Democratizing Innovation by Eric von Hippel - HTML preview

PLEASE NOTE: This is an HTML preview only and some elements such as links or page numbers may be incorrect.
Download the book in PDF, ePub, Kindle for a complete version.
Democratizing Innovation

Eric von Hipel

copy @ web.mit.edu/evhippel/www/

Copyright © 2005 Eric von Hippel.

Exclusive rights to pub-

lish and sell this book in print form in English are licensed to

The MIT Press.

All other rights are reserved by the author.

An electronic version of this book is available under a Creative

Commons license.;

License:

Creative Commons US Attribution-NonCommercial-

NoDerivs license 2.0. ‹http://creativecommons.org/licenses/by-nc-nd/2.

0/legalcode› Some Rights Reserved. You are free to copy, distribute, display and perform the work, under the following conditions: Attribution, you must give the original author credit; you may not

use this work for commercial purposes; No Derivative Works, you

may not alter, transform, or build-upon this work. For reuse or

distribution you must make clear to others the license terms of this

work. Any conditions can be waived if you get permission from

the copyright holder. Your fair use and other rights are in no way

affected by the above.

SiSU

web.mit.edu/evhippel/www/

ii

Contents

Contents

Democratizing Innovation,

Eric von Hipel

1

Attribution . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

Acknowledgements

. . . . . . . . . . . . . . . . . . . . .

2

Democratizing Innovation

3

1 Introduction and Overview . . . . . . . . . . . . . . . . .

3

2 Development of Products by Lead Users

. . . . . . . .

15

3 Why Many Users Want Custom Products . . . . . . . .

23

4 Users' Innovate-or-Buy Decisions

. . . . . . . . . . . .

31

5 Users' Low-Cost Innovation Niches . . . . . . . . . . . .

41

6 Why Users Often Freely Reveal Their Innovations . . .

50

7 Innovation Communities . . . . . . . . . . . . . . . . . .

59

8 Adapting Policy to User Innovation . . . . . . . . . . . .

69

9 Democratizing Innovation . . . . . . . . . . . . . . . . .

77

10 Application: Searching for Lead User Innovations . . .

84

11 Application: Toolkits for User Innovation and Custom

Design . . . . . . . . . . . . . . . . . . . . . . . . . .

93

12 Linking User Innovation to Other Phenomena and Fields105

Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

Bibliography

. . . . . . . . . . . . . . . . . . . . . . . . . 116

Index

127

Metadata

134

SiSU Metadata, document information . . . . . . . . . . . 134

SiSU

web.mit.edu/evhippel/www/

iii

Democratizing Innovation

1

Democratizing Innovation,

Attribution

Eric von Hipel

Dedicated to all who are building the information commons.

SiSU

web.mit.edu/evhippel/www/

1

Democratizing Innovation

Acknowledgements

Networks, John Martin from Verizon, Ben Hyde from the Apache

Foundation, Brian Behlendorf from the Apache Foundation and

Early in my research on the democratization of innovation I was

CollabNet, and Joan Churchill and Susan Hiestand from Lead User

very fortunate to gain five major academic mentors and friends.

Concepts. Thank you so much for the huge (and often humbling)

Nathan Rosenberg, Richard Nelson, Zvi Griliches, Edwin Mans-

insights that your and our field experimentation has provided!

field, and Ann Carter all provided crucial support as I adopted eco-

I am also eager to acknowledge and thank my family for the joy and

nomics as the organizing framework and toolset for my work. Later,

learning they experience and share with me. My wife Jessie is a

I collaborated with a number of wonderful co-authors, all of whom

professional editor and edited my first book in a wonderful way. For

are friends as well: Stan Finkelstein, Nikolaus Franke, Dietmar

this book, however, time devoted to bringing up the children made

Harhoff, Joachim Henkel, Cornelius Herstatt, Ralph Katz, Georg

a renewed editorial collaboration impossible. I hope the reader will

von Krogh, Karim Lakhani, Gary Lilien, Christian Luthje, Pamela

not suffer unduly as a consequence! My children Christiana Dag-

Morrison, William Riggs, John Roberts, Stephan Schrader, Mary

mar and Eric James have watched me work on the book---indeed

Sonnack, Stefan Thomke, Marcie Tyre, and Glen Urban. Other ex-

they could not avoid it as I often write at home. I hope they have

cellent research collaborators and friends of long standing include

been drawing the lesson that academic research can be really fun.

