1
00:00:05,368 --> 00:00:07,750
ERIK DEMAINE: I think fun
is a very important part

2
00:00:07,750 --> 00:00:10,890
to life in general, and
in particular, work.

3
00:00:10,890 --> 00:00:13,420
Maybe not so common
to view it that way,

4
00:00:13,420 --> 00:00:15,909
but I think if you
have fun, if you enjoy,

5
00:00:15,909 --> 00:00:17,700
and if you're passionate
about what you do,

6
00:00:17,700 --> 00:00:19,050
then you'll do it well.

7
00:00:19,050 --> 00:00:20,880
And you'll be super-productive.

8
00:00:20,880 --> 00:00:23,390
And people will love you for it.

9
00:00:23,390 --> 00:00:27,260
And so my guiding principle
in the research that I do

10
00:00:27,260 --> 00:00:31,646
has been to always pursue
what I view as fun.

11
00:00:31,646 --> 00:00:34,020
And sometimes it's things that
are very relatable as fun,

12
00:00:34,020 --> 00:00:35,492
like games and puzzles.

13
00:00:35,492 --> 00:00:36,950
Sometimes it's more
obscure things,

14
00:00:36,950 --> 00:00:38,610
but I personally
find them really fun.

15
00:00:38,610 --> 00:00:41,160
And maybe they're
more practical.

16
00:00:41,160 --> 00:00:43,400
And so it's also nice
to have a mixture.

17
00:00:43,400 --> 00:00:47,900
But in teaching, I think
learning should also be fun.

18
00:00:47,900 --> 00:00:50,730
My dad raised me this way
when he was homeschooling me--

19
00:00:50,730 --> 00:00:53,870
and that always,
you should learn

20
00:00:53,870 --> 00:00:55,170
what you're curious about.

21
00:00:55,170 --> 00:00:58,020
And so conversely
as a teacher, I

22
00:00:58,020 --> 00:01:02,680
want to make material really
interesting and intriguing

23
00:01:02,680 --> 00:01:06,590
and just fun to learn about,
because I think students will

24
00:01:06,590 --> 00:01:08,760
be more excited and engage
more with the material

25
00:01:08,760 --> 00:01:10,190
if they enjoy it.

26
00:01:10,190 --> 00:01:12,090
Seems natural.

27
00:01:12,090 --> 00:01:14,320
And so in this
class, I really tried

28
00:01:14,320 --> 00:01:16,930
to bring lots of fun
examples-- I mean,

29
00:01:16,930 --> 00:01:19,977
maybe there are some
more serious problems

30
00:01:19,977 --> 00:01:21,560
that people have
proved hard, and then

31
00:01:21,560 --> 00:01:23,222
there's more
entertaining problems

32
00:01:23,222 --> 00:01:24,430
that people have proved hard.

33
00:01:24,430 --> 00:01:27,320
So I often cover both.

34
00:01:27,320 --> 00:01:29,120
But given the choice,
I would prefer

35
00:01:29,120 --> 00:01:31,504
the one that's more
entertaining and something

36
00:01:31,504 --> 00:01:34,170
that people can relate to, like,
yeah, I played that video game.

37
00:01:34,170 --> 00:01:35,060
And it was really tough.

38
00:01:35,060 --> 00:01:36,976
And now we can actually
prove that it's tough,

39
00:01:36,976 --> 00:01:40,040
then that's what
makes it exciting.

40
00:01:40,040 --> 00:01:41,970
And so it's really
nice-- I mean,

41
00:01:41,970 --> 00:01:45,670
I view teaching as kind
a entertainment of sorts.

42
00:01:45,670 --> 00:01:48,357
I think it's nice to
have that as a goal.

43
00:01:48,357 --> 00:01:50,440
And if people are entertained,
they'll be engaged.

44
00:01:50,440 --> 00:01:51,370
And they'll learn the material.

45
00:01:51,370 --> 00:01:52,911
And they'll care
more deeply about it

46
00:01:52,911 --> 00:01:56,140
and hopefully do research
in the field and all that.

47
00:01:56,140 --> 00:01:58,750
When I was younger, when
I was a graduate student,

48
00:01:58,750 --> 00:02:02,150
a lot of my more senior
colleagues, other professors,

49
00:02:02,150 --> 00:02:07,780
would warn me that it was maybe
dangerous to pursue fun things,

50
00:02:07,780 --> 00:02:09,509
that people would not
view it as serious.

