TEDxSydney
Tom Griffiths: 3 ways to make better decisions -- by thinking like a computer
Filmed:
Readability: 4.2
3,652,976 views
If you ever struggle to make decisions, here's a talk for you. Cognitive scientist Tom Griffiths shows how we can apply the logic of computers to untangle tricky human problems, sharing three practical strategies for making better decisions -- on everything from finding a home to choosing which restaurant to go to tonight.
Tom Griffiths - Psychologist, cognitive scientist
Tom Griffiths uses ideas from computer science to understand how human minds work. Full bio
Tom Griffiths uses ideas from computer science to understand how human minds work. Full bio
Double-click the English transcript below to play the video.
00:13
If there's one city in the world
0
1407
1581
00:15
where it's hard to find
a place to buy or rent,
a place to buy or rent,
1
3012
2333
00:17
it's Sydney.
2
5369
1150
00:19
And if you've tried
to find a home here recently,
to find a home here recently,
3
7043
2367
00:21
you're familiar with the problem.
4
9434
1840
00:23
Every time you walk into an open house,
5
11298
2014
00:25
you get some information
about what's out there
about what's out there
6
13336
2191
00:27
and what's on the market,
7
15551
1393
00:28
but every time you walk out,
8
16968
1430
00:30
you're running the risk
of the very best place passing you by.
of the very best place passing you by.
9
18422
3214
00:33
So how do you know when
to switch from looking
to switch from looking
10
21660
2820
00:36
to being ready to make an offer?
11
24504
1562
00:39
This is such a cruel and familiar problem
12
27663
2401
00:42
that it might come as a surprise
that it has a simple solution.
that it has a simple solution.
13
30088
3074
00:45
37 percent.
14
33588
1203
00:46
(Laughter)
15
34815
1828
00:48
If you want to maximize the probability
that you find the very best place,
that you find the very best place,
16
36667
3936
00:52
you should look at 37 percent
of what's on the market,
of what's on the market,
17
40627
2825
00:55
and then make an offer
on the next place you see,
on the next place you see,
18
43476
2310
00:57
which is better than anything
that you've seen so far.
that you've seen so far.
19
45810
2524
01:00
Or if you're looking for a month,
take 37 percent of that time --
take 37 percent of that time --
20
48358
3805
01:04
11 days, to set a standard --
21
52187
2915
01:07
and then you're ready to act.
22
55126
1575
01:09
We know this because
trying to find a place to live
trying to find a place to live
23
57829
2709
01:12
is an example of an optimal
stopping problem.
stopping problem.
24
60562
2325
01:14
A class of problems that has been
studied extensively
studied extensively
25
62911
2483
01:17
by mathematicians and computer scientists.
26
65418
2105
01:21
I'm a computational cognitive scientist.
27
69502
2519
01:24
I spend my time trying to understand
28
72045
1960
01:26
how it is that human minds work,
29
74029
1798
01:27
from our amazing successes
to our dismal failures.
to our dismal failures.
30
75851
3671
01:32
To do that, I think about
the computational structure
the computational structure
31
80552
2493
01:35
of the problems
that arise in everyday life,
that arise in everyday life,
32
83069
2659
01:37
and compare the ideal
solutions to those problems
solutions to those problems
33
85752
2375
01:40
to the way that we actually behave.
34
88151
1747
01:42
As a side effect,
35
90725
1185
01:43
I get to see how applying
a little bit of computer science
a little bit of computer science
36
91934
2715
01:46
can make human decision-making easier.
37
94673
1860
01:49
I have a personal motivation for this.
38
97828
1847
01:52
Growing up in Perth
as an overly cerebral kid ...
as an overly cerebral kid ...
39
100153
3166
01:55
(Laughter)
40
103343
4731
02:00
I would always try and act in the way
that I thought was rational,
that I thought was rational,
41
108098
3147
02:03
reasoning through every decision,
42
111269
1597
02:04
trying to figure out
the very best action to take.
the very best action to take.
43
112890
2403
02:07
But this is an approach
that doesn't scale up
that doesn't scale up
44
115972
2190
02:10
when you start to run into
the sorts of problems
the sorts of problems
45
118186
2242
02:12
that arise in adult life.
46
120452
1500
02:13
At one point, I even tried
to break up with my girlfriend
to break up with my girlfriend
47
121976
2728
02:16
because trying to take into account
her preferences as well as my own
her preferences as well as my own
48
124728
3350
02:20
and then find perfect solutions --
49
128102
1683
02:21
(Laughter)
50
129809
2259
02:24
was just leaving me exhausted.
