1
00:00:00,460 --> 00:00:05,190
This chapter is a very brief introduction
to some of the most important

2
00:00:05,190 --> 00:00:10,200
fundamentals of communicating
ideas using visual signs.

3
00:00:11,300 --> 00:00:15,770
There's some very basic principles
that if you keep in mind,

4
00:00:15,770 --> 00:00:20,890
as you design your visualizations,
they really become essential and valuable.

5
00:00:20,890 --> 00:00:26,110
As you make your own design your own
visualizations to present your own data.

6
00:00:26,110 --> 00:00:29,200
And you'll find that there's some
very simple things you can do

7
00:00:29,200 --> 00:00:32,070
that I think will help tremendously.

8
00:00:32,070 --> 00:00:33,280
So the first, and

9
00:00:33,280 --> 00:00:39,760
probably one of the most often overlooked
ideas, is the concept of The Grid System.

10
00:00:39,760 --> 00:00:47,100
So this is a page I've taken from
Josef Mueller-Brockman's extensive

11
00:00:47,100 --> 00:00:53,570
book on grid systems, and
what you can see here is he's laid out

12
00:00:53,570 --> 00:00:59,500
the idea of how a grid organizes
data when it's presented on a page.

13
00:00:59,500 --> 00:01:05,770
And what you can see here is that
the information's organized into columns.

14
00:01:05,770 --> 00:01:13,278
Each column is left-justified to each
of the each of the different grids.

15
00:01:13,278 --> 00:01:15,700
[COUGH] Excuse me.

16
00:01:15,700 --> 00:01:19,910
The margins between the grids
are all the same size,

17
00:01:20,950 --> 00:01:24,410
and the line height is the same size.

18
00:01:24,410 --> 00:01:27,788
Now, this gives incredible meaning and

19
00:01:27,788 --> 00:01:32,730
organization to the,
to what's being presented.

20
00:01:32,730 --> 00:01:38,780
So first of all you'll see that the eye
very naturally flows across and

21
00:01:38,780 --> 00:01:40,860
down each one of the columns.

22
00:01:40,860 --> 00:01:45,420
Separately, following that gestalt
principle that we talked about before,

23
00:01:45,420 --> 00:01:50,000
of proximity,
things that are in the same column,

24
00:01:50,000 --> 00:01:55,420
are our mind reads them as having
perceived similarity of meaning.

25
00:01:56,710 --> 00:02:03,020
They can see how the grid system,
is utilized by a newspaper

26
00:02:03,020 --> 00:02:09,030
like the New York Times to present what
is truly a very dense amount of data.

27
00:02:09,030 --> 00:02:12,280
And one of the reasons that it
makes sense is because they have so

28
00:02:12,280 --> 00:02:13,880
strictly enforced the grid.

29
00:02:15,640 --> 00:02:20,070
There's another major principle that
the New York Times is abiding by here.

30
00:02:20,070 --> 00:02:24,530
And it's something that we can
find when we go back even to

31
00:02:24,530 --> 00:02:30,710
ancient Egyptian papyrus, and it's this
idea that's called hierarchy of size.

32
00:02:30,710 --> 00:02:35,360
And in this beautiful scene taken
from the Egyptian Book of the Dead,

33
00:02:35,360 --> 00:02:40,200
the most important figures are look,
the biggest.

34
00:02:40,200 --> 00:02:46,730
And so what you see here is in the center
on the bottom row is the god Anubis,

35
00:02:46,730 --> 00:02:51,299
and Anubis is twice the size of
the human being on his left.

36
00:02:52,520 --> 00:02:57,850
Now, we interpret this in modern times
not to mean that Anubis was twice as

37
00:02:57,850 --> 00:03:04,290
tall as a man, but that he was twice
as important as the other figures.

38
00:03:04,290 --> 00:03:06,220
And you'll find that the royal,

39
00:03:06,220 --> 00:03:11,430
the people on the left who are the same
size as the gods are the rulers.

