ABOUT THE SPEAKER
Hannah Fry - Complexity theorist
Hannah Fry researches the trends in our civilization and ways we can forecast its future.

Why you should listen

Hannah Fry completed her PhD in fluid dynamics in early 2011 with an emphasis on how liquid droplets move. Then, after working as an aerodynamicist in the motorsport industry, she began work on an interdisciplinary project in complexity sciences at University College London. Hannah’s current research focusses on discovering new connections between mathematically described systems and human interaction at the largest scale.

More profile about the speaker
Hannah Fry | Speaker | TED.com
TEDxUCL

Hannah Fry: Is life really that complex?

Filmed:
819,007 views

Can an algorithm forecast the site of the next riot? In this accessible talk, mathematician Hannah Fry shows how complex social behavior can be analyzed and perhaps predicted through analogies to natural phenomena, like the patterns of a leopard's spots or the distribution of predators and prey in the wild.
- Complexity theorist
Hannah Fry researches the trends in our civilization and ways we can forecast its future. Full bio

Double-click the English transcript below to play the video.

00:10
Thanks very much.
0
24
1157
00:11
I am Hannah Fry, the badass.
1
1205
1848
00:13
And today I'm asking the question:
2
3077
1680
00:14
Is life really that complex?
3
4781
1756
00:16
Now, I've only got nine minutes
to try and provide you with an answer,
4
6561
3325
00:19
so what I've done
is split this neatly into two parts:
5
9910
2716
00:22
part one: yes;
6
12650
2353
00:25
and later on, part two: no.
7
15027
2528
00:27
Or, to be more accurate: no?
8
17579
2544
00:30
(Laughter)
9
20147
1204
00:31
So first of all, let me try and define
what I mean by "complex."
10
21375
3006
00:34
Now, I could give you
a host of formal definitions,
11
24405
2441
00:36
but in the simplest terms,
12
26870
1253
00:38
any problem in complexity is something
that Einstein and his peers can't do.
13
28147
4899
00:43
So, let's imagine --
if the clicker works ... there we go.
14
33070
3240
00:46
Einstein is playing a game of snooker.
15
36334
2103
00:48
He's a clever chap, so he knows
that when he hits the cue ball,
16
38461
3479
00:51
he could write you an equation
17
41964
1441
00:53
and tell you exactly where the red ball
is going to hit the sides,
18
43429
3128
00:56
how fast it's going
and where it's going to end up.
19
46581
2439
00:59
Now, if you scale these snooker balls
up to the size of the solar system,
20
49044
3469
01:02
Einstein can still help you.
21
52537
1959
01:04
Sure, the physics changes,
22
54520
1245
01:05
but if you wanted to know about
the path of the Earth around the Sun,
23
55789
3282
01:09
Einstein could write you an equation
24
59095
1733
01:10
telling you where both objects are
at any point in time.
25
60852
2643
01:13
Now, with a surprising
increase in difficulty,
26
63519
2204
01:15
Einstein could include
the Moon in his calculations.
27
65747
2452
01:18
But as you add more and more planets,
Mars and Jupiter, say,
28
68223
3067
01:21
the problem gets too tough for Einstein
to solve with a pen and paper.
29
71314
3764
01:25
Now, strangely, if instead of having
a handful of planets,
30
75102
2843
01:27
you had millions of objects
or even billions,
31
77969
2672
01:30
the problem actually becomes much simpler,
32
80665
2271
01:32
and Einstein is back in the game.
33
82960
1922
01:34
Let me explain what I mean by this,
34
84906
1846
01:36
by scaling these objects back down
to a molecular level.
35
86776
3294
01:40
If you wanted to trace the erratic path
of an individual air molecule,
36
90094
3747
01:43
you'd have absolutely no hope.
37
93865
1842
01:45
But when you have millions
of air molecules all together,
38
95731
2711
01:48
they start to act in a way
which is quantifiable, predictable
39
98466
3877
01:52
and well-behaved.
40
102367
1170
01:53
And thank goodness air is well-behaved,
41
103561
1885
01:55
because if it wasn't,
planes would fall out of the sky.
42
105470
2910
01:58
Now, on an even bigger scale,
across the whole of the world,
43
108404
3064
02:01
the idea is exactly the same
with all of these air molecules.
