WEBVTT

1
00:00:00.040 --> 00:00:03.600
<v Speaker 1>All right, ready to dive deep. Today we're tackling networks,

2
00:00:03.839 --> 00:00:05.799
<v Speaker 1>but not the social media kind, but we'll touch on

3
00:00:05.799 --> 00:00:10.080
<v Speaker 1>those two. We're talking about the structures underpinning practically everything,

4
00:00:10.359 --> 00:00:14.359
<v Speaker 1>internet genes, even how rumors spread. You've sent over a

5
00:00:14.400 --> 00:00:18.239
<v Speaker 1>ton on network connectivity. Great stuff, so, expert speaker, you're

6
00:00:18.280 --> 00:00:22.120
<v Speaker 1>with us. Why are these connections so important? Especially when

7
00:00:22.120 --> 00:00:23.199
<v Speaker 1>things get disrupted?

8
00:00:24.239 --> 00:00:27.879
<v Speaker 2>It's really quite amazing how vital these connections are. We're

9
00:00:27.879 --> 00:00:31.600
<v Speaker 2>talking about systems that, when disrupted, can have huge ripple

10
00:00:31.640 --> 00:00:34.240
<v Speaker 2>effects in ways we might not expect. A lot of

11
00:00:34.240 --> 00:00:36.679
<v Speaker 2>the source material jumps right into how we even measure

12
00:00:36.759 --> 00:00:38.840
<v Speaker 2>the strength and resilience of a network.

13
00:00:39.000 --> 00:00:40.719
<v Speaker 1>Yeah, and I have to admit at first it seemed

14
00:00:40.799 --> 00:00:43.679
<v Speaker 1>like whoa are we doing math today? Yeah? But it

15
00:00:43.679 --> 00:00:46.119
<v Speaker 1>makes sense. Different networks, different measuring sticks.

16
00:00:45.880 --> 00:00:48.920
<v Speaker 2>Right, absolutely, And the source material highlights a few key

17
00:00:48.960 --> 00:00:52.560
<v Speaker 2>ways to measure, each revealing something different about a network's resilience.

18
00:00:52.679 --> 00:00:54.759
<v Speaker 2>Like one that pops up is path capacity.

19
00:00:55.119 --> 00:00:59.240
<v Speaker 1>Okay, So like planning a road trip, more road options

20
00:00:59.280 --> 00:00:59.960
<v Speaker 1>the better.

21
00:01:00.520 --> 00:01:04.280
<v Speaker 2>Exactly, path capacity is all about redundancy. Your usual routes blocked?

22
00:01:04.280 --> 00:01:07.879
<v Speaker 2>How many alternate paths? More options? More resilient than network

23
00:01:08.040 --> 00:01:08.480
<v Speaker 2>makes sense.

24
00:01:08.519 --> 00:01:12.120
<v Speaker 1>More paths, more flexibility, less chance of a total meltdown.

25
00:01:12.400 --> 00:01:16.159
<v Speaker 1>So that's paths. But then there's this triangle capacity. Are

26
00:01:16.200 --> 00:01:17.560
<v Speaker 1>you talking like geometry here?

27
00:01:17.599 --> 00:01:20.959
<v Speaker 2>Not geometry? No, but it does get at how interconnected

28
00:01:21.000 --> 00:01:25.439
<v Speaker 2>a network is. Think of it as a measure of cliqueishness.

29
00:01:25.879 --> 00:01:30.359
<v Speaker 2>High triangle capacity, lots of interconnected groups, like everyone knows everyone.

30
00:01:30.560 --> 00:01:33.680
<v Speaker 1>Okay, but isn't that a double edged sword. Close connections

31
00:01:33.680 --> 00:01:36.159
<v Speaker 1>are strong but also potentially more vulnerable?

32
00:01:36.200 --> 00:01:40.760
<v Speaker 2>You got it? Think about misinformation spreading online. Those tightly

33
00:01:40.840 --> 00:01:45.079
<v Speaker 2>connected groups. They become like echo chambers, amplifying certain messages

34
00:01:45.120 --> 00:01:47.319
<v Speaker 2>even if they're not accurate, like a big game of

35
00:01:47.359 --> 00:01:50.280
<v Speaker 2>telephone with well potentially big consequences.