40
00:03:12,650 --> 00:03:19,250
So this hierarchy of size is imposed,
also in The New York Times,

41
00:03:19,250 --> 00:03:26,770
where you'll see that the main what is
the biggest visual item on the page?

42
00:03:26,770 --> 00:03:28,200
Well, it's the headline, right?

43
00:03:28,200 --> 00:03:30,050
It's The New York Times logo.

44
00:03:30,050 --> 00:03:32,510
What's the second biggest
thing on the page?

45
00:03:32,510 --> 00:03:35,880
It's the central image,
and the central column.

46
00:03:35,880 --> 00:03:41,550
So your eye is drawn to the most
visually dominant aspect of the page.

47
00:03:41,550 --> 00:03:45,740
So the reason why making things
large gives them more meaning.

48
00:03:45,740 --> 00:03:49,680
So the idea of hierarchy of size and

49
00:03:49,680 --> 00:03:53,270
grid are really being strictly being
enforced by the New York Times.

50
00:03:53,270 --> 00:03:56,230
What are some other basic
elements of visual communication?

51
00:03:58,110 --> 00:04:01,510
Another one of the most common
is this idea of grouping.

52
00:04:01,510 --> 00:04:06,560
And you can see how this appears
in the periodic table here,

53
00:04:06,560 --> 00:04:10,050
and very often one of the most
important parts about

54
00:04:10,050 --> 00:04:14,200
the periodic table is the way that
the columns themselves are organized.

55
00:04:19,690 --> 00:04:21,356
And what you have on the left
are all of the metals and

56
00:04:21,356 --> 00:04:24,310
what you have on the right are all of the
non-metals, and in the middle are all of

57
00:04:24,310 --> 00:04:29,500
the transition metals, and on the very
bottom is the lanthanum series.

58
00:04:29,500 --> 00:04:36,500
And these groups are given meaning
by being next to one another.

59
00:04:36,500 --> 00:04:39,180
In other words,
it's very clear that hydrogen and

60
00:04:39,180 --> 00:04:43,660
lithium share a common quality
by being in the same column.

61
00:04:46,420 --> 00:04:49,300
There's one additional
principle that I think is very,

62
00:04:49,300 --> 00:04:54,100
very often overlooked in
scientific visualization.

63
00:04:54,100 --> 00:05:00,480
Tufte brings this about, and he calls
this idea one plus one equals three.

64
00:05:00,480 --> 00:05:01,470
So let me explain it.

65
00:05:01,470 --> 00:05:02,990
Let's take a moment.

66
00:05:02,990 --> 00:05:05,160
Look at the screen and
tell me what you see.

67
00:05:06,720 --> 00:05:09,340
I mean most people would
say they see two lines.

68
00:05:11,360 --> 00:05:12,640
Let's go again.

69
00:05:12,640 --> 00:05:14,250
Now tell me what you see.

70
00:05:14,250 --> 00:05:17,360
I think most people consistently say,
I see two lines.

71
00:05:19,210 --> 00:05:21,690
Let's do it one more time,
now what do you see?

72
00:05:22,740 --> 00:05:25,940
Most people will say,
I see two lines and a space.

73
00:05:26,950 --> 00:05:32,360
It's very much like the two lines now
create a road through this, graphic.

74
00:05:33,510 --> 00:05:39,540
Essentially, the content is the same,
but the spacing is different.

75
00:05:39,540 --> 00:05:42,700
And there's a perceptive
psychology effect here, and

76
00:05:42,700 --> 00:05:45,220
Tufte calls this,
one plus one equals three.

77
00:05:46,350 --> 00:05:52,270
And what he means is that even though you
have two graphical elements on the page,

78
00:05:52,270 --> 00:05:57,230
what you really have to pay attention
to is the third graphical element,

79
00:05:57,230 --> 00:06:02,800
which is the negative space that's created
between two strong graphical elements.

