1
00:00:01,275 --> 00:00:06,307
In this video I’d like to talk about the power of creating and comparing alternatives.

2
00:00:06,307 --> 00:00:08,723
And to do that I’m going to share some research

3
00:00:08,723 --> 00:00:13,572
that Steven Dow did as a postdoctoral scholar with me at Stanford University.

4
00:00:14,741 --> 00:00:18,340
When you’re designing, does it make more sense to go for quality

5
00:00:18,340 --> 00:00:21,940
and try to come up with the best possible design?


6
00:00:21,940 --> 00:00:29,131
Or does it make more sense to go for quantity first as a path to try and learn and understand?

7
00:00:29,131 --> 00:00:34,967
There’s a story that Bayles and Orland tell about an art teacher who divides the class in half,

8
00:00:34,967 --> 00:00:37,046
and he tells one half of the class,

9
00:00:37,046 --> 00:00:42,779
“You’re going to be graded exclusively on the quality of the very best thing that you make.”

10
00:00:42,779 --> 00:00:44,557
He tells the other half of the class,

11
00:00:44,557 --> 00:00:48,281
“You’re going to be graded on the quantity of things that you make.

12
00:00:48,281 --> 00:00:52,446
Doesn’t matter how good it is; all that matters is how much that you make.”

13
00:00:52,446 --> 00:00:57,799
And what this teacher found was that while the quantity group was busily churning our piles of work —

14
00:00:57,799 --> 00:00:59,842
and learning from their mistakes —

15
00:00:59,842 --> 00:01:02,613
the quality group sat around theorizing, and at the end of the day

16
00:01:02,613 --> 00:01:07,705
they had little more to show for their efforts than grandiose theories and piles of dead clay.


17
00:01:07,705 --> 00:01:14,096
So this gives us some intuition that rapidly producing many alternatives has a lot of value.

18
00:01:14,096 --> 00:01:19,095
And to explore this further, Steven and I had people create egg drop devices.

19
00:01:19,095 --> 00:01:21,096
You may have done this when you were in high school.

20
00:01:21,096 --> 00:01:23,898
If you haven’t, it’s a lot of fun, and I suggest trying it out.

21
00:01:23,898 --> 00:01:25,708
And what you can do with an egg drop device,

22
00:01:25,708 --> 00:01:30,286
is you’re building a contraption that will protect an egg from a fall.

23
00:01:30,286 --> 00:01:36,432
Here we threw one out my third-story office window and, lo and behold, the egg survives.


24
00:01:36,432 --> 00:01:42,083
And we tested a whole bunch of people in variance of this design and people come up with all sorts of stuff.

25
00:01:42,083 --> 00:01:47,588
They come up with good ideas, and bad ideas, and creative solutions, and really unimaginative ones.


26
00:01:47,588 --> 00:01:50,460
And one thing that is really interesting is that,

27
00:01:50,460 --> 00:01:58,479
in aggregate, people often pick one idea early on, and they stick with it to their detriment.

28
00:01:58,479 --> 00:02:02,582
And so here is a couple participants talking about that experience.

29
00:02:02,582 --> 00:02:07,395
(No, I don’t know, for some reason this is… this seems to be the only idea,

30
00:02:07,395 --> 00:02:12,163
in that there needs to be a platform and then it’s going to cushion, if possible, with the materials.

31
00:02:12,163 --> 00:02:14,286
I… I don’t see any, any other way.

32
00:02:14,286 --> 00:02:18,331
>> I’m not a very good outside-the-box thinker,

33
00:02:18,331 --> 00:02:21,977
so I kind of just had one idea and I was going to try and make it work.


34
00:02:21,977 --> 00:02:27,512
>> I kind of went with the whole parachute idea, and what I had from the beginning. So.


35
00:02:27,512 --> 00:02:33,249
>> This is the best approach for such a design.)

36
00:02:34,172 --> 00:02:38,970
What we see here is an example of what Karl Duncker called “functional fixation.”

37
00:02:38,970 --> 00:02:44,585
In a number of experiments that he ran in the 1940’s he gave people tasks like this:

38
00:02:44,585 --> 00:02:50,103
“Attach the candle to the wall such that none of the wax drips on the table.”

