

160
00:10:08,360 --> 00:10:14,760
Speaker 1: strange sphere, YouTube occupies a unique area in the horizon

161
00:10:14,840 --> 00:10:17,440
Speaker 1: of digital entertainment. They're right on the edge and they're

162
00:10:17,520 --> 00:10:20,960
Speaker 1: super huge, so whenever they move there are going to

163
00:10:21,040 --> 00:10:25,720
Speaker 1: be waves and unforeseen ripple effects throughout the industry. There

164
00:10:25,720 --> 00:10:30,199
Speaker 1: are one point five billion YouTube users in the world today,

165
00:10:30,520 --> 00:10:34,120
Speaker 1: and that's more than the number of households that own televisions.

166
00:10:34,320 --> 00:10:37,560
Speaker 1: What they watch is shaped by this algorithm that you

167
00:10:37,679 --> 00:10:41,800
Speaker 1: just described NOL. It's automated largely. It's skims and ranks

168
00:10:42,120 --> 00:10:46,200
Speaker 1: billions of videos to identify twenty clips that will show

169
00:10:46,320 --> 00:10:49,520
Speaker 1: up in the right rail of the YouTube screen that

170
00:10:49,600 --> 00:10:52,400
Speaker 1: are called up next, and they're supposed to be relevant

171
00:10:52,440 --> 00:10:57,440
Speaker 1: to a previous video you just watched, and statistically, they're

172
00:10:57,480 --> 00:11:03,360
Speaker 1: supposed to be entire highly designed or entirely curated for

173
00:11:03,440 --> 00:11:08,439
Speaker 1: the sole purpose of keeping your eyes on the screen.

174
00:11:08,559 --> 00:11:10,679
Speaker 1: And they're very, very good at it. But there are

175
00:11:10,800 --> 00:11:14,960
Speaker 1: very very few public explanations about how this actually works.

176
00:11:15,000 --> 00:11:18,600
Speaker 1: It's a bit of a black box. YouTube engineers, in

177
00:11:18,840 --> 00:11:22,040
Speaker 1: some of their papers published on this describe it as

178
00:11:22,360 --> 00:11:26,679
Speaker 1: one of the largest scale and most sophisticated industrial recommendation

179
00:11:26,760 --> 00:11:31,320
Speaker 1: systems in existence. So in many ways, the story of

180
00:11:31,320 --> 00:11:35,960
Speaker 1: how YouTube figures out what you watch is a story

181
00:11:36,000 --> 00:11:41,720
Speaker 1: of incredible innovation and amazing breakthroughs in communication. However, according

182
00:11:41,760 --> 00:11:45,640
Speaker 1: to a growing body of critics, there's much much more

183
00:11:45,760 --> 00:11:50,520
Speaker 1: to the story, and you should be disturbed. Here's where

184
00:11:50,559 --> 00:11:54,400
Speaker 1: it gets crazy. You're gonna disturb me, Ben, I think

185
00:11:54,440 --> 00:11:57,079
Speaker 1: we'll all be disturbed here. You're gonna shake my worldview.

186
00:11:58,240 --> 00:12:01,440
Speaker 1: I don't know, but I might shake up your YouTube views.

187
00:12:01,600 --> 00:12:07,040
Speaker 1: Let's see. So first things first, this algorithm is proprietary,

188
00:12:07,200 --> 00:12:10,319
Speaker 1: and we know that it works hard stop. And it's

189
00:12:10,360 --> 00:12:15,280
Speaker 1: also constantly being tinkered with. The engineers and the coders

190
00:12:15,600 --> 00:12:19,520
Speaker 1: in that cadre we mentioned earlier are constantly going back

191
00:12:19,559 --> 00:12:22,720
Speaker 1: and saying, well, what if we just this value, or

192
00:12:22,720 --> 00:12:26,680
Speaker 1: what if we move what if we prioritize for instance, uh,