51
00:02:09,509 --> 00:02:13,040
And maybe you won't get
a job when you graduate.

52
00:02:13,040 --> 00:02:15,970
And I kind of
ignored those people.

53
00:02:15,970 --> 00:02:19,540
And it was a bit
of a bold guess.

54
00:02:19,540 --> 00:02:22,240
But it worked out well for me.

55
00:02:22,240 --> 00:02:24,250
And later, those same
people apologized

56
00:02:24,250 --> 00:02:26,340
for making that suggestion.

57
00:02:26,340 --> 00:02:30,086
I think, again, if you do
things you're passionate about,

58
00:02:30,086 --> 00:02:31,210
you'll do them really well.

59
00:02:31,210 --> 00:02:32,460
And you'll be very productive.

60
00:02:32,460 --> 00:02:37,250
And so you can make up
for the topics being

61
00:02:37,250 --> 00:02:40,810
more recreational by
showing that they're

62
00:02:40,810 --> 00:02:44,410
really powerful and really
deep mathematically.

63
00:02:44,410 --> 00:02:48,080
And that's turned
out to work for me.

64
00:02:48,080 --> 00:02:53,060
Since being at MIT, it was,
again, like well, like, tenure.

65
00:02:53,060 --> 00:02:55,180
Will people appreciate
the fun side of things?

66
00:02:55,180 --> 00:02:57,175
And I do lots of serious
things in addition

67
00:02:57,175 --> 00:02:58,130
to lots of fun things.

68
00:02:58,130 --> 00:02:59,560
But it worked out.

69
00:02:59,560 --> 00:03:00,580
I got tenure.

70
00:03:00,580 --> 00:03:06,360
And in fact, I think it's an
important part of who I am.

71
00:03:06,360 --> 00:03:08,840
One analogy is to
Martin Gardner.

72
00:03:08,840 --> 00:03:10,130
He's sort of an idol of mine.

73
00:03:10,130 --> 00:03:11,630
I never got to meet
him, although we

74
00:03:11,630 --> 00:03:13,740
corresponded a little bit.

75
00:03:13,740 --> 00:03:18,040
And he is called the father
of recreational mathematics.

76
00:03:18,040 --> 00:03:21,270
He wrote a column for
Scientific American for decades.

77
00:03:21,270 --> 00:03:24,830
And it really influenced a lot
of present day mathematicians

78
00:03:24,830 --> 00:03:26,980
and computer scientists
to get into the field,

79
00:03:26,980 --> 00:03:28,230
because they read this column.

80
00:03:28,230 --> 00:03:29,810
And it's talking
about open problems,

81
00:03:29,810 --> 00:03:30,960
but in a very accessible way.

82
00:03:30,960 --> 00:03:32,334
He wasn't really
a mathematician.

83
00:03:32,334 --> 00:03:34,770
He just learned
from mathematicians

84
00:03:34,770 --> 00:03:37,630
that he knew and wrote about the
cool problems, very accessible

85
00:03:37,630 --> 00:03:41,170
problems, and fun problems,
that people were working on

86
00:03:41,170 --> 00:03:43,660
and what kind of
solutions they had.

87
00:03:43,660 --> 00:03:46,600
And so Martin Gardner's
kind of a model--

88
00:03:46,600 --> 00:03:48,370
although I do it more
in the teaching side

89
00:03:48,370 --> 00:03:50,680
than in the writing
side-- of a way

90
00:03:50,680 --> 00:03:54,600
to engage new people
into the field,

91
00:03:54,600 --> 00:03:56,090
get people excited
about science,

92
00:03:56,090 --> 00:03:58,490
and technology, and mathematics.

93
00:03:58,490 --> 00:04:01,350
And I think MIT
appreciates that a lot,

94
00:04:01,350 --> 00:04:04,450
that by working on fun problems,
you can reach a broad audience.

95
00:04:04,450 --> 00:04:06,760
And when we proved Super
Mario Brothers as NP complete,

96
00:04:06,760 --> 00:04:10,120
all the newspapers were
covering it, like, wow.

97
00:04:10,120 --> 00:04:12,166
Hard Nintendo games
really are hard.