80
00:06:03,980 --> 00:06:06,640
So when you create lines,

81
00:06:06,640 --> 00:06:12,500
what's very important is that you realize
that you're not just drawing a line,

82
00:06:12,500 --> 00:06:17,940
but by dividing space you're creating
all the negative space around it.

83
00:06:17,940 --> 00:06:21,060
So it's very easy in drawings for

84
00:06:21,060 --> 00:06:27,430
us to create many lines that all,
call for our attention.

85
00:06:28,900 --> 00:06:33,400
So what is one of the ways that we can
offer understanding the principles of

86
00:06:33,400 --> 00:06:38,590
psychology, to, and perception, for
how to minimize something like this?

87
00:06:38,590 --> 00:06:43,020
Well, let's look at what
Tufte would call reduction of

88
00:06:43,020 --> 00:06:47,710
noise in this classic box and
whisker plot.

89
00:06:47,710 --> 00:06:54,120
So very briefly, the plot shows a number
of categories along the X axis.

90
00:06:54,120 --> 00:07:00,250
What you see is the max and
min values along the lines, and

91
00:07:00,250 --> 00:07:06,600
then the box and those are the whiskers,
and the box shows the standard deviation,

92
00:07:06,600 --> 00:07:11,610
the you know, the majority of the, 95% and
the line in the middle shows the median.

93
00:07:12,760 --> 00:07:14,820
So, this is a,

94
00:07:14,820 --> 00:07:20,351
traditional scientific graph that
really came out of two key in the 50s.

95
00:07:21,900 --> 00:07:28,440
So, what is it about
this graph that really

96
00:07:28,440 --> 00:07:32,043
confronts this concept of
one plus one equals three?

97
00:07:32,043 --> 00:07:37,100
Well, Tufte has another idea that
he calls the graphical integrity.

98
00:07:37,100 --> 00:07:43,270
The main idea is that every line in
a visualization should convey meaning, and

99
00:07:43,270 --> 00:07:48,040
you should convey meaning in
the most compact form possible.

100
00:07:48,040 --> 00:07:49,400
So let's take an example.

101
00:07:49,400 --> 00:07:51,630
What are lines that don't
need to be on this screen?

102
00:07:52,710 --> 00:07:55,070
Well, let's take the most obvious ones.

103
00:07:55,070 --> 00:07:58,460
The first is the background lines.

104
00:07:58,460 --> 00:08:04,730
They don't really need to
communicate what can be shown in

105
00:08:04,730 --> 00:08:10,860
the very small little ticks that are now
placed on the left, and it reduces

106
00:08:10,860 --> 00:08:16,370
the graphical noise significantly and
it doesn't change the meaning at all.

107
00:08:16,370 --> 00:08:19,130
Well, let's continue to reduce.

108
00:08:19,130 --> 00:08:22,230
What are other lines on this
page that we don't need?

109
00:08:23,350 --> 00:08:24,410
Well very often,

110
00:08:24,410 --> 00:08:30,570
you'll see that scientists will
create figures in boxes, right?

111
00:08:30,570 --> 00:08:36,448
But the box very clearly violates this
principle of one plus one equals three,

112
00:08:36,448 --> 00:08:38,870
because what does it do?

113
00:08:38,870 --> 00:08:41,199
It draws attention to the box,

114
00:08:42,710 --> 00:08:45,260
what you really want to pay
attention to is the data.

115
00:08:46,260 --> 00:08:48,060
So how do we get rid of the box?

116
00:08:48,060 --> 00:08:51,010
We can very simply remove the top and

117
00:08:51,010 --> 00:08:55,440
the right,
which don't add any meaning to the figure.

118
00:08:55,440 --> 00:09:00,480
So if you're learning anything from this,
it's, think about the graphs that

119
00:09:00,480 --> 00:09:05,600
you're drawing, and how much
additional line they are including for

120
00:09:05,600 --> 00:09:07,360
the data that they communicate.

121
00:09:08,975 --> 00:09:12,914
Tufte goes even further,
and he says you know what?