193
00:12:26,960 --> 00:12:30,600
Speaker 1: watch time or some other some other aspect of the

194
00:12:30,720 --> 00:12:34,160
Speaker 1: of the great formula. So the nuts and bolts of

195
00:12:34,200 --> 00:12:37,480
Speaker 1: this process are literally the stuff they don't want you

196
00:12:37,559 --> 00:12:40,400
Speaker 1: to know. YouTube does not want you to know the

197
00:12:40,480 --> 00:12:44,000
Speaker 1: intricacies of how this works. It's like a it's like

198
00:12:44,040 --> 00:12:47,600
Speaker 1: the Coca cola recipe. It's a secret formula. And that's

199
00:12:47,640 --> 00:12:50,720
Speaker 1: not necessarily a bad thing, not to be alarmist or anything.

200
00:12:50,760 --> 00:12:53,679
Speaker 1: It's it's not necessarily sinister. They just want to keep

201
00:12:53,720 --> 00:12:56,840
Speaker 1: their competitors from stealing their work or one upping them,

202
00:12:56,960 --> 00:13:00,840
Speaker 1: or they want to protect their jobs. But it does

203
00:13:00,960 --> 00:13:06,400
Speaker 1: lead us to ask, what keeps someone glued to the screen?

204
00:13:07,080 --> 00:13:10,840
Speaker 1: Is it? Is it something that makes them happy? Is

205
00:13:10,880 --> 00:13:15,080
Speaker 1: it just smiling children and someone softly playing acoustic guitar?

206
00:13:15,400 --> 00:13:20,600
Speaker 1: Sounds nice? It does right, Well, the algorithm has determined

207
00:13:20,600 --> 00:13:23,320
Speaker 1: the certain videos, or certain kinds of videos, will tend

208
00:13:23,360 --> 00:13:25,920
Speaker 1: to have a much higher likelihood of keeping people glued

209
00:13:25,960 --> 00:13:28,880
Speaker 1: to YouTube. But these are not nice videos for the

210
00:13:28,960 --> 00:13:33,600
Speaker 1: most part. In fact, they're they're divisive, they're downright someone

211
00:13:33,800 --> 00:13:38,640
Speaker 1: argue inappropriate, especially considering that it's pretty easy for a

212
00:13:38,720 --> 00:13:42,280
Speaker 1: kid to just watch whatever they want on YouTube. One

213
00:13:42,320 --> 00:13:43,800
Speaker 1: of the things we have to point out though, and

214
00:13:43,920 --> 00:13:49,120
Speaker 1: all it's so true YouTube loves videos about conspiracy theories. Yeah.

215
00:13:49,160 --> 00:13:51,880
Speaker 1: Have you ever been on YouTube and you're watching something

216
00:13:51,920 --> 00:13:53,480
Speaker 1: and then all of a sudden pops up with a

217
00:13:54,480 --> 00:13:58,680
Speaker 1: and Alex Jones or a flat earth thing? So nuts?

218
00:13:58,679 --> 00:14:00,559
Speaker 1: And it makes me wonder you We do a lot

219
00:14:00,559 --> 00:14:03,120
Speaker 1: of weird research here, but in some cases I felt

220
00:14:03,160 --> 00:14:06,480
Speaker 1: insulted when that stuff popped up. You know what's been

221
00:14:06,480 --> 00:14:10,440
Speaker 1: stalking me around YouTube recently. It's a how stuff works

222
00:14:10,520 --> 00:14:13,880
Speaker 1: video about corn. I don't know why did you have

223
00:14:13,880 --> 00:14:17,120
Speaker 1: anything to do with said video? Man? I don't think so. No,

224
00:14:17,480 --> 00:14:21,120
Speaker 1: it's just a video about corn. Interesting. I haven't watched it,

225
00:14:21,200 --> 00:14:27,320
Speaker 1: but maybe I'll cave. So this, this tendency of conspiracy

226
00:14:27,400 --> 00:14:30,760
Speaker 1: videos to get promoted on the trending or up next