Carliss Baldwin, Sonali Shah, Sarah Slaughter, and Lars Jeppe-

Certainly, that is the lesson I drew from my father, Arthur von Hip-

sen.

pel. He wrote his books in his study upstairs when I was a child and

At some point as interest in a topic grows, there is a transition from

would often come down to the kitchen for a cup of coffee. In transit,

dyadic academic relationships to a real research community. In my

he would throw up his hands and say, to no one in particular, “Why

case, the essential person in enabling that transition was my close

do I choose to work on such difficult problems?” And then he would

friend and colleague Dietmar Harhoff. He began to send wonderful

look deeply happy. Dad, I noticed the smile!

Assistant Professors (Habilitanden) over from his university, Lud-

Finally my warmest thanks to my MIT colleagues and students and

wig Maximilians Universität in Munich, to do collaborative research

also to MIT as an institution. MIT is a really inspiring place to work

with me as MIT Visiting Scholars. They worked on issues related

and learn from others. We all understand the requirements for

to the democratization of innovation while at MIT and then carried

good research and learning, and we all strive to contribute to a

on when they returned to Europe. Now they are training others in

very supportive academic environment. And, of course, new peo-

their turn.

ple are always showing up with new and interesting ideas, so fun

I have also greatly benefited from close contacts with colleagues in

and learning are always being renewed!

industry. As Director of the MIT Innovation Lab, I work together with

senior innovation managers in just a few companies to develop and

try out innovation tools in actual company settings. Close intellec-

tual colleagues and friends of many years standing in this sphere

include Jim Euchner from Pitney-Bowes, Mary Sonnack and Roger

Lacey from 3M, John Wright from IFF, Dave Richards from Nortel

SiSU

web.mit.edu/evhippel/www/

2

Democratizing Innovation

2

Democratizing Innovation

1 Introduction and Overview

3

When I say that innovation is being democratized, I mean that users

4

of products and services---both firms and individual consumers---

are increasingly able to innovate for themselves. User-centered in-

novation processes offer great advantages over the manufacturer-

centric innovation development systems that have been the main-

stay of commerce for hundreds of years. Users that innovate can

develop exactly what they want, rather than relying on manufactur-

ers to act as their (often very imperfect) agents. Moreover, individ-

ual users do not have to develop everything they need on their own:

they can benefit from innovations developed and freely shared by

others.

The trend toward democratization of innovation applies to informa-

5

tion products such as software and also to physical products. As

a quick illustration of the latter, consider the development of high-

performance windsurfing techniques and equipment in Hawaii by

an informal user group. High-performance windsurfing involves ac-

robatics such as jumps and flips and turns in mid-air. Larry Stanley,

a pioneer in high-performance windsurfing, described the develop-

ment of a major innovation in technique and equipment to Sonali

Shah:

In 1978 Jürgen Honscheid came over from West Germany for the

6

first Hawaiian World Cup and discovered jumping, which was new

to him, although Mike Horgan and I were jumping in 1974 and 1975.

There was a new enthusiasm for jumping and we were all trying to

outdo each other by jumping higher and higher. The problem was

that . . . the riders flew off in mid-air because there was no way to

keep the board with you---and as a result you hurt your feet, your

legs, and the board.

Then I remembered the “Chip,” a small experimental board we had

7

built with footstraps, and thought “it's dumb not to use this for jump-

SiSU

web.mit.edu/evhippel/www/

3

Democratizing Innovation

ing.” That's when I first started jumping with footstraps and discov-

The ongoing shift of innovation to users has some very attractive

12

ering controlled flight. I could go so much faster than I ever thought

qualities. It is becoming progressively easier for many users to get

and when you hit a wave it was like a motorcycle rider hitting a

precisely what they want by designing it for themselves. And in-

ramp; you just flew into the air. All of a sudden not only could you

novation by users appears to increase social welfare. At the same

fly into the air, but you could land the thing, and not only that, but

time, the ongoing shift of product-development activities from man-

you could change direction in the air!

ufacturers to users is painful and difficult for many manufacturers.

Open, distributed innovation is “attacking” a major structure of the

8

The whole sport of high-performance windsurfing really started

social division of labor. Many firms and industries must make fun-

from that. As soon as I did it, there were about ten of us who sailed

damental changes to long-held business models in order to adapt.

all the time together and within one or two days there were various

Further, governmental policy and legislation sometimes preferen-

boards out there that had footstraps of various kinds on them, and

tially supports innovation by manufacturers. Considerations of so-

we were all going fast and jumping waves and stuff. It just kind of

cial welfare suggest that this must change. The workings of the

snowballed from there. (Shah 2000)

intellectual property system are of special concern. But despite the

9

By 1998, more than a million people were engaged in windsurf-

difficulties, a democratized and user-centric system of innovation

ing, and a large fraction of the boards sold incorporated the user-

appears well worth striving for.

developed innovations for the high-performance sport.