36
00:01:50.480 --> 00:01:54.400
<v Speaker 1>Powerful analogy. So we've got paths for flow these interconnected clusters.

37
00:01:54.400 --> 00:01:59.040
<v Speaker 1>There's also loop capacity in the source material, natural connectivity,

38
00:01:59.040 --> 00:01:59.840
<v Speaker 1>they call it too. What's that?

39
00:02:00.719 --> 00:02:04.159
<v Speaker 2>Loop capacity is about feedback, about recovery. It measures how

40
00:02:04.159 --> 00:02:07.400
<v Speaker 2>easily information can flow back to its source, forming a loop.

41
00:02:07.879 --> 00:02:11.680
<v Speaker 2>High loop capacity it suggests a network that can self correct,

42
00:02:12.080 --> 00:02:13.680
<v Speaker 2>bounce back from disruptions.

43
00:02:14.280 --> 00:02:16.400
<v Speaker 1>So we've got these different ways to measure a network.

44
00:02:16.400 --> 00:02:19.319
<v Speaker 1>It's paths, how interconnected it is, its ability to self correct.

45
00:02:19.319 --> 00:02:22.080
<v Speaker 1>But what about real world examples? Where does it get

46
00:02:22.199 --> 00:02:22.960
<v Speaker 1>really interesting?

47
00:02:23.080 --> 00:02:25.520
<v Speaker 2>Well, that's the thing about these measures, they're not just theoretical,

48
00:02:25.560 --> 00:02:28.840
<v Speaker 2>they're everywhere all around us. The power grid, for instance,

49
00:02:29.159 --> 00:02:31.800
<v Speaker 2>remember that massive blackout in the Northeast back in two

50
00:02:31.800 --> 00:02:33.879
<v Speaker 2>thousand and three, millions in the dark.

51
00:02:34.000 --> 00:02:35.479
<v Speaker 1>Yeah, talk about a domino effect.

52
00:02:35.560 --> 00:02:37.639
<v Speaker 2>One thing goes down exactly and it ripples through the

53
00:02:37.639 --> 00:02:41.120
<v Speaker 2>whole system. In that case, overloaded power lines, they tripped offline,

54
00:02:41.120 --> 00:02:44.719
<v Speaker 2>and because the grid lack redundancy, so low path capacity

55
00:02:44.759 --> 00:02:49.280
<v Speaker 2>that disruption it cascaded really quickly. That event really showed

56
00:02:49.280 --> 00:02:52.520
<v Speaker 2>how crucial path capacity is, especially for something as vital

57
00:02:52.520 --> 00:02:53.479
<v Speaker 2>as the power grid.

58
00:02:53.639 --> 00:02:56.520
<v Speaker 1>Right, keep the lights on. But then what about social networks?

59
00:02:56.520 --> 00:02:59.560
<v Speaker 1>How do these measures apply in like the online world?

60
00:03:00.080 --> 00:03:03.319
<v Speaker 2>About how trends go viral, or how news spreads online

61
00:03:03.439 --> 00:03:06.840
<v Speaker 2>or even frankly, misinformation, how fast it can all spread.

62
00:03:07.199 --> 00:03:09.879
<v Speaker 2>That's a direct result of how online networks are structured.

63
00:03:10.080 --> 00:03:13.159
<v Speaker 2>Often with that high triangle capacity we talked about, those

64
00:03:13.199 --> 00:03:16.919
<v Speaker 2>tightly knit groups, they can amplify certain messages really effectively,

65
00:03:17.319 --> 00:03:19.599
<v Speaker 2>which of course can have real world consequences.

66
00:03:19.800 --> 00:03:22.159
<v Speaker 1>It's like that game of telephone again. But on a

67
00:03:22.199 --> 00:03:23.800
<v Speaker 1>massive scale, precisely.