51
132092
1752
02:25
(Laughter)
52
133868
2533
02:28
She pointed out that I was taking
the wrong approach
the wrong approach
53
136425
2429
02:30
to solving this problem --
54
138878
1609
02:32
and she later became my wife.
55
140511
1382
02:33
(Laughter)
56
141917
2062
02:36
(Applause)
57
144003
4971
02:40
Whether it's as basic as trying to decide
what restaurant to go to
what restaurant to go to
58
148998
3461
02:44
or as important as trying to decide
who to spend the rest of your life with,
who to spend the rest of your life with,
59
152483
4055
02:48
human lives are filled
with computational problems
with computational problems
60
156562
2375
02:50
that are just too hard to solve
by applying sheer effort.
by applying sheer effort.
61
158961
4196
02:55
For those problems,
62
163650
1166
02:56
it's worth consulting the experts:
63
164840
1961
02:58
computer scientists.
64
166825
1151
03:00
(Laughter)
65
168000
1784
03:01
When you're looking for life advice,
66
169808
1819
03:03
computer scientists probably aren't
the first people you think to talk to.
the first people you think to talk to.
67
171651
3640
03:07
Living life like a computer --
68
175315
1875
03:09
stereotypically deterministic,
exhaustive and exact --
exhaustive and exact --
69
177214
2578
03:11
doesn't sound like a lot of fun.
70
179816
1552
03:14
But thinking about the computer science
of human decisions
of human decisions
71
182153
2927
03:17
reveals that in fact,
we've got this backwards.
we've got this backwards.
72
185104
2408
03:19
When applied to the sorts
of difficult problems
of difficult problems
73
187536
2198
03:21
that arise in human lives,
74
189758
1239
03:23
the way that computers
actually solve those problems
actually solve those problems
75
191021
2727
03:25
looks a lot more like the way
that people really act.
that people really act.
76
193772
2896
03:29
Take the example of trying to decide
what restaurant to go to.
what restaurant to go to.
77
197267
2922
03:33
This is a problem that has
a particular computational structure.
a particular computational structure.
78
201090
3088
03:36
You've got a set of options,
79
204202
1524
03:37
you're going to choose
one of those options,
one of those options,
80
205750
2049
03:39
and you're going to face
exactly the same decision tomorrow.
exactly the same decision tomorrow.
81
207823
2811
03:42
In that situation,
82
210658
1151
03:43
you run up against
what computer scientists call
what computer scientists call
83
211833
2295
03:46
the "explore-exploit trade-off."
84
214152
2372
03:49
You have to make a decision
85
217115
1334
03:50
about whether you're going
to try something new --
to try something new --
86
218473
2334
03:52
exploring, gathering some information
87
220831
2183
03:55
that you might be able
to use in the future --
to use in the future --
88
223038
2285
03:57
or whether you're going to go to a place
that you already know is pretty good --
that you already know is pretty good --
89
225347
3793
04:01
exploiting the information
that you've already gathered so far.
that you've already gathered so far.
90
229164
3003
04:05
The explore/exploit trade-off
shows up any time you have to choose
shows up any time you have to choose
91
233488
3096
04:08
between trying something new
92
236608
1350
04:09
and going with something
that you already know is pretty good,
that you already know is pretty good,
93
237982
2980
04:12
whether it's listening to music
94
240986
1550
04:14
or trying to decide
who you're going to spend time with.
who you're going to spend time with.
95
242560
2634
04:17
It's also the problem
that technology companies face
that technology companies face
96
245218
2494
04:19
when they're trying to do something
like decide what ad to show on a web page.
like decide what ad to show on a web page.
97
247736
3667
04:23
Should they show a new ad
and learn something about it,
and learn something about it,
98
251417
2572
04:26
or should they show you an ad
99
254003
1383
04:27
that they already know there's a good
chance you're going to click on?
chance you're going to click on?
100
255420
3306
04:30
Over the last 60 years,
101
258750
1151
04:31
computer scientists have made
a lot of progress understanding
a lot of progress understanding
102
259925
2892
04:34
the explore/exploit trade-off,
103
262841
1452
04:36
and their results offer
some surprising insights.
some surprising insights.
104
264317
2398
04:39
When you're trying to decide
what restaurant to go to,
what restaurant to go to,
105
267291
2598
04:41
the first question you should ask yourself
106
269913
2000
04:43
is how much longer
you're going to be in town.
you're going to be in town.