44
111492
3122
02:04
It's true that you can't take
an individual rain droplet
45
114638
2918
02:07
and say where it's come from
or where it's going to end up.
46
117580
2785
02:10
But you can say with pretty good certainty
47
120389
2034
02:12
whether it will be cloudy tomorrow.
48
122447
1813
02:14
So that's it.
49
124284
1158
02:15
In Einstein's time,
this is how far science had got.
50
125466
2683
02:18
We could do really small problems
with a few objects
51
128173
3613
02:21
with simple interactions,
52
131810
1275
02:23
or we could do huge problems
with millions of objects
53
133109
2602
02:25
and simple interactions.
54
135735
1323
02:27
But what about everything in the middle?
55
137082
1937
02:29
Well, just seven years
before Einstein's death,
56
139043
2693
02:31
an American scientist called
Warren Weaver made exactly this point.
57
141760
3658
02:35
He said that scientific methodology
has gone from one extreme to another,
58
145442
3624
02:39
leaving out an untouched
great middle region.
59
149090
3071
02:42
Now, this middle region
is where complexity science lies,
60
152185
2699
02:44
and this is what I mean by complex.
61
154908
2259
02:47
Now, unfortunately, almost
every single problem you can think of
62
157191
3525
02:50
to do with human behavior
63
160740
1222
02:51
lies in this middle region.
64
161986
2069
02:54
Einstein's got absolutely no idea
how to model the movement of a crowd.
65
164079
4290
02:58
There are too many people
to look at them all individually
66
168393
2801
03:01
and too few to treat them as a gas.
67
171218
1872
03:03
Similarly, people are prone
to annoying things like decisions
68
173114
3356
03:06
and not wanting to walk into each other,
69
176494
2014
03:08
which makes the problem
all the more complicated.
70
178532
2613
03:11
Einstein also couldn't tell you
71
181169
1703
03:12
when the next stock market crash
is going to be.
72
182896
2349
03:15
Einstein couldn't tell you
how to improve unemployment.
73
185269
2764
03:18
Einstein can't even tell you
74
188057
1401
03:19
whether the next iPhone
is going to be a hit or a flop.
75
189482
3382
03:22
So to conclude part one:
we're completely screwed.
76
192888
2606
03:25
We've got no tools to deal with this,
and life is way too complex.
77
195518
4500
03:30
But maybe there's hope,
78
200042
1796
03:31
because in the last few years,
79
201862
1534
03:33
we've begun to see the beginnings
of a new area of science
80
203420
3837
03:37
using mathematics
to model our social systems.
81
207281
3027
03:40
And I'm not just talking here
about statistics and computer simulations.
82
210332
3484
03:43
I'm talking about writing down
equations about our society
83
213840
2996
03:46
that will help us understand
what's going on
84
216860
2069
03:48
in the same way as with the snooker balls
or the weather prediction.
85
218953
3215
03:52
And this has come about
because people have begun to realize
86
222192
2842
03:55
that we can use and exploit analogies
87
225058
2343
03:57
between our human systems
and those of the physical world around us.
88
227425
3942
04:01
Now, to give you an example:
89
231938
1464
04:03
the incredibly complex problem
of migration across Europe.
90
233426
3454
04:06
Actually, as it turns out, when you view
all of the people together,
91
236904
3332
04:10
collectively, they behave as though
they're following the laws of gravity.
92
240260
4043
04:14
But instead of planets
being attracted to one another,
93
244327
3128
04:17
it's people who are attracted
to areas with better job opportunities,
94
247479
4210
04:21
higher pay, better quality of life
and lower unemployment.
95
251713
4015
04:25
And in the same way as people
are more likely to go for opportunities
96
255752
3528
04:29
close to where they live already --
London to Kent, for example,
97
259304
3035
04:32
as opposed to London to Melbourne --
98
262363
1792
04:34
the gravitational effect of planets
far away is felt much less.
99
264179
4136
04:38
So, to give you another example:
100
268997
2067
04:41
in 2008, a group in UCLA
were looking into the patterns
101
271088
4225
04:45
of burglary hot spots in the city.
102
275337
2712
04:48
Now, one thing about burglaries