227
00:14:31,160 --> 00:14:34,400
Speaker 1: AutoPlay kind of stuff has disturbed a lot of people

228
00:14:34,520 --> 00:14:38,680
Speaker 1: because these videos there seems to be a relationship between

229
00:14:38,720 --> 00:14:41,840
Speaker 1: how outlandish they are and how well they do with

230
00:14:41,880 --> 00:14:45,560
Speaker 1: this algorithm. YouTube came under fire a while back for

231
00:14:45,680 --> 00:14:50,040
Speaker 1: promoting videos claiming that the Las Vegas mass shooting was

232
00:14:50,080 --> 00:14:54,320
Speaker 1: a hoax because survivors and the families of the victims

233
00:14:54,640 --> 00:14:58,920
Speaker 1: were stunned and outraged. This was just a few days

234
00:14:59,000 --> 00:15:02,200
Speaker 1: after the shooting, which are curd at Mandalaid Bay. This

235
00:15:02,280 --> 00:15:07,400
Speaker 1: kind of proliferation concerns educators and media pundance because they

236
00:15:07,440 --> 00:15:10,360
Speaker 1: feel like people will fall into a sort of autopilot

237
00:15:10,440 --> 00:15:13,080
Speaker 1: and you know, instead of seeing a video pop up

238
00:15:13,080 --> 00:15:16,120
Speaker 1: and thinking I better check another source to see if

239
00:15:16,160 --> 00:15:20,440
Speaker 1: that's correct. Their concern is someone will read something about

240
00:15:22,160 --> 00:15:27,720
Speaker 1: the US capturing all guns and forcing people to use

241
00:15:27,840 --> 00:15:31,560
Speaker 1: gold for all financial transactions, and they won't go to

242
00:15:31,640 --> 00:15:34,320
Speaker 1: another source to check it. They'll just figure since it

243
00:15:34,400 --> 00:15:38,160
Speaker 1: popped up on YouTube, it must be true. I mean,

244
00:15:38,200 --> 00:15:42,040
Speaker 1: that's what I would assume. I'm kidding, but yeah, I mean,

245
00:15:42,040 --> 00:15:46,400
Speaker 1: they're definitely capitalizing on people's kind of gullibility here. So

246
00:15:46,960 --> 00:15:50,360
Speaker 1: what do you do gold when you're showing your kid

247
00:15:50,440 --> 00:15:53,880
Speaker 1: YouTube stop first she's watched it. Yeah, I don't know.

248
00:15:53,920 --> 00:15:56,080
Speaker 1: I'm gonna sort of peg myself here as not being

249
00:15:56,120 --> 00:15:58,360
Speaker 1: the most conscientious parent with that because I sort of like,

250
00:15:58,720 --> 00:16:00,840
Speaker 1: I kind of trust her to what she's doing, but

251
00:16:00,920 --> 00:16:04,840
Speaker 1: I know there's a lot of nefarious stuff going on there. Um.

252
00:16:04,920 --> 00:16:08,840
Speaker 1: She typically watches d I y videos and like gamer

253
00:16:08,840 --> 00:16:11,960
Speaker 1: type videos like Dan TDM and stuff like that, and

254
00:16:12,080 --> 00:16:14,640
Speaker 1: a lot of like kind of crafty videos um or

255
00:16:14,680 --> 00:16:19,000
Speaker 1: like some of those, Oh here's a satisfying um slice

256
00:16:19,000 --> 00:16:22,880
Speaker 1: of cake or whatever, you know. But she's so good

257
00:16:22,920 --> 00:16:27,320
Speaker 1: at self monitoring that I really don't worry about it.

258
00:16:27,400 --> 00:16:29,680
Speaker 1: But maybe maybe I'm gonna learn something in this episode

259
00:16:29,680 --> 00:16:33,040
Speaker 1: today that's gonna make me feel like I should. Yes, well,

260
00:16:33,280 --> 00:16:38,000
Speaker 1: we'll see forward into the breach, into the digital void.