68
00:03:24.280 --> 00:03:27.000
<v Speaker 2>And it's not just social media either. Researchers are using

69
00:03:27.000 --> 00:03:29.840
<v Speaker 2>this kind of network analysis to understand all sorts of things,

70
00:03:29.840 --> 00:03:33.479
<v Speaker 2>how disease is spread, even how effective different drug treatments are.

71
00:03:33.840 --> 00:03:37.360
<v Speaker 1>Speaking of health, the source material mentions using networks for

72
00:03:37.879 --> 00:03:40.479
<v Speaker 1>like personalized medicine, how does that even work?

73
00:03:40.840 --> 00:03:44.479
<v Speaker 2>Imagine your genes as a network, right, each gene interacting

74
00:03:44.520 --> 00:03:48.120
<v Speaker 2>with others, and it's incredibly complex. But by analyzing these

75
00:03:48.159 --> 00:03:52.000
<v Speaker 2>genetic networks, researchers can start to identify individuals who might

76
00:03:52.039 --> 00:03:55.639
<v Speaker 2>respond differently to certain drugs or treatments, all based on

77
00:03:55.719 --> 00:03:57.120
<v Speaker 2>that unique genetic makeup.

78
00:03:57.319 --> 00:03:59.439
<v Speaker 1>So instead of one size fits all medicine, it's like

79
00:03:59.639 --> 00:04:03.479
<v Speaker 1>tailor to your personal genetic network. That's wild. But these

80
00:04:03.520 --> 00:04:07.080
<v Speaker 1>networks aren't static, right, They're constantly changing. How do we

81
00:04:07.199 --> 00:04:09.879
<v Speaker 1>keep up with those shifts, let alone predict how they'll

82
00:04:09.919 --> 00:04:11.400
<v Speaker 1>impact the network's connectivity.

83
00:04:11.479 --> 00:04:14.759
<v Speaker 2>That's a really important point. Networks are dynamic, They change

84
00:04:14.800 --> 00:04:17.439
<v Speaker 2>all the time, and that's where network inference comes in.

85
00:04:17.920 --> 00:04:20.600
<v Speaker 2>Is basically trying to predict the future of a network

86
00:04:20.639 --> 00:04:24.360
<v Speaker 2>by analyzing its past, like forecasting, but for networks instead

87
00:04:24.360 --> 00:04:28.040
<v Speaker 2>of weather exactly. Yeah, and the Source Material mentions an

88
00:04:28.079 --> 00:04:30.079
<v Speaker 2>algorithm called TRIP TRIP.

89
00:04:30.319 --> 00:04:31.879
<v Speaker 1>What does that stand for, Well.

90
00:04:31.720 --> 00:04:35.759
<v Speaker 2>It stands for tracking relative importance of paths. But what

91
00:04:35.800 --> 00:04:38.560
<v Speaker 2>it does is it's designed to track changes in a

92
00:04:38.639 --> 00:04:43.279
<v Speaker 2>network's eigenfunctions over time. Eigenfunctions, Yeah, it's a mouthful. Think

93
00:04:43.319 --> 00:04:46.560
<v Speaker 2>of them as a way to capture a network's essential characteristics,

94
00:04:47.000 --> 00:04:51.000
<v Speaker 2>its overall connectivity, its tendency to form clusters, how easily

95
00:04:51.079 --> 00:04:54.360
<v Speaker 2>information flows through all those things. And what TRIP does

96
00:04:54.480 --> 00:04:57.600
<v Speaker 2>is it analyzes how those characteristics change over time.

97
00:04:57.720 --> 00:05:00.480
<v Speaker 1>So it's like tracking the fingerprints of a network as it.

98
00:05:00.439 --> 00:05:04.720
<v Speaker 2>Evolves exactly, and that can help researchers understand how even

99
00:05:04.800 --> 00:05:07.959
<v Speaker 2>subtle shifts in a network might lead to big changes

100
00:05:07.959 --> 00:05:10.920
<v Speaker 2>and its behavior down the line, Like it might reveal

101
00:05:10.959 --> 00:05:14.040
<v Speaker 2>whether a power grid is becoming more vulnerable to blackouts,

102
00:05:14.319 --> 00:05:17.240
<v Speaker 2>or if a social network is becoming more prone to say,

103
00:05:17.639 --> 00:05:21.759
<v Speaker 2>the rapid spread of misinformation, all by analyzing these changes

104
00:05:21.800 --> 00:05:22.319
<v Speaker 2>over time.