Users, as the term will be used in this book, are firms or individual

13

10

The user-centered innovation process just illustrated is in sharp

consumers that expect to benefit from using a product or a ser-

contrast to the traditional model, in which products and services

vice. In contrast, manufacturers expect to benefit from selling a

are developed by manufacturers in a closed way, the manufac-

product or a service. A firm or an individual can have different re-

turers using patents, copyrights, and other protections to prevent

lationships to different products or innovations. For example, Boe-

imitators from free riding on their innovation investments. In this

ing is a manufacturer of airplanes, but it is also a user of machine

traditional model, a user's only role is to have needs, which man-

tools. If we were examining innovations developed by Boeing for

ufacturers then identify and fill by designing and producing new

the airplanes it sells, we would consider Boeing a manufacturer-

products. The manufacturer-centric model does fit some fields and

innovator in those cases. But if we were considering innovations in

conditions. However, a growing body of empirical work shows that

metal-forming machinery developed by Boeing for in-house use in

users are the first to develop many and perhaps most new industrial

building airplanes, we would categorize those as user-developed

and consumer products. Further, the contribution of users is grow-

innovations and would categorize Boeing as a user-innovator in

ing steadily larger as a result of continuing advances in computer

those cases.

and communications capabilities.

11

In this book I explain in detail how the emerging process of user-

Innovation user and innovation manufacturer are the two general

14

centric, democratized innovation works. I also explain how innova-

“functional” relationships between innovator and innovation. Users

tion by users provides a very necessary complement to and feed-

are unique in that they alone benefit directly from innovations. All

stock for manufacturer innovation.

others (here lumped under the term “manufacturers”) must sell

SiSU

web.mit.edu/evhippel/www/

4

Democratizing Innovation

innovation-related products or services to users, indirectly or di-

of the majority of users in their populations with respect to an impor-

rectly, in order to profit from innovations. Thus, in order to profit,

tant market trend, and they expect to gain relatively high benefits

inventors must sell or license knowledge related to innovations,

from a solution to the needs they have encountered there. The cor-

and manufacturers must sell products or services incorporating in-

relations found between innovation by users and lead user status

novations. Similarly, suppliers of innovation-related materials or

are highly significant, and the effects are very large.

services---unless they have direct use for the innovations---must

sell the materials or services in order to profit from the innova-

tions.

Since lead users are at the leading edge of the market with respect

20

15

The user and manufacturer categorization of relationships between

to important market trends, one can guess that many of the novel

innovator and innovation can be extended to specific functions, at-

products they develop for their own use will appeal to other users

tributes, or features of products and services. When this is done, it

too and so might provide the basis for products manufacturers

may turn out that different parties are associated with different at-

would wish to commercialize. This turns out to be the case. A num-

tributes of a particular product or service. For example, household-

ber of studies have shown that many of the innovations reported

ers are the users of the switching attribute of a household electric

by lead users are judged to be commercially attractive and/or have

light switch---they use it to turn lights on and off. However, switches

actually been commercialized by manufacturers.

also have other attributes, such as “easy wiring” qualities, that may

be used only by the electricians who install them. Therefore, if

an electrician were to develop an improvement to the installation

attributes of a switch, it would be considered a user-developed in-

Research provides a firm grounding for these empirical findings.

21

novation.

The two defining characteristics of lead users and the likelihood

that they will develop new or modified products have been found

16

A brief overview of the contents of the book follows.

to be highly correlated (Morrison et al. 2004). In addition, it has

17

Development of Products by Lead Users (Chapter 2)

been found that the higher the intensity of lead user characteristics

displayed by an innovator, the greater the commercial attractive-

18

Empirical studies show that many users---from 10 percent to nearly

ness of the innovation that the lead user develops (Franke and von

40 percent---engage in developing or modifying products. About

Hippel 2003a). In figure 1.1, the increased concentration of inno-

half of these studies do not determine representative innovation

vations toward the right indicates that the likelihood of innovating is

frequencies; they were designed for other purposes. Nonetheless,

higher for users having higher lead user index values. The rise in

when taken together, the findings make it very clear that users

average innovation attractiveness as one moves from left to right

are doing a lot of product modification and product development

indicates that innovations developed by lead users tend to be more

in many fields.

commercially attractive. (Innovation attractiveness is the sum of

19

Studies of innovating users (both individuals and firms) show them

the novelty of the innovation and the expected future generality of

to have the characteristics of “lead users.” That is, they are ahead

market demand.)