261
00:16:38,280 --> 00:16:48,920
Speaker 1: After that is a word from our sponsors, and we're back. Yes,

262
00:16:49,040 --> 00:16:52,000
Speaker 1: there are controversies in the algorithm. There are people alleging

263
00:16:52,040 --> 00:16:54,720
Speaker 1: that there was a ghost in the machine. It gets

264
00:16:54,760 --> 00:17:00,400
Speaker 1: worse than just maybe a disrespectful video in the wake

265
00:17:00,480 --> 00:17:04,280
Speaker 1: of a tragedy. Some unidentified groups. And this is something

266
00:17:04,320 --> 00:17:07,120
Speaker 1: you and I have talked about. Off Ernol have conspired

267
00:17:07,160 --> 00:17:11,560
Speaker 1: to game YouTube's algorithm and hack things like YouTube Kids.

268
00:17:12,000 --> 00:17:18,840
Speaker 1: They're targeting children with um strange prank content. So there's

269
00:17:18,880 --> 00:17:23,320
Speaker 1: a character beloved by British children named Pepper Pig. About

270
00:17:23,359 --> 00:17:25,679
Speaker 1: Pepper Pig, I'm sure, yeah, you do, Okay, Well, tell

271
00:17:25,720 --> 00:17:28,040
Speaker 1: us a little bit about Pepper Pig. It's a two

272
00:17:28,040 --> 00:17:33,840
Speaker 1: dimensional British pig, British yep TV dimensional British pig. And

273
00:17:33,920 --> 00:17:37,840
Speaker 1: this this pig does the kinds of things you would

274
00:17:37,880 --> 00:17:40,480
Speaker 1: expect it to do on a children's show, but someone

275
00:17:40,520 --> 00:17:43,400
Speaker 1: went through and took advantage of the algorithm such that

276
00:17:43,640 --> 00:17:48,320
Speaker 1: children would watch maybe several clips, one to three clips

277
00:17:48,440 --> 00:17:51,080
Speaker 1: or so of Pepper Pig just on auto play, and

278
00:17:51,119 --> 00:17:53,280
Speaker 1: the next thing you know, it shows Pepper. There's a

279
00:17:53,280 --> 00:17:56,600
Speaker 1: new clip that shows Pepper Pig eating her father or

280
00:17:56,720 --> 00:18:00,480
Speaker 1: drinking bleach. This is very, very disturbed and stuff for

281
00:18:00,560 --> 00:18:03,239
Speaker 1: children to see. I'm actually texting with my kid right

282
00:18:03,280 --> 00:18:04,840
Speaker 1: now and I asked her to tell me about Pepper

283
00:18:04,920 --> 00:18:07,720
Speaker 1: Pig and uh, let's see if she's seen any of

284
00:18:07,760 --> 00:18:09,960
Speaker 1: these crazy Pepper Pig videos. But I don't think she has,

285
00:18:10,000 --> 00:18:13,360
Speaker 1: she would have said something about it. Well, we'll we'll

286
00:18:13,359 --> 00:18:16,760
Speaker 1: get an update in real time. Then, lud and violent

287
00:18:16,920 --> 00:18:20,120
Speaker 1: videos also do very well in this algorithm, or at

288
00:18:20,200 --> 00:18:23,760
Speaker 1: least they get a lot of support in the system.

289
00:18:23,800 --> 00:18:27,680
Speaker 1: There was one YouTube creator who was banned from monetizing

290
00:18:27,800 --> 00:18:32,400
Speaker 1: or making ad revenue from his strange videos because they

291
00:18:32,400 --> 00:18:35,800
Speaker 1: were disturbing things. There were videos of his children getting

292
00:18:35,840 --> 00:18:39,879
Speaker 1: flu shots, removing ear wax, crying because their pets died.