107
271937
2179
04:46
If you're just going to be there
for a short time,
for a short time,
108
274505
2342
04:48
then you should exploit.
109
276871
1515
04:50
There's no point gathering information.
110
278410
1905
04:52
Just go to a place
you already know is good.
you already know is good.
111
280339
2048
04:54
But if you're going to be there
for a longer time, explore.
for a longer time, explore.
112
282411
2929
04:57
Try something new,
because the information you get
because the information you get
113
285364
2387
04:59
is something that can improve
your choices in the future.
your choices in the future.
114
287775
2974
05:02
The value of information increases
115
290773
1979
05:04
the more opportunities
you're going to have to use it.
you're going to have to use it.
116
292776
2600
05:08
This principle can give us insight
117
296193
1754
05:09
into the structure
of a human life as well.
of a human life as well.
118
297971
2070
05:13
Babies don't have a reputation
for being particularly rational.
for being particularly rational.
119
301104
3053
05:17
They're always trying new things,
120
305098
1825
05:18
and you know, trying to stick them
in their mouths.
in their mouths.
121
306947
2836
05:22
But in fact, this is exactly
what they should be doing.
what they should be doing.
122
310621
2952
05:25
They're in the explore
phase of their lives,
phase of their lives,
123
313597
2430
05:28
and some of those things
could turn out to be delicious.
could turn out to be delicious.
124
316051
2621
05:32
At the other end of the spectrum,
125
320040
1572
05:33
the old guy who always goes
to the same restaurant
to the same restaurant
126
321636
2506
05:36
and always eats the same thing
127
324166
1635
05:37
isn't boring --
128
325825
1328
05:39
he's optimal.
129
327177
1509
05:40
(Laughter)
130
328710
3830
05:44
He's exploiting the knowledge
that he's earned
that he's earned
131
332564
2228
05:46
through a lifetime's experience.
132
334816
1767
05:50
More generally,
133
338406
1151
05:51
knowing about
the explore/exploit trade-off
the explore/exploit trade-off
134
339581
2031
05:53
can make it a little easier for you
to sort of relax and go easier on yourself
to sort of relax and go easier on yourself
135
341636
3667
05:57
when you're trying to make a decision.
136
345327
1810
05:59
You don't have to go
to the best restaurant every night.
to the best restaurant every night.
137
347161
2728
06:01
Take a chance, try something new, explore.
138
349913
2979
06:04
You might learn something.
139
352916
1627
06:06
And the information that you gain
140
354567
1589
06:08
is going to be worth more
than one pretty good dinner.
than one pretty good dinner.
141
356180
2543
06:12
Computer science can also help
to make it easier on us
to make it easier on us
142
360178
2699
06:14
in other places at home and in the office.
143
362901
2152
06:17
If you've ever had
to tidy up your wardrobe,
to tidy up your wardrobe,
144
365860
2453
06:20
you've run into a particularly
agonizing decision:
agonizing decision:
145
368337
2671
06:23
you have to decide what things
you're going to keep
you're going to keep
146
371032
2382
06:25
and what things you're going to give away.
147
373438
2009
06:27
Martha Stewart turns out
to have thought very hard about this --
to have thought very hard about this --
148
375974
3000
06:30
(Laughter)
149
378998
1205
06:32
and she has some good advice.
150
380227
1678
06:33
She says, "Ask yourself four questions:
151
381929
2295
06:36
How long have I had it?
152
384248
1526
06:37
Does it still function?
153
385798
1450
06:39
Is it a duplicate
of something that I already own?
of something that I already own?
154
387272
3199
06:42
And when was the last time
I wore it or used it?"
I wore it or used it?"
155
390495
2411
06:46
But there's another group of experts
156
394648
1715
06:48
who perhaps thought
even harder about this problem,
even harder about this problem,
157
396387
3130
06:51
and they would say one of these questions
is more important than the others.
is more important than the others.
158
399541
3660
06:55
Those experts?
159
403819
1150
06:57
The people who design
the memory systems of computers.
the memory systems of computers.
160
405281
2848
07:00
Most computers have
two kinds of memory systems:
two kinds of memory systems:
161
408153
2254
07:02
a fast memory system,
162
410431
1387
07:03
like a set of memory chips
that has limited capacity,
that has limited capacity,
163
411842
3169
07:07
because those chips are expensive,
164
415035
2052
07:09
and a slow memory system,
which is much larger.
which is much larger.