98
00:04:12,166 --> 00:04:14,540
You know, mathematicians proved
this thing everyone knew.

99
00:04:14,540 --> 00:04:18,620
It sounds funny, and so
it makes for good press.

100
00:04:18,620 --> 00:04:21,360
And I think people are intrigued
by, oh, what does that mean?

101
00:04:21,360 --> 00:04:24,560
Then they can go and read
the paper or take this class,

102
00:04:24,560 --> 00:04:27,250
and it covers how do you
prove something like that?

103
00:04:27,250 --> 00:04:29,190
And how do you
actually-- it's really

104
00:04:29,190 --> 00:04:31,570
fun to prove that Super
Mario Brothers is hard.

105
00:04:31,570 --> 00:04:35,830
You basically construct little
levels that build a computer.

106
00:04:35,830 --> 00:04:37,750
So you build a Super
Mario Brothers level that

107
00:04:37,750 --> 00:04:39,010
represents an entire computer.

108
00:04:39,010 --> 00:04:41,187
And that's a fun thing to do.

109
00:04:41,187 --> 00:04:42,770
And I think it can
get a lot of people

110
00:04:42,770 --> 00:04:45,470
excited about computer science.

111
00:04:45,470 --> 00:04:48,270
When I first came
to MIT, I used this

112
00:04:48,270 --> 00:04:52,610
as a principle to get new
graduate students started.

113
00:04:52,610 --> 00:04:55,190
So they're trying to
find a problem to solve.

114
00:04:55,190 --> 00:04:57,440
They don't really know
what they're interested in.

115
00:04:57,440 --> 00:04:59,460
So often, I'll suggest
a game or a puzzle.

116
00:04:59,460 --> 00:05:00,826
Hey, let's prove this hard.

117
00:05:00,826 --> 00:05:02,200
One of the first
ones was Tetris.

118
00:05:02,200 --> 00:05:06,540
That was done with two MIT
students back in 2003, I think,

119
00:05:06,540 --> 00:05:08,079
just after I started.

120
00:05:08,079 --> 00:05:10,120
It's like, I've always
been curious about Tetris.

121
00:05:10,120 --> 00:05:11,830
It seems really tricky.

122
00:05:11,830 --> 00:05:13,990
And then together, we
proved that it was hard.

123
00:05:13,990 --> 00:05:15,570
And that was their first paper.

124
00:05:15,570 --> 00:05:17,950
And it got them kind
of psyched up and ready

125
00:05:17,950 --> 00:05:19,800
to solve more problems.

126
00:05:19,800 --> 00:05:22,490
They don't do hardness
of games anymore,

127
00:05:22,490 --> 00:05:25,030
but it was like a nice warmup
to build their confidence

128
00:05:25,030 --> 00:05:26,814
and show that they can do this.

129
00:05:26,814 --> 00:05:28,230
So I think a lot
of these problems

130
00:05:28,230 --> 00:05:30,690
are it's hard just made easy.

131
00:05:30,690 --> 00:05:32,530
It becomes quite accessible.

132
00:05:32,530 --> 00:05:35,560
One of the things that really
made me happy about this class

133
00:05:35,560 --> 00:05:39,260
is-- I teach a lot of
advanced theory classes.

134
00:05:39,260 --> 00:05:42,600
And each one of them has a
final project at the end.

135
00:05:42,600 --> 00:05:45,420
And this class also
had a final project.

136
00:05:45,420 --> 00:05:47,280
And there are many
types you can do.

137
00:05:47,280 --> 00:05:49,540
But the type I'm always
most excited about

138
00:05:49,540 --> 00:05:51,700
is to solve an open problem.

139
00:05:51,700 --> 00:05:56,280
And in this hardness proof
class, most of the student

140
00:05:56,280 --> 00:05:57,850
solved an open problem.

141
00:05:57,850 --> 00:06:01,580
Compared to any other
class I've taught or seen,

142
00:06:01,580 --> 00:06:03,510
many more students
solve the open problems,

143
00:06:03,510 --> 00:06:05,980
because a lot of them are
relatively accessible.

144
00:06:05,980 --> 00:06:08,585
You can take a game or puzzle
that just you like playing

145
00:06:08,585 --> 00:06:10,860
and you have expertise
in, but no one else

146
00:06:10,860 --> 00:06:12,800
has tried to analyze
mathematically.