122
00:09:12,914 --> 00:09:16,415
The box is a two-dimensional box, but

123
00:09:16,415 --> 00:09:23,240
what it's really communicating
is the top and the bottom point.

124
00:09:23,240 --> 00:09:27,550
And the line is a point,
it doesn't need to be a line.

125
00:09:27,550 --> 00:09:30,750
So if we want to reduce this even further,

126
00:09:30,750 --> 00:09:36,740
we can actually remove the box entirely
and replace the midpoint with a dot.

127
00:09:38,370 --> 00:09:44,620
Now you have a graph that has no
less meaning, and far less ink.

128
00:09:46,060 --> 00:09:50,180
In the way that Tufte
describes data graphics,

129
00:09:50,180 --> 00:09:53,640
this is a high value for us to strive for.

130
00:09:55,320 --> 00:10:00,230
Another thing that we can
do that really allows us to

131
00:10:00,230 --> 00:10:03,690
focus on this idea of one
plus one equals three.

132
00:10:03,690 --> 00:10:06,810
Look at each one of
the graphical lines here.

133
00:10:06,810 --> 00:10:11,400
They're really significant, and,
when you look at the image,

134
00:10:11,400 --> 00:10:13,980
you don't just see the data.

135
00:10:13,980 --> 00:10:16,890
You see the line.

136
00:10:16,890 --> 00:10:22,760
But the line is really pointing you,
just to the endpoints, right?

137
00:10:22,760 --> 00:10:25,480
What it's really showing you,
is, the top line is

138
00:10:25,480 --> 00:10:29,940
showing you the maximum data point,
and, the top of the standard deviation.

139
00:10:30,940 --> 00:10:38,110
So, what if you also diminish
the intensity, or the width, of each line?

140
00:10:39,150 --> 00:10:44,210
And now what you do is you've
taken very thick lines,

141
00:10:44,210 --> 00:10:49,530
which draw attention to themselves,
and made it so

142
00:10:49,530 --> 00:10:54,810
that what you're paying attention
to are the end points of the lines.

143
00:10:54,810 --> 00:10:59,510
So you can even further
reduce the noise or

144
00:10:59,510 --> 00:11:05,140
the ink necessary to communicate
the number of data points

145
00:11:05,140 --> 00:11:11,150
in a graph by turning down
the intensity of the lines.

146
00:11:13,125 --> 00:11:18,580
Tufte talked about this as expressiveness.

147
00:11:18,580 --> 00:11:23,840
What's important is that data
graphics are expressive and

148
00:11:23,840 --> 00:11:30,470
here I've chosen McKinley's description
which is the idea that, visual,

149
00:11:30,470 --> 00:11:38,410
graphics need to express all the facts in
the data and only the facts in the data.

150
00:11:39,590 --> 00:11:41,190
This is something that's very easy to do,

151
00:11:41,190 --> 00:11:45,590
to accidentally, violate, and
I'll give you a quick example.

152
00:11:46,960 --> 00:11:52,790
So here you see a data set that shows
you the relationship between a different

153
00:11:52,790 --> 00:11:56,740
number of car manufacturers and
the country in which they're manufactured.

154
00:11:57,750 --> 00:12:00,760
So look at this graph,
what is wrong with it?

155
00:12:02,110 --> 00:12:09,720
So you see that, you can easily look at
this, and tell that the various cars,

156
00:12:09,720 --> 00:12:14,340
come from different nations, but
what's misleading about this graphic?

157
00:12:15,560 --> 00:12:21,520
Well, what's misleading about this
graphic, is the bar communicates

158
00:12:21,520 --> 00:12:27,350
an idea of a continuous value,
and that the bar for

159
00:12:27,350 --> 00:12:33,300
Honda is bigger than the bar for
the ACM, for the AMC.

160
00:12:34,310 --> 00:12:42,660
And actually, there's no quantity
communicated in each of these values.