165
417111
3260
07:13
In order for the computer to operate
as efficiently as possible,
as efficiently as possible,
166
421239
3000
07:16
you want to make sure
167
424263
1151
07:17
that the pieces of information
you want to access
you want to access
168
425438
2352
07:19
are in the fast memory system,
169
427814
1477
07:21
so that you can get to them quickly.
170
429315
1754
07:23
Each time you access
a piece of information,
a piece of information,
171
431093
2061
07:25
it's loaded into the fast memory
172
433178
1525
07:26
and the computer has to decide which item
it has to remove from that memory,
it has to remove from that memory,
173
434727
3877
07:30
because it has limited capacity.
174
438628
1741
07:33
Over the years,
175
441794
1151
07:34
computer scientists have tried
a few different strategies
a few different strategies
176
442969
2739
07:37
for deciding what to remove
from the fast memory.
from the fast memory.
177
445732
2292
07:40
They've tried things like choosing
something at random
something at random
178
448048
3176
07:43
or applying what's called
the "first-in, first-out principle,"
the "first-in, first-out principle,"
179
451248
2906
07:46
which means removing the item
180
454178
1399
07:47
which has been in the memory
for the longest.
for the longest.
181
455601
2176
07:50
But the strategy that's most effective
182
458091
2713
07:52
focuses on the items
which have been least recently used.
which have been least recently used.
183
460828
3229
07:56
This says if you're going to decide
to remove something from memory,
to remove something from memory,
184
464972
3191
08:00
you should take out the thing which was
last accessed the furthest in the past.
last accessed the furthest in the past.
185
468187
4968
08:05
And there's a certain
kind of logic to this.
kind of logic to this.
186
473179
2159
08:07
If it's been a long time since you last
accessed that piece of information,
accessed that piece of information,
187
475362
3598
08:10
it's probably going to be a long time
188
478984
1762
08:12
before you're going to need
to access it again.
to access it again.
189
480770
2192
08:15
Your wardrobe is just like
the computer's memory.
the computer's memory.
190
483787
2774
08:18
You have limited capacity,
191
486585
2030
08:20
and you need to try and get in there
the things that you're most likely to need
the things that you're most likely to need
192
488639
5327
08:25
so that you can get to them
as quickly as possible.
as quickly as possible.
193
493990
2419
08:29
Recognizing that,
194
497016
1151
08:30
maybe it's worth applying
the least recently used principle
the least recently used principle
195
498191
2813
08:33
to organizing your wardrobe as well.
196
501028
1949
08:35
So if we go back
to Martha's four questions,
to Martha's four questions,
197
503001
2082
08:37
the computer scientists
would say that of these,
would say that of these,
198
505107
2861
08:39
the last one is the most important.
199
507992
1941
08:43
This idea of organizing things
200
511473
2055
08:45
so that the things you are most
likely to need are most accessible
likely to need are most accessible
201
513552
3129
08:48
can also be applied in your office.
202
516705
1676
08:51
The Japanese economist Yukio Noguchi
203
519091
1930
08:53
actually invented a filing system
that has exactly this property.
that has exactly this property.
204
521045
3055
08:57
He started with a cardboard box,
205
525161
1609
08:58
and he put his documents into the box
from the left-hand side.
from the left-hand side.
206
526794
2940
09:02
Each time he'd add a document,
207
530067
1434
09:03
he'd move what was in there along
208
531525
1621
09:05
and he'd add that document
to the left-hand side of the box.
to the left-hand side of the box.
209
533170
2810
09:08
And each time he accessed
a document, he'd take it out,
a document, he'd take it out,
210
536004
2667
09:10
consult it and put it back in
on the left-hand side.
on the left-hand side.
211
538695
2608
09:13
As a result, the documents would be
ordered from left to right
ordered from left to right
212
541327
3281
09:16
by how recently they had been used.
213
544632
1761
09:18
And he found he could quickly find
what he was looking for
what he was looking for
214
546417
2774
09:21
by starting at the left-hand
side of the box
side of the box
215
549215
2054
09:23
and working his way to the right.
216
551293
1572
09:25
Before you dash home
and implement this filing system --
and implement this filing system --
217
553215
2621
09:27
(Laughter)
218
555860
1731
09:29
it's worth recognizing
that you probably already have.
that you probably already have.
219
557615
2638
09:32
(Laughter)
220
560891
3310
09:36
That pile of papers on your desk ...