293
00:18:40,480 --> 00:18:43,960
Speaker 1: And this guy's defense was that he had only been

294
00:18:44,000 --> 00:18:49,679
Speaker 1: responding to the demands of the YouTube formula, the algorithm.

295
00:18:49,720 --> 00:18:51,879
Speaker 1: That's what got us out there in popular he said.

296
00:18:52,119 --> 00:18:54,359
Speaker 1: We learned to fuel it and do whatever it took

297
00:18:54,400 --> 00:18:57,560
Speaker 1: to please the algorithm. That's kind of weird, you know,

298
00:18:57,680 --> 00:19:01,640
Speaker 1: filming bad stuff happening to your children. Yeah, that's insane,

299
00:19:01,960 --> 00:19:05,200
Speaker 1: And we should stop for a second to talk about monetization.

300
00:19:06,200 --> 00:19:10,040
Speaker 1: People make videos that feed this algorithm because once those

301
00:19:10,119 --> 00:19:12,679
Speaker 1: videos are monetized, meaning they can slap and add on

302
00:19:12,720 --> 00:19:16,399
Speaker 1: the front of it, more views will equal more money eventually,

303
00:19:16,920 --> 00:19:20,520
Speaker 1: and research suggests that this creates a feedback loop of sorts,

304
00:19:20,760 --> 00:19:25,879
Speaker 1: with conspiratorial, divisive, sensational, and violent videos becoming the best

305
00:19:26,040 --> 00:19:28,520
Speaker 1: use of time for creators, at least in terms of

306
00:19:28,520 --> 00:19:33,000
Speaker 1: the bang for their buck clockwise. And up to this point,

307
00:19:33,840 --> 00:19:38,200
Speaker 1: we have just been discussing creeps on the internet, soulless

308
00:19:38,400 --> 00:19:43,400
Speaker 1: sky net like algorithms exploiting the same primal emotional responses

309
00:19:43,440 --> 00:19:47,240
Speaker 1: that make us slow down and gawk at car accidents.

310
00:19:47,840 --> 00:19:50,520
Speaker 1: But this is the disturbing part. This is a very

311
00:19:50,560 --> 00:19:53,440
Speaker 1: big butt. This is much more dangerous than a few

312
00:19:53,440 --> 00:19:58,960
Speaker 1: creeps trolling within appropriate cartoons. According to a former YouTube engineer,

313
00:19:59,000 --> 00:20:03,520
Speaker 1: a guy named yah Child's Lot, YouTube is something that

314
00:20:03,600 --> 00:20:06,800
Speaker 1: looks like reality but is distorted to make you spend

315
00:20:06,840 --> 00:20:10,800
Speaker 1: more time online, and he says the recommendation algorithm is

316
00:20:10,800 --> 00:20:14,280
Speaker 1: not optimizing for things that are truthful, or balanced or

317
00:20:14,320 --> 00:20:20,880
Speaker 1: healthy for democracy, and so when he left YouTube, officially

318
00:20:21,040 --> 00:20:23,439
Speaker 1: he just he wasn't up to snuff performance wise, but

319
00:20:23,560 --> 00:20:27,359
Speaker 1: he said he was pushed out because he was agitating

320
00:20:27,400 --> 00:20:32,439
Speaker 1: with other engineers to get a less dangerous algorithm or

321
00:20:32,440 --> 00:20:37,520
Speaker 1: to change what it values over just watch time. And

322
00:20:38,680 --> 00:20:43,359
Speaker 1: he went on to make his own efforts at transparency

323
00:20:43,440 --> 00:20:47,919
Speaker 1: to show the public how the YouTube algorithm works. He

324
00:20:48,000 --> 00:20:51,600
Speaker 1: believes that he's found shocking examples of this stuff going

325
00:20:51,800 --> 00:20:58,160
Speaker 1: very wrong. One was the run up to the presidential election.

326
00:20:58,320 --> 00:21:01,600
Speaker 1: Could you tell us a little bit about this? Absolutely so.