SiSU

web.mit.edu/evhippel/www/

5

index-9_1.png

Democratizing Innovation

well enough to induce purchase from and capture significant profits

from a large number of customers. When users' needs are hetero-

geneous, this strategy of “a few sizes fit all” will leave many users

somewhat dissatisfied with the commercial products on offer and

probably will leave some users seriously dissatisfied. In a study of

a sample of users of the security features of Apache web server

software, Franke and von Hippel (2003b) found that users had a

very high heterogeneity of need, and that many had a high willing-

ness to pay to get precisely what they wanted. Nineteen percent of

the users sampled actually innovated to tailor Apache more closely

to their needs. Those who did were found to be significantly more

satisfied.

Users' Innovate-or-Buy Decisions (Chapter 4)

27

22

Even if many users want “exactly right products” and are willing

28

and able to pay for their development, why do users often do this

23

Figure 1.1 User-innovators with stronger “lead user” characteris-

for themselves rather than hire a custom manufacturer to develop

tics develop innovations having higher appeal in the general mar-

a special just-right product for them? After all, custom manufactur-

ketplace. Estimated OLS function: Y = 2.06 + 0.57x, where Y rep-

ers specialize in developing products for one or a few users. Since

resents attractiveness of innovation and x represents lead-user-

these firms are specialists, it is possible that they could design and

ness of respondent. Adjusted R2 = 0.281; p = 0.002; n = 30.

build custom products for individual users or user firms faster, bet-

Source of data: Franke and von Hippel 2003.

ter, or cheaper than users could do this for themselves. Despite

this possibility, several factors can drive users to innovate rather

24

Why Many Users Want Custom Products (Chapter 3)

than buy. Both in the case of user firms and in the case of individ-

25

Why do so many users develop or modify products for their own

ual user-innovators, agency costs play a major role. In the case

use? Users may innovate if and as they want something that is

of individual user-innovators, enjoyment of the innovation process

not available on the market and are able and willing to pay for its

can also be important.

development. It is likely that many users do not find what they

With respect to agency costs, consider that when a user develops

want on the market. Meta-analysis of market-segmentation studies

29

its own custom product that user can be trusted to act in its own best

suggests that users' needs for products are highly heterogeneous

interests. When a user hires a manufacturer to develop a custom

in many fields (Franke and Reisinger 2003).

product, the situation is more complex. The user is then a principal

26

Mass manufacturers tend to follow a strategy of developing prod-

that has hired the custom manufacturer to act as its agent. If the

ucts that are designed to meet the needs of a large market segment

interests of the principal and the agent are not the same, there will

SiSU

web.mit.edu/evhippel/www/

6

Democratizing Innovation

be agency costs. In general terms, agency costs are (1) costs in-

shows in a quantitative way that user firms with unique needs will

curred to monitor the agent to ensure that it (or he or she) follows

always be better off developing new products for themselves. It

the interests of the principal, (2) the cost incurred by the agent to

also shows that development by manufacturers can be the most

commit itself not to act against the principal's interest (the “bonding

economical option when n or more user firms want the same thing.

cost”), and (3) costs associated with an outcome that does not fully

However, when the number of user firms wanting the same thing

serve the interests of the principal (Jensen and Meckling 1976). In

falls between 1 and n, manufacturers may not find it profitable to

the specific instance of product and service development, a ma-

develop a new product for just a few users. In that case, more than

jor divergence of interests between user and custom manufacturer

one user may invest in developing the same thing independently,

does exist: the user wants to get precisely what it needs, to the

owing to market failure. This results in a waste of resources from

extent that it can afford to do so. In contrast, the custom manu-

the point of view of social welfare. The problem can be addressed

facturer wants to lower its development costs by incorporating so-

by new institutional forms, such as the user innovation communities

lution elements it already has or that it predicts others will want in

that will be studied later in this book.

the future---even if by doing so it does not serve its present client's

needs as well as it could.