161
00:12:42,660 --> 00:12:47,690
And a better way for us to represent
this fact would be by showing points.

162
00:12:49,330 --> 00:12:53,120
What we actually have is
that we're mapping ordinal,

163
00:12:53,120 --> 00:12:56,530
excuse me, nominal values, right?

164
00:12:56,530 --> 00:13:01,380
Category labels,
which is the country with the origin of

165
00:13:01,380 --> 00:13:06,280
the manufacturer of each car and
nominal values don't have magnitude.

166
00:13:07,470 --> 00:13:11,578
So when you show them using
a bar chart like that,

167
00:13:11,578 --> 00:13:16,790
you give a, the false impression
that there's meaning in the data.

168
00:13:16,790 --> 00:13:19,800
That implies a magnitude that's not there.

169
00:13:21,210 --> 00:13:26,980
So, this graph is not expressive in
the way that McKinley describes it,

170
00:13:26,980 --> 00:13:31,360
because it's communicating information
that's not actually there.

171
00:13:33,370 --> 00:13:38,990
Another thing that Tufte talks about,
with this idea of data ink.

172
00:13:38,990 --> 00:13:46,530
If we're going to make the graph
as have as few pixels as possible,

173
00:13:46,530 --> 00:13:53,218
to communicate the most meaning, what is
the furthest possible way that we can go.

174
00:13:53,218 --> 00:13:56,930
Is what Tufte calls a chart junk,

175
00:13:57,980 --> 00:14:02,480
and chart junk is
a wonderful example of how

176
00:14:03,720 --> 00:14:09,220
data graphics appear very often in
popular culture in magazines and

177
00:14:09,220 --> 00:14:15,110
they're designed to give this huge
emotional impact, like you can see

178
00:14:15,110 --> 00:14:20,970
in this graphic,
what's trying to be shown by the newspaper

179
00:14:20,970 --> 00:14:27,140
is this idea that the costs of
elections are becoming monstrous.

180
00:14:27,140 --> 00:14:31,510
And so they have this
Tasmanian devil serpent, and

181
00:14:31,510 --> 00:14:37,320
the teeth of the devil serpent represent
the spending levels, and they're going up.

182
00:14:37,320 --> 00:14:39,630
But if you look at the graph to the right,

183
00:14:39,630 --> 00:14:42,850
what's represented is the same value,
empirically.

184
00:14:42,850 --> 00:14:48,280
So in the way that Tufte
represents graphics,

185
00:14:48,280 --> 00:14:53,510
the graphic on the right is
the more meaningful graphic,

186
00:14:53,510 --> 00:15:00,310
because it communicates the same
data with far less editorial,

187
00:15:00,310 --> 00:15:05,250
right, far less unnecessary graphic,
graphical marks.

188
00:15:05,250 --> 00:15:08,491
Now it's also important to note is,
while I think Tufte,

189
00:15:08,491 --> 00:15:13,430
is a purist, in the way representation
should include meaning and

190
00:15:13,430 --> 00:15:19,757
only meaning,
experiments have shown that readers

191
00:15:19,757 --> 00:15:24,940
remember infographics with this
kind of emotional content better.

192
00:15:25,990 --> 00:15:28,920
So, as a designer of graphics you'll have

193
00:15:28,920 --> 00:15:33,940
the option of deciding the faithfulness
of the representation.

194
00:15:33,940 --> 00:15:38,370
So what I hope comes clear to you
from this particular chapter is that

195
00:15:38,370 --> 00:15:43,865
there are a number of ways that you can
choose to represent the data, and that,

196
00:15:43,865 --> 00:15:49,980
they, are small changes can give
you significant multiplier in

197
00:15:49,980 --> 00:15:56,220
the organization and the legibility of
your graphics as well as their integrity.

198
00:15:56,220 --> 00:16:00,530
And I hope that you feel more well
equipped to make decisions about how

199
00:16:00,530 --> 00:16:03,390
to decide how to encode
particular kinds of data.