327
00:21:01,680 --> 00:21:05,280
Speaker 1: He actually observed in a short um relatively under the

328
00:21:05,359 --> 00:21:08,600
Speaker 1: radar blog spot post that was published after Trump was

329
00:21:08,640 --> 00:21:12,359
Speaker 1: elected um that the impact of YouTube's recommendation algorithm was

330
00:21:12,440 --> 00:21:16,240
Speaker 1: not a neutral thing at all during the presidential race.

331
00:21:16,280 --> 00:21:21,320
Speaker 1: In fact, it was pushing videos that were largely helpful

332
00:21:21,760 --> 00:21:25,640
Speaker 1: to the Trump campaign and damaging to the Clinton campaign.

333
00:21:26,040 --> 00:21:29,800
Speaker 1: He said. It was strange, Um, wherever you started, he said,

334
00:21:30,760 --> 00:21:33,440
Speaker 1: whether it was from a Trump search or a Clinton search,

335
00:21:33,480 --> 00:21:36,679
Speaker 1: the recommendation algorithm was much more likely to push you

336
00:21:37,240 --> 00:21:42,560
Speaker 1: in a pro Trump direction, which is fascinating because regardless

337
00:21:42,600 --> 00:21:46,520
Speaker 1: of our individual opinions for or against the particular candidate,

338
00:21:46,960 --> 00:21:49,919
Speaker 1: the problem here is not personally what anyone would vote for,

339
00:21:50,200 --> 00:21:52,280
Speaker 1: you know what I mean. It's bigger than that. It's

340
00:21:52,280 --> 00:21:54,919
Speaker 1: deeper than that. The problem here is that YouTube is

341
00:21:54,960 --> 00:21:59,800
Speaker 1: functioning under the illusion of neutrality, like an old television

342
00:22:00,000 --> 00:22:03,840
Speaker 1: that you are supposed to control the tube, but increasingly

343
00:22:03,880 --> 00:22:07,000
Speaker 1: it appears the opposite is true and the tube is

344
00:22:07,040 --> 00:22:11,480
Speaker 1: controlling you. The illusion neutrality goes hand in hand with

345
00:22:11,720 --> 00:22:18,160
Speaker 1: the illusion of agency, right or choice. This leads us

346
00:22:18,200 --> 00:22:22,119
Speaker 1: to some unanswered questions. We don't know what will change

347
00:22:22,200 --> 00:22:25,920
Speaker 1: or how we can check back in. Did did the

348
00:22:25,960 --> 00:22:28,440
Speaker 1: kiddo get back to you? She just said she hasn't

349
00:22:28,440 --> 00:22:30,440
Speaker 1: seen any of the creepy pepper pig videos. Oh well

350
00:22:30,440 --> 00:22:34,160
Speaker 1: that's great. Yes, that's crazy. So we have some morals

351
00:22:34,160 --> 00:22:37,320
Speaker 1: to the story. We have some unanswered questions. One, you

352
00:22:37,359 --> 00:22:40,119
Speaker 1: cannot let your kids run wild on YouTube. You have

353
00:22:40,160 --> 00:22:45,080
Speaker 1: to be very careful taking that one under. And we

354
00:22:45,160 --> 00:22:48,600
Speaker 1: also have to remember to discern fact from fiction. Just

355
00:22:48,600 --> 00:22:50,760
Speaker 1: because the flat Earth video pops up in your feet

356
00:22:50,800 --> 00:22:53,919
Speaker 1: doesn't make her flat. It just means you should double

357
00:22:54,040 --> 00:22:57,960
Speaker 1: check anything you see. And there's very very little chance

358
00:22:58,000 --> 00:23:00,679
Speaker 1: that YouTube will change the core functional of a system

359
00:23:00,720 --> 00:23:04,840
Speaker 1: anytime soon. There's just so much money on the table.