Chapter 4 concludes by pointing out that an additional incentive

32

can drive individual user-innovators to innovate rather than buy:

30

A user wants to preserve its need specification because that spec-

ification is chosen to make that user's overall solution quality as

they may value the process of innovating because of the enjoy-

high as possible at the desired price. For example, an individ-

ment or learning that it brings them. It might seem strange that

ual user may specify a mountain-climbing boot that will precisely

user-innovators can enjoy product development enough to want to

fit his unique climbing technique and allow him to climb Everest

do it themselves---after all, manufacturers pay their product devel-

more easily. Any deviations in boot design will require compensat-

opers to do such work! On the other hand, it is also clear that enjoy-

ing modifications in the climber's carefully practiced and deeply in-

ment of problem solving is a motivator for many individual problem

grained climbing technique---a much more costly solution from the

solvers in at least some fields. Consider for example the millions of

user's point of view. A custom boot manufacturer, in contrast, will

crossword-puzzle aficionados. Clearly, for these individuals enjoy-

have a strong incentive to incorporate the materials and processes

ment of the problem-solving process rather than the solution is the

it has in stock and expects to use in future even if this produces

goal. One can easily test this by attempting to offer a puzzle solver

a boot that is not precisely right for the present customer. For ex-

a completed puzzle---the very output he or she is working so hard

ample, the manufacturer will not want to learn a new way to bond

to create. One will very likely be rejected with the rebuke that one

boot components together even if that would produce the best cus-

should not spoil the fun! Pleasure as a motivator can apply to the

tom result for one client. The net result is that when one or a few

development of commercially useful innovations as well. Studies

users want something special they will often get the best result by

of the motivations of volunteer contributors of code to widely used

innovating for themselves.

software products have shown that these individuals too are often

strongly motivated to innovate by the joy and learning they find in

31

A small model of the innovate-or-buy decision follows. This model

this work (Hertel et al. 2003; Lakhani and Wolf 2005).

SiSU

web.mit.edu/evhippel/www/

7

Democratizing Innovation

33

Users' Low-Cost Innovation Niches (Chapter 5)

will also differ among individual users and manufacturers. The in-

formation assets of some particular user (or some particular man-

34

An exploration of the basic processes of product and service devel-

ufacturer) will be closest to what is required to develop a particular

opment show that users and manufacturers tend to develop differ-

innovation, and so the cost of developing that innovation will be rel-

ent types of innovations. This is due in part to information asymme-

atively low for that user or manufacturer. The net result is that user

tries: users and manufacturers tend to know different things. Prod-

innovation activities will be distributed across many users accord-

uct developers need two types of information in order to succeed

ing to their information endowments. With respect to innovation,

at their work: need and context-of-use information (generated by

one user is by no means a perfect substitute for another.

users) and generic solution information (often initially generated by

manufacturers specializing in a particular type of solution). Bring-

Why Users Often Freely Reveal Their Innovations (Chapter

37

ing these two types of information together is not easy. Both need

6)

information and solution information are often very “sticky”---that is,

The social efficiency of a system in which individual innovations are

costly to move from the site where the information was generated

38

developed by individual users is increased if users somehow dif-

to other sites. As a result, users generally have a more accurate

fuse what they have developed to others. Manufacturer-innovators

and more detailed model of their needs than manufacturers have,

partially achieve this when they sell a product or a service on the

while manufacturers have a better model of the solution approach

open market (partially because they diffuse the product incorpo-

in which they specialize than the user has.

rating the innovation, but often not all the information that others

35

When information is sticky, innovators tend to rely largely on infor-

would need to fully understand and replicate it). If user-innovators

mation they already have in stock. One consequence of the infor-

do not somehow also diffuse what they have done, multiple users

mation asymmetry between users and manufacturers is that users

with very similar needs will have to independently develop very

tend to develop innovations that are functionally novel, requiring

similar innovations---a poor use of resources from the viewpoint

a great deal of user-need information and use-context information

of social welfare. Empirical research shows that users often do

for their development. In contrast, manufacturers tend to develop

achieve widespread diffusion by an unexpected means: they often

innovations that are improvements on well-known needs and that

“freely reveal” what they have developed. When we say that an in-

require a rich understanding of solution information for their de-

novator freely reveals information about a product or service it has

velopment. For example, firms that use inventory-management

developed, we mean that all intellectual property rights to that infor-

systems, such as retailers, tend to be the developers of new ap-

mation are voluntarily given up by the innovator, and all interested

proaches to inventory management. In contrast, manufacturers of

parties are given access to it---the information becomes a public

inventory-management systems and equipment tend to develop

good.

improvements to the equipment used to implement these user-

The empirical finding that users often freely reveal their innova-

devised approaches (Ogawa 1998).

39

tions has been a major surprise to innovation researchers. On the

36

If we extend the information-asymmetry argument one step further,

face of it, if a user-innovator's proprietary information has value to

we see that information stickiness implies that information on hand

others, one would think that the user would strive to prevent free

SiSU

web.mit.edu/evhippel/www/

8

Democratizing Innovation

diffusion rather than help others to free ride on what it has devel-

novators with significant private benefits as well as losses or risks

oped at private cost. Nonetheless, it is now very