221
564225
2866
09:39
typically maligned
as messy and disorganized,
as messy and disorganized,
222
567115
2474
09:41
a pile of papers is, in fact,
perfectly organized --
perfectly organized --
223
569613
2565
09:44
(Laughter)
224
572202
1000
09:45
as long as you, when you take a paper out,
225
573226
2014
09:47
put it back on the top of the pile,
226
575264
2432
09:49
then those papers are going
to be ordered from top to bottom
to be ordered from top to bottom
227
577720
2826
09:52
by how recently they were used,
228
580570
1552
09:54
and you can probably quickly find
what you're looking for
what you're looking for
229
582146
2701
09:56
by starting at the top of the pile.
230
584871
1700
09:59
Organizing your wardrobe or your desk
231
587988
1848
10:01
are probably not the most pressing
problems in your life.
problems in your life.
232
589860
2681
10:05
Sometimes the problems we have to solve
are simply very, very hard.
are simply very, very hard.
233
593588
3602
10:09
But even in those cases,
234
597716
1150
10:10
computer science can offer some strategies
235
598890
2021
10:12
and perhaps some solace.
236
600935
1669
10:16
The best algorithms are about doing
what makes the most sense
what makes the most sense
237
604048
3075
10:19
in the least amount of time.
238
607147
1502
10:22
When computers face hard problems,
239
610627
1958
10:24
they deal with them by making them
into simpler problems --
into simpler problems --
240
612609
2762
10:27
by making use of randomness,
241
615395
1346
10:28
by removing constraints
or by allowing approximations.
or by allowing approximations.
242
616765
3817
10:32
Solving those simpler problems
243
620606
1494
10:34
can give you insight
into the harder problems,
into the harder problems,
244
622124
2920
10:37
and sometimes produces
pretty good solutions in their own right.
pretty good solutions in their own right.
245
625068
3031
10:41
Knowing all of this has helped me
to relax when I have to make decisions.
to relax when I have to make decisions.
246
629658
3569
10:45
You could take the 37 percent rule
for finding a home as an example.
for finding a home as an example.
247
633251
3255
10:49
There's no way that you can
consider all of the options,
consider all of the options,
248
637015
2702
10:51
so you have to take a chance.
249
639741
1580
10:53
And even if you follow
the optimal strategy,
the optimal strategy,
250
641888
2860
10:56
you're not guaranteed a perfect outcome.
251
644772
1931
10:59
If you follow the 37 percent rule,
252
647198
2154
11:01
the probability that you find
the very best place is --
the very best place is --
253
649376
3279
11:04
funnily enough ...
254
652679
1508
11:06
(Laughter)
255
654211
1014
11:07
37 percent.
256
655249
1420
11:09
You fail most of the time.
257
657966
1689
11:12
But that's the best that you can do.
258
660522
1744
11:14
Ultimately, computer science
can help to make us more forgiving
can help to make us more forgiving
259
662846
3052
11:17
of our own limitations.
260
665922
1909
11:20
You can't control outcomes,
just processes.
just processes.
261
668333
2539
11:22
And as long as you've used
the best process,
the best process,
262
670896
2119
11:25
you've done the best that you can.
263
673039
1763
11:26
Sometimes those best processes
involve taking a chance --
involve taking a chance --
264
674826
3688
11:30
not considering all of your options,
265
678538
2115
11:32
or being willing to settle
for a pretty good solution.
for a pretty good solution.
266
680677
2729
11:35
These aren't the concessions
that we make when we can't be rational --
that we make when we can't be rational --
267
683430
3350
11:38
they're what being rational means.
268
686804
1629
11:40
Thank you.
269
688833
1151
11:42
(Applause)
270
690008
4547
ABOUT THE SPEAKER
Tom Griffiths - Psychologist, cognitive scientistTom Griffiths uses ideas from computer science to understand how human minds work.
Why you should listen
Tom Griffiths's research explores connections between natural and artificial intelligence to discover how people solve the challenging computational problems they encounter in everyday life. Currently the Henry R. Luce Professor of Information Technology, Consciousness, and Culture at Princeton University, his work has received awards from organizations ranging from the American Psychological Association to the Sloan Foundation.
In 2016, Griffiths and his friend and collaborator Brian Christian published Algorithms to Live By, a book that illustrates how understanding the algorithms used by computers can inform human decision-making (and vice versa). The book was named one of the Amazon.com "Best Science Books of 2016" and appeared on Forbes's "Must-read brain books of 2016" list as well as the MIT Technology Review's "Best books of 2016" list.
Tom Griffiths | Speaker | TED.com