is this idea of repeat victimization.
103
278073
5519
04:53
So if you have a group of burglars
who manage to successfully rob an area,
104
283616
4237
04:57
they'll tend to return to that area
and carry on burgling it.
105
287877
3790
05:01
So they learn the layout of the houses,
106
291691
2856
05:04
the escape routes
107
294571
1694
05:06
and the local security measures
that are in place.
108
296289
3004
05:09
And this will continue to happen
109
299317
1685
05:11
until local residents and police
ramp up the security,
110
301026
3181
05:14
at which point, the burglars
will move off elsewhere.
111
304231
2771
05:17
And it's that balance
between burglars and security
112
307026
2808
05:19
which creates these dynamic
hot spots of the city.
113
309858
3037
05:22
As it turns out,
this is exactly the same process
114
312919
3544
05:26
as how a leopard gets its spots,
115
316487
2242
05:28
except in the leopard example,
it's not burglars and security,
116
318753
2936
05:31
it's the chemical process
that creates these patterns
117
321713
3465
05:35
and something called "morphogenesis."
118
325202
1995
05:37
We actually know an awful lot
about the morphogenesis of leopard spots.
119
327221
4256
05:41
Maybe we can use this to try and spot
some of the warning signs with burglaries
120
331501
4644
05:46
and perhaps, also to create
better crime strategies to prevent crime.
121
336169
4107
05:50
There's a group here at UCL
122
340300
1572
05:51
who are working with
the West Midlands police right now
123
341896
2825
05:54
on this very question.
124
344745
1641
05:56
I could give you
plenty of examples like this,
125
346410
2915
05:59
but I wanted to leave you
with one from my own research
126
349349
2643
06:02
on the London riots.
127
352016
1166
06:03
Now, you probably
don't need me to tell you
128
353206
2015
06:05
about the events of last summer,
129
355245
1567
06:06
where London and the UK saw
the worst sustained period
130
356836
3030
06:09
of violent looting and arson
131
359890
1526
06:11
for over twenty years.
132
361440
1613
06:13
It's understandable that, as a society,
we want to try and understand
133
363077
3287
06:16
exactly what caused these riots,
134
366388
1794
06:18
but also, perhaps, to equip our police
with better strategies
135
368206
3885
06:22
to lead to a swifter
resolution in the future.
136
372115
3781
06:25
Now, I don't want to upset
the sociologists here,
137
375920
2356
06:28
so I absolutely cannot talk about
the individual motivations for a rioter,
138
378300
4857
06:33
but when you look at
the rioters all together,
139
383181
2168
06:35
mathematically, you can separate it
into a three-stage process
140
385373
3208
06:38
and draw analogies accordingly.
141
388605
1975
06:40
So, step one: let's say
you've got a group of friends.
142
390604
3177
06:43
None of them are involved in the riots,
143
393805
1875
06:45
but one of them walks past
a Foot Locker which is being raided,
144
395704
3682
06:49
and goes in and bags himself
a new pair of trainers.
145
399410
2513
06:51
He texts one of his friends and says,
"Come on down to the riots."
146
401947
4089
06:56
So his friend joins him,
147
406060
1421
06:57
and then the two of them text
more of their friends, who join them,
148
407505
3157
07:00
and text more of their friends
149
410686
1581
07:02
and more and more, and so it continues.
150
412291
2374
07:04
This process is identical to the way
that a virus spreads through a population.
151
414689
4583
07:09
If you think about the bird flu epidemic
of a couple of years ago,
152
419296
3100
07:12
the more people that were infected,
the more people that got infected,
153
422420
3303
07:15
and the faster the virus spread
154
425747
1588
07:17
before the authorities managed
to get a handle on events.
155
427359
3141
07:20
And it's exactly the same process here.
156
430988
2515
07:23
So let's say you've got a rioter,
he's decided he's going to riot.
157
433527
3276
07:26
The next thing he has to do
is pick a riot site.
158
436827
2535
07:30
Now, what you should know
about rioters is that, um ...
159
440274
3624
07:33
Oops, clicker's gone. There we go.
160
443922
1642
07:35
What you should know about rioters is,