39
00:02:50,103 --> 00:02:53,352
Ten, twenty percent of the people figured it out.

40
00:02:53,352 --> 00:03:03,245
Take a moment and see if you can figure it out.

41
00:03:03,245 --> 00:03:07,665
The solution — as a couple of you have got, but I bet many people didn’t —

42
00:03:07,665 --> 00:03:13,495
is to take the box that holds the tacks and use that as a container for the candle.

43
00:03:13,495 --> 00:03:16,619
That will protect the wax from dripping on the table.

44
00:03:16,619 --> 00:03:24,239
And what’s interesting about this is that, because the tacks are in a box, we don’t see the box.


45
00:03:24,239 --> 00:03:29,640
If you give people the exact same set up, where the tacks are outside the box,

46
00:03:29,640 --> 00:03:33,851
all of a sudden the box becomes obviously available as a resource

47
00:03:33,851 --> 00:03:37,074
and nearly everybody solves exactly the same problem.

48
00:03:37,074 --> 00:03:39,463
So Stephen and I set off and tried to figure out

49
00:03:39,463 --> 00:03:44,788
whether we could augment people’s design process to get them to explore more alternatives.

50
00:03:44,788 --> 00:03:49,533
And one of the things that we did, is we forced people to come up with multiple alternatives in parallel.


51
00:03:49,533 --> 00:03:52,162
We call this parallel prototyping,

52
00:03:52,162 --> 00:03:58,653
and in this particular study we had people design graphical advertisements for the web.

53
00:03:58,653 --> 00:04:01,509
So, we’re going to put people in one of two conditions:

54
00:04:01,509 --> 00:04:07,824
You’re either going to be in a serial condition, where you iteratively create six prototypes from start to finish;

55
00:04:07,824 --> 00:04:13,635
or in a parallel condition, where you create three alternatives, get feedback, create two more,


56
00:04:13,635 --> 00:04:16,080
get feedback, and then make a final one.

57
00:04:16,080 --> 00:04:20,834
I should clarify that the amount of time that was available was exactly the same in both conditions,


58
00:04:20,834 --> 00:04:24,506
and in both conditions people got exactly the same amount of feedback.

59
00:04:24,506 --> 00:04:27,631
The only difference is when and how they got it.

60
00:04:29,354 --> 00:04:33,709
And, again, people come up with all sorts of stuff: Creative ideas and crummy ideas,

61
00:04:33,709 --> 00:04:41,923
well executed and poorly executed, and, overall, we’re able to measure, using web analytics,


62
00:04:41,923 --> 00:04:46,617
the click-through rate that people clicked on these advertisements.

63
00:04:46,617 --> 00:04:52,580
And so, over the past several years, we’ve run millions of advertisements out on the web.

64
00:04:52,580 --> 00:04:57,868
And what we see, in aggregate, is that participants who got a parallel design medicine —

65
00:04:57,868 --> 00:05:02,806
who were forced to create multiple alternatives in parallel — had a higher click-through rate:


66
00:05:02,806 --> 00:05:07,168
The ads they created were clicked on more than ads in the serial condition.

67
00:05:07,168 --> 00:05:13,247
And not only that, but the people who clicked on those ads and then went to the site subsequently

68
00:05:13,247 --> 00:05:16,329
spent a whole lot more time on that site

69
00:05:16,329 --> 00:05:21,500
and what this is telling us is that we’re getting the right people through to those ads.

70
00:05:22,039 --> 00:05:27,799
We also had experts — both advertising professionals and clients for this website —

71
00:05:27,799 --> 00:05:34,171
rate the quality of the advertisements and the experts also rated the quality of the parallel ads to be higher.

72
00:05:34,171 --> 00:05:38,720
And then we had the ads rated by a crowd online for the diversity of the ads.

73
00:05:38,720 --> 00:05:42,592
And what we see is that the ads in the parallel condition are also more diverse.

74
00:05:44,330 --> 00:05:48,625
And so why does a parallel approach yield better results?

75
00:05:49,379 --> 00:05:53,754
I think one of the important things that creating multiple alternatives in parallel does,

76
00:05:53,754 --> 00:05:56,904
is it separates your ego from the thing that you make.