360
00:23:05,320 --> 00:23:09,200
Speaker 1: So it is up to us to change our own behavior,

361
00:23:09,359 --> 00:23:11,439
Speaker 1: or at least be more aware of the ways in

362
00:23:11,480 --> 00:23:16,680
Speaker 1: which we interact with this digital behemoth. Which I'm gonna

363
00:23:16,680 --> 00:23:19,800
Speaker 1: go ahead and say it all at all, you know,

364
00:23:19,840 --> 00:23:22,199
Speaker 1: taking account all the pros and all the cons. I

365
00:23:22,240 --> 00:23:25,680
Speaker 1: think YouTube is actually a great thing. But we want

366
00:23:25,720 --> 00:23:28,840
Speaker 1: to hear from you. Are we living in an increasingly

367
00:23:29,040 --> 00:23:35,640
Speaker 1: distorted reality? Are our decisions being guided by um algorithms

368
00:23:35,640 --> 00:23:37,560
Speaker 1: that may not be as fair as they would have

369
00:23:37,680 --> 00:23:41,040
Speaker 1: us believe. Or is the opposite the case. Is this

370
00:23:41,160 --> 00:23:43,280
Speaker 1: not a big deal, No one should worry about it.

371
00:23:43,720 --> 00:23:47,080
Speaker 1: Just skip the weird stuff. Let us know what you think. Yeah,

372
00:23:47,080 --> 00:23:49,520
Speaker 1: you can write to us on Facebook. We have a

373
00:23:49,560 --> 00:23:52,119
Speaker 1: private group they're called Here's where it Gets Crazy. Or

374
00:23:52,119 --> 00:23:53,480
Speaker 1: you can just go to the Facebook page where you

375
00:23:53,480 --> 00:23:55,640
Speaker 1: can find us at conspiracy Stuff, or if you want

376
00:23:55,640 --> 00:23:58,080
Speaker 1: to go to Instagram where a conspiracy stuff show. Twitter,

377
00:23:58,080 --> 00:24:01,040
Speaker 1: we're just conspiracy stuff, um. And we've got that that

378
00:24:01,200 --> 00:24:02,920
Speaker 1: handy dandy phone number that you can call in and

379
00:24:03,000 --> 00:24:05,080
Speaker 1: leave us some voicemails if you so choose. What's that

380
00:24:05,160 --> 00:24:08,080
Speaker 1: number billion and it's one eight three three st d

381
00:24:08,560 --> 00:24:12,159
Speaker 1: w y t K. Finally got in on it. That well,

382
00:24:12,200 --> 00:24:14,280
Speaker 1: you guys always doing and I never catch it in time,

383
00:24:14,600 --> 00:24:16,760
Speaker 1: and this time I was on the ball. There you

384
00:24:16,840 --> 00:24:20,199
Speaker 1: go on the ball. And if you would like to

385
00:24:20,240 --> 00:24:23,960
Speaker 1: hang out with us live, we have great news. We're

386
00:24:24,000 --> 00:24:27,479
Speaker 1: going on tour. We're hitting the road. Uh will this

387
00:24:27,520 --> 00:24:31,240
Speaker 1: episode be out while we're on tour? I'm not quite sure. Ben. Well,

388
00:24:31,480 --> 00:24:34,040
Speaker 1: if you're listening to this and it is after October.

389
00:24:35,359 --> 00:24:38,320
Speaker 1: Just find us. Go find a live show tab on

390
00:24:38,320 --> 00:24:40,600
Speaker 1: stuff they'll once you know dot com, and catch us

391
00:24:40,640 --> 00:24:43,000
Speaker 1: in a city near you. Please do and if then

392
00:24:43,080 --> 00:24:44,560
Speaker 1: that appeals to you, you can send us a good

393
00:24:44,560 --> 00:24:48,080
Speaker 1: old fashioned email where we are conspiracy at how stuff

394
00:24:48,080 --> 00:25:09,280
Speaker 1: Works dot com