147
00:06:12,800 --> 00:06:15,470
And now, suddenly, you can
bring these new tools to it

148
00:06:15,470 --> 00:06:19,340
and prove your result
and write a paper.

149
00:06:19,340 --> 00:06:21,790
Fun is like the
hook that gets you

150
00:06:21,790 --> 00:06:27,030
excited about the research,
the problems, and the material.

151
00:06:27,030 --> 00:06:30,930
So we analyze a lot
of games and puzzles.

152
00:06:30,930 --> 00:06:33,390
And most of them, to kind
of keep it interesting,

153
00:06:33,390 --> 00:06:36,115
we analyze real world games
and puzzles, like video games

154
00:06:36,115 --> 00:06:38,040
that people play.

155
00:06:38,040 --> 00:06:40,450
And one example is
Super Mario Brothers,

156
00:06:40,450 --> 00:06:43,380
where we can prove that it's NP
hard to figure out whether you

157
00:06:43,380 --> 00:06:44,930
can solve a given level.

158
00:06:44,930 --> 00:06:48,310
So there's no good algorithm
to, given a Super Mario Brothers

159
00:06:48,310 --> 00:06:51,150
level, figure out
how to solve it.

160
00:06:51,150 --> 00:06:54,170
This, I guess, becomes
particularly relevant

161
00:06:54,170 --> 00:06:57,020
once you can make your
own custom levels.

162
00:06:57,020 --> 00:07:01,980
So with the upcoming release
of Super Mario Maker, now

163
00:07:01,980 --> 00:07:05,090
people can actually
build their own levels.

164
00:07:05,090 --> 00:07:07,630
This'll be really fun from
a research perspective,

165
00:07:07,630 --> 00:07:10,400
because we can take all the
levels that we've designed.

166
00:07:10,400 --> 00:07:12,972
When we design them, we just
arrange them as pixel art,

167
00:07:12,972 --> 00:07:15,180
so that you can see the
level, but you can't actually

168
00:07:15,180 --> 00:07:16,964
experience the level.

169
00:07:16,964 --> 00:07:18,880
And in order to actually
experience the level,

170
00:07:18,880 --> 00:07:20,780
you'd have to modify
the ROM of the game.

171
00:07:20,780 --> 00:07:24,340
And it's very tricky and
technically annoying,

172
00:07:24,340 --> 00:07:27,230
and maybe also
illegal to do that.

173
00:07:27,230 --> 00:07:30,340
Super Mario Maker suddenly
makes it accessible

174
00:07:30,340 --> 00:07:32,140
that everyone can
make custom levels.

175
00:07:32,140 --> 00:07:34,140
I think it also makes the
results more relevant.

176
00:07:34,140 --> 00:07:36,690
It says, hey, if you
download a new level,

177
00:07:36,690 --> 00:07:39,330
figuring out how to solve it is
going to be a real challenge.

178
00:07:39,330 --> 00:07:41,570
And I think this
is a big reason why

179
00:07:41,570 --> 00:07:44,710
people find video
games fun, because they

180
00:07:44,710 --> 00:07:45,915
find them challenging.

181
00:07:45,915 --> 00:07:48,040
I think if a game is too
easy-- there are exception

182
00:07:48,040 --> 00:07:52,190
but if a game is too easy, it
gets boring pretty quickly.

183
00:07:52,190 --> 00:07:54,872
And when there's
this nice-- when

184
00:07:54,872 --> 00:07:56,580
you can set the
difficulty of the problem

185
00:07:56,580 --> 00:07:59,600
so it's not impossible, but
at least it's a big challenge

186
00:07:59,600 --> 00:08:03,710
to solve it, I think
that's where you get

187
00:08:03,710 --> 00:08:05,970
the sweet spot in having fun.

188
00:08:05,970 --> 00:08:08,130
And this whole area
of hardness proofs

189
00:08:08,130 --> 00:08:11,070
gives you a mathematical way
of analyzing games and puzzles

190
00:08:11,070 --> 00:08:13,050
to show, yeah, they
are really challenging.

191
00:08:13,050 --> 00:08:15,270
They're hard even for
computers to solve them.

192
00:08:15,270 --> 00:08:19,000
So for humans, it could be
also really challenging.