77
00:05:56,904 --> 00:06:02,655
If I have only one idea and you critique it, I’m going to treat that as feedback about me;

78
00:06:02,655 --> 00:06:06,412
whereas if I have multiple different ideas and I get critique about them,

79
00:06:06,412 --> 00:06:12,150
I can see that its feedback about the ideas and not a referendum on me as a person,

80
00:06:12,150 --> 00:06:18,083
Also, automatically, by creating multiple alternatives, people are inspired to compare what they’ve created

81
00:06:18,083 --> 00:06:23,213
and try and transfer what they’ve learned from one design as they go forward in the future.

82
00:06:24,029 --> 00:06:27,173
And we see this transfer across a wide variety of domains.

83
00:06:27,173 --> 00:06:30,847
For example, in Dedre Gentner’s research on business negotiation,

84
00:06:30,847 --> 00:06:34,524
she had participants read business school cases,

85
00:06:34,524 --> 00:06:40,286
and she either had people read the cases one at a time and think about each individually,

86
00:06:40,286 --> 00:06:44,555
or she had people read them two at a time and compare them.

87
00:06:44,555 --> 00:06:49,219
And what she found was that having people compare two cases —

88
00:06:49,219 --> 00:06:53,057
to be able to contrast them and see similarities —

89
00:06:53,057 --> 00:06:58,804
yielded to a three-fold increase in the amount of wisdom that they were able to get

90
00:06:58,804 --> 00:07:04,506
out of those cases and transfer to a new negotiation task.

91
00:07:04,506 --> 00:07:10,410
So, what we got out of this is that maybe there’s some big benefits of creating multiple alternatives,

92
00:07:10,410 --> 00:07:13,980
especially for design teams and not just for individual design.

93
00:07:13,980 --> 00:07:18,925
So the next experiment we ran looked at sharing multiple alternatives.

94
00:07:20,064 --> 00:07:23,846
Same basic idea — we have a new client this time.

95
00:07:23,846 --> 00:07:27,048
And we’re going to have people either create and share multiple,

96
00:07:27,048 --> 00:07:29,423
create multiple and share their best,

97
00:07:29,423 --> 00:07:32,319
or create and share only one.

98
00:07:32,319 --> 00:07:35,758
Participants came up with lots of different designs.

99
00:07:35,758 --> 00:07:37,781
And [what] you can see is that

100
00:07:37,781 --> 00:07:42,327
the “share multiple” condition drastically outperforms the other two conditions.

101
00:07:42,327 --> 00:07:48,257
So being able to create and share multiple designs has especially significant benefits for teams.

102
00:07:48,257 --> 00:07:50,826
And there are a number of reasons for this.

103
00:07:50,826 --> 00:07:55,117
I’d like to point out one in particular, which is the increase in group rapport.

104
00:07:55,117 --> 00:07:59,745
When we asked people how they felt about their teammate, both before and after the task,

105
00:07:59,745 --> 00:08:06,784
in the create- and share-one conditions, people felt worse about their teammate afterwards —

106
00:08:06,784 --> 00:08:13,907
the single design approach can create enmity between teammates, and hostility —

107
00:08:13,907 --> 00:08:19,404
whereas, when creating and sharing multiple designs, people felt better about their teammates afterwards.

108
00:08:20,465 --> 00:08:25,290
One important benefit of sharing multiple designs, both with users and with designers,

109
00:08:25,290 --> 00:08:30,898
is that alternatives provide a vocabulary for talking about the space of possible designs.

110
00:08:30,898 --> 00:08:35,072
As Tohidi and colleagues showed, this could be especially valuable for users

111
00:08:35,072 --> 00:08:38,220
because users don’t know what the space of possible designs is.

112
00:08:38,220 --> 00:08:42,407
And so having multiple alternatives gives this vocabulary.

113
00:08:43,268 --> 00:08:46,559
I hope that today’s lecture has provided you with the conceptual tools

114
00:08:46,559 --> 00:08:49,490
for why it’s valuable to create many different alternatives.

115
00:08:49,490 --> 00:08:53,302
And I hope that this will be really useful for you as you go about your design projects.


116
00:08:53,302 --> 00:08:56,640
I’ll see you next time.