105
00:05:22.600 --> 00:05:26.480
<v Speaker 1>Wow. So we've gone from paths and triangles to predicting

106
00:05:26.519 --> 00:05:29.240
<v Speaker 1>the future of networks. But it sounds like the Source

107
00:05:29.279 --> 00:05:31.839
<v Speaker 1>Material takes it even a step further with this idea

108
00:05:31.879 --> 00:05:33.959
<v Speaker 1>of multi layered networks, And.

109
00:05:33.800 --> 00:05:36.360
<v Speaker 2>This is where things get really interesting.

110
00:05:36.040 --> 00:05:39.120
<v Speaker 1>Multi layered networks. It's like that saying everything's connected, but

111
00:05:39.160 --> 00:05:41.399
<v Speaker 1>it's not just as saying, right, is how these complex

112
00:05:41.399 --> 00:05:43.319
<v Speaker 1>systems actually work exactly.

113
00:05:43.439 --> 00:05:48.000
<v Speaker 2>We're never dealing with just one network in isolation. Transportation, communication,

114
00:05:48.199 --> 00:05:52.560
<v Speaker 2>financial systems, even our personal relationships. It's all interconnected, these

115
00:05:52.639 --> 00:05:56.839
<v Speaker 2>intricate webs of interdependence. And the thing is, these multi

116
00:05:56.920 --> 00:05:59.560
<v Speaker 2>layered networks, they're often more fragile than we might.

117
00:05:59.480 --> 00:06:01.560
<v Speaker 1>Think, be kind of daunting when you think about it,

118
00:06:02.000 --> 00:06:05.360
<v Speaker 1>like one little hiccup in one network could snowball across

119
00:06:05.360 --> 00:06:07.720
<v Speaker 1>all these different layers. How do we even start to

120
00:06:07.720 --> 00:06:10.639
<v Speaker 1>get a handle on these systems, let alone predict where

121
00:06:10.639 --> 00:06:11.439
<v Speaker 1>the wheat points are.

122
00:06:11.800 --> 00:06:14.879
<v Speaker 2>That's where the really cutting edge research comes in. The

123
00:06:14.920 --> 00:06:18.920
<v Speaker 2>source material mentions this model called Moulan for a multi

124
00:06:19.000 --> 00:06:20.120
<v Speaker 2>layered network.

125
00:06:19.759 --> 00:06:21.519
<v Speaker 1>Analysis sounds complicated.

126
00:06:21.680 --> 00:06:25.439
<v Speaker 2>It is complex, yeah, but it's also really elegant. Basically,

127
00:06:25.480 --> 00:06:29.120
<v Speaker 2>it provides this framework for understanding how these different networks

128
00:06:29.120 --> 00:06:33.480
<v Speaker 2>interact and importantly for identifying those really critical nodes, the

129
00:06:33.480 --> 00:06:36.600
<v Speaker 2>ones that if their compromise, could take down the whole system.

130
00:06:36.720 --> 00:06:39.720
<v Speaker 1>So MEI lan's like what x ray vision for interconnected systems.

131
00:06:40.079 --> 00:06:42.879
<v Speaker 1>We can spot those potential points of failure before they

132
00:06:42.920 --> 00:06:43.959
<v Speaker 1>become a major problem.

133
00:06:44.120 --> 00:06:48.000
<v Speaker 2>That's the idea. By mapping out those dependencies between different layers,

134
00:06:48.040 --> 00:06:49.759
<v Speaker 2>like if this one thing goes down, how does it

135
00:06:49.800 --> 00:06:53.079
<v Speaker 2>impact everything else? And by looking at how information or

136
00:06:53.120 --> 00:06:56.680
<v Speaker 2>resources flow through the system, it can help pinpoint those

137
00:06:56.720 --> 00:06:57.920
<v Speaker 2>pressure points.

