1
00:00:09,740 --> 00:00:14,355
Edward Tufte's emphasis on simplicity and
minimalism has grown a great following.

2
00:00:14,355 --> 00:00:18,340
He's largely seen as one of the most
influential writers on the topic of

3
00:00:18,340 --> 00:00:20,250
visualizations for charts and graphs.

4
00:00:21,400 --> 00:00:23,330
In his book, Beautiful Evidence,

5
00:00:23,330 --> 00:00:27,530
he introduced the idea that took
minimalism in charts to a new level.

6
00:00:27,530 --> 00:00:31,570
Instead of making a chart a separate
artifact to be studied on it's own,

7
00:00:31,570 --> 00:00:36,330
he suggested it be reduced and
embedded within the context of discussion.

8
00:00:36,330 --> 00:00:41,130
He argued that a small graph, say a time
series line graph, could convey so

9
00:00:41,130 --> 00:00:45,600
much more information quickly that
it would be invaluable to readers.

10
00:00:45,600 --> 00:00:50,020
He called these items sparklines and
referred to them as data words.

11
00:00:50,020 --> 00:00:53,520
An intriguing way to bridge
the gap between text and figures.

12
00:00:54,810 --> 00:00:59,180
He suggested that sparklines could not
only be represented directly in text, but

13
00:00:59,180 --> 00:01:03,320
also could be embedded in tables
along with the data they describe.

14
00:01:03,320 --> 00:01:07,090
This thought became so natural that it's
even been pick up by one of the most

15
00:01:07,090 --> 00:01:11,800
common pieces of spreadsheet
analysis software, Microsoft Excel.

16
00:01:11,800 --> 00:01:15,819
Here is a picture of tabular data
from the NASDAQ stock exchange,

17
00:01:15,819 --> 00:01:20,655
showing the opening price of four
technology stocks over one-month period.

18
00:01:20,655 --> 00:01:22,799
Take a moment to consider this data.

19
00:01:22,799 --> 00:01:26,233
Which stock has had a bad
month with a downward trend?

20
00:01:26,233 --> 00:01:28,430
And which has had a good
month with an upward trend?

21
00:01:29,470 --> 00:01:32,750
What if I added sparklines
in the cells below?

22
00:01:32,750 --> 00:01:35,760
Here you can see immediately that
Apple has had a poor month with

23
00:01:35,760 --> 00:01:37,250
decreased stock price.

24
00:01:37,250 --> 00:01:41,190
And IBM has had good month
with increasing stock prices.

25
00:01:41,190 --> 00:01:44,945
With Amazon and Intel it's a little less
clear, and we see some spikes throughout.

26
00:01:47,193 --> 00:01:51,333
Now these sparklines don't help us to
answer who has the highest stock price, or

27
00:01:51,333 --> 00:01:53,233
who had the greatest wins or losses.

28
00:01:53,233 --> 00:01:58,190
But they do give us a general feeling for
the trend behind the data.

29
00:01:58,190 --> 00:02:01,060
And actually, sparklines are used for
data where trends or

30
00:02:01,060 --> 00:02:03,470
distribution characteristics
are important.

31
00:02:04,530 --> 00:02:08,905
Here's another example of sparklines
used in Google Finance's website.

32
00:02:08,905 --> 00:02:12,305
Notice how you can actually quickly zoom
in on pieces of the distribution with

33
00:02:12,305 --> 00:02:13,204
these sparklines.

34
00:02:13,204 --> 00:02:15,800
So they're not just
a representation of the data but

35
00:02:15,800 --> 00:02:17,570
a method of interacting with them.

36
00:02:19,200 --> 00:02:21,950
There have been lots of
different uses of sparklines.

37
00:02:21,950 --> 00:02:26,420
And customized interfaces,
such as video games, are a great example.

38
00:02:26,420 --> 00:02:29,940
A modern take on sparklines,
which I think of as kind of neat,

39
00:02:29,940 --> 00:02:31,770
is called the sparktweet.

40
00:02:31,770 --> 00:02:36,350
Where Unicode block characters are used
to display a bar graph inside the 140

41
00:02:36,350 --> 00:02:38,343
characters allotted by Twitter.

42
00:02:38,343 --> 00:02:39,318
For instance,

43
00:02:39,318 --> 00:02:44,273
here a user has tweeted her hours of sleep
over the month of April as a bar graph.

44
00:02:47,452 --> 00:02:51,020
The last Tufte principle that I want to
share with you is one called lie factor.

45
00:02:52,110 --> 00:02:55,630
Lie factor is the size of
an effect shown in the graphic

46
00:02:55,630 --> 00:02:59,360
divided by the size of
the effect actually in the data.

47
00:02:59,360 --> 00:03:01,540
It's often done unconsciously, to try and

48
00:03:01,540 --> 00:03:05,700
help tell a narrative, however,
it's misleading to the observers.

49
00:03:05,700 --> 00:03:08,510
There are lots of different
examples of lie factor.

50
00:03:08,510 --> 00:03:12,410
This one from Time Magazine in 1979
provides a great example, I think.

51
00:03:14,070 --> 00:03:18,695
In it, various barrels of oil are shown,
showing the price of oil over six years.

52
00:03:18,695 --> 00:03:22,521
But it's unclear to the viewer what
the size of one barrel means in relation

53
00:03:22,521 --> 00:03:26,300
to the others, in part because there's
an element of perspective here.

54
00:03:26,300 --> 00:03:29,613
Are the barrels different sizes,
or do they just appear to be so

55
00:03:29,613 --> 00:03:31,750
because some are in front of the others?

56
00:03:32,800 --> 00:03:36,690
Is it the volume of the barrel that
represents the growth of the cost, or

57
00:03:36,690 --> 00:03:37,530
the height of the barrel?

58
00:03:39,600 --> 00:03:44,006
Sparklines and lie factor are just two
more tools that Edward Tufte has given us

59
00:03:44,006 --> 00:03:46,152
to understand and communicate data.