they're not prepared to travel
161
445588
3344
07:38
that far from where they live,
162
448956
1451
07:40
unless it's a really juicy riot site.
163
450431
1852
07:42
(Laughter)
164
452307
1075
07:43
So you can see that here from this graph,
165
453406
2069
07:45
with an awful lot of rioters
having traveled less than a kilometer
166
455499
3391
07:48
to the site that they went to.
167
458914
1679
07:50
Now, this pattern is seen
in consumer models of retail spending,
168
460617
4909
07:55
i.e., where we choose to go shopping.
169
465550
2309
07:57
So, of course, people like
to go to local shops,
170
467883
2922
08:00
but you'd be prepared
to go a little bit further
171
470829
2592
08:03
if it was a really good retail site.
172
473445
2116
08:05
And this analogy, actually, was already
picked up by some of the papers,
173
475585
3442
08:09
with some tabloid press calling the events
"Shopping with violence,"
174
479051
3262
08:12
which probably sums it up
in terms of our research.
175
482337
2788
08:15
Oh! -- we're going backwards.
176
485673
1476
08:19
OK, step three.
177
489730
1456
08:21
Finally, the rioter is at his site,
178
491210
1817
08:23
and he wants to avoid
getting caught by the police.
179
493051
4572
08:27
The rioters will avoid
the police at all times,
180
497647
2701
08:30
but there is some safety in numbers.
181
500372
2094
08:32
And on the flip side, the police,
with their limited resources,
182
502490
3061
08:35
are trying to protect
as much of the city as possible,
183
505575
2579
08:38
arrest rioters wherever possible
184
508178
2013
08:40
and to create a deterrent effect.
185
510215
2041
08:45
And actually, as it turns out,
186
515510
1491
08:47
this mechanism between the two species,
so to speak, of rioters and police,
187
517025
4623
08:51
is identical to predators
and prey in the wild.
188
521672
2649
08:54
So if you can imagine rabbits and foxes,
189
524345
2197
08:56
rabbits are trying to avoid
foxes at all costs,
190
526566
2750
08:59
while foxes are patrolling the space,
trying to look for rabbits.
191
529340
3687
09:03
We actually know an awful lot
about the dynamics of predators and prey.
192
533051
3354
09:06
We also know a lot about
consumer spending flows.
193
536429
4979
09:11
And we know a lot about
how viruses spread through a population.
194
541432
3163
09:14
So if you take these three analogies
together and exploit them,
195
544619
3033
09:17
you can come up with a mathematical
model of what actually happened,
196
547676
3236
09:20
that's capable of replicating
the general patterns
197
550936
2404
09:23
of the riots themselves.
198
553364
1343
09:25
Now, once we've got this,
we can almost use this as a petri dish
199
555678
3086
09:28
and start having conversations
200
558788
1623
09:30
about which areas of the city
were more susceptible than others
201
560435
3139
09:33
and what police tactics could be used
202
563598
1877
09:35
if this were ever to happen
again in the future.
203
565499
2307
09:37
Even twenty years ago, modeling
of this sort was completely unheard of.
204
567830
4003
09:41
But I think that these analogies
are an incredibly important tool
205
571857
4444
09:46
in tackling problems with our society,
206
576325
2491
09:48
and perhaps, ultimately improving
our society overall.
207
578840
3406
09:52
So, to conclude: life is complex,
208
582270
2389
09:54
but perhaps understanding it need not
necessarily be that complicated.
209
584683
3357
09:58
Thank you.
210
588064
1158
09:59
(Applause)
211
589246
1386

▲Back to top

ABOUT THE SPEAKER
Hannah Fry - Complexity theorist
Hannah Fry researches the trends in our civilization and ways we can forecast its future.

Why you should listen

Hannah Fry completed her PhD in fluid dynamics in early 2011 with an emphasis on how liquid droplets move. Then, after working as an aerodynamicist in the motorsport industry, she began work on an interdisciplinary project in complexity sciences at University College London. Hannah’s current research focusses on discovering new connections between mathematically described systems and human interaction at the largest scale.

More profile about the speaker
Hannah Fry | Speaker | TED.com