138
00:06:57.879 --> 00:07:02.279
<v Speaker 1>Which is huge for designing more resiltsllient systems, from infrastructure

139
00:07:02.319 --> 00:07:07.639
<v Speaker 1>to strategies for dealing with pandemics, misinformation, you name it,

140
00:07:08.319 --> 00:07:11.040
<v Speaker 1>anything that relies on these interconnected networks. So if we

141
00:07:11.120 --> 00:07:13.759
<v Speaker 1>know all of this about networks, how to analyze them,

142
00:07:13.879 --> 00:07:17.199
<v Speaker 1>how to spot their weaknesses, can we actually use that

143
00:07:17.319 --> 00:07:20.040
<v Speaker 1>to make them stronger, more resilient. Can we get ahead

144
00:07:20.079 --> 00:07:21.600
<v Speaker 1>of these potential catastrophes?

145
00:07:21.680 --> 00:07:25.399
<v Speaker 2>The million dollar question, right, and the source material does

146
00:07:25.480 --> 00:07:29.279
<v Speaker 2>offer some hope. Researchers are developing algorithms specifically designed to

147
00:07:29.319 --> 00:07:33.399
<v Speaker 2>do just that. Optimized network connectivity make them less susceptible

148
00:07:33.439 --> 00:07:35.279
<v Speaker 2>to disruptions in the first place.

149
00:07:35.319 --> 00:07:39.000
<v Speaker 1>Like that contain algorithm mentioned in one of the papers,

150
00:07:39.079 --> 00:07:42.160
<v Speaker 1>So it actually contains the damage like a firewall.

151
00:07:42.199 --> 00:07:43.480
<v Speaker 2>Yeah, it's a good way to think about it. I mean,

152
00:07:43.519 --> 00:07:46.759
<v Speaker 2>it's not an impenetrable barrier, but it's designed to minimize

153
00:07:46.800 --> 00:07:49.560
<v Speaker 2>the impact of a destruction. Think of it like having

154
00:07:49.560 --> 00:07:52.720
<v Speaker 2>an architect for your network, figuring out where to add redundancies,

155
00:07:52.720 --> 00:07:55.959
<v Speaker 2>how to strengthen those critical connections, where the weak points

156
00:07:56.000 --> 00:07:57.720
<v Speaker 2>are that need extra reinforcement.

157
00:07:57.959 --> 00:08:00.759
<v Speaker 1>So it's not just about reacting after something goes wrong,

158
00:08:00.879 --> 00:08:02.920
<v Speaker 1>but building in that resilience from.

159
00:08:02.800 --> 00:08:06.399
<v Speaker 2>The get go, exactly, and that has implications for everything

160
00:08:06.560 --> 00:08:09.879
<v Speaker 2>keeping the lights on, making sure our communication systems work,

161
00:08:10.160 --> 00:08:14.800
<v Speaker 2>maybe even influencing how diseases spread or how information flows online.

162
00:08:15.000 --> 00:08:17.600
<v Speaker 2>The possibilities are pretty amazing when you think about it,

163
00:08:17.680 --> 00:08:18.399
<v Speaker 2>they really are.

164
00:08:19.000 --> 00:08:22.319
<v Speaker 1>It's been a fascinating deep dive today expert speaker learning

165
00:08:22.360 --> 00:08:25.800
<v Speaker 1>about all these hidden networks that shape our world, everything

166
00:08:25.839 --> 00:08:29.160
<v Speaker 1>from the electricity powering our homes to like you said,

167
00:08:29.160 --> 00:08:32.519
<v Speaker 1>those whispers that can shape public opinion. Listener, I bet

168
00:08:32.559 --> 00:08:34.360
<v Speaker 1>you're seeing the world a little differently now, all these

169
00:08:34.399 --> 00:08:38.840
<v Speaker 1>invisible connections influencing everything around us. It really makes you

170
00:08:38.879 --> 00:08:42.759
<v Speaker 1>wonder what other hidden networks are out there, quietly shaping

171
00:08:42.799 --> 00:08:45.440
<v Speaker 1>the world we live in. Something to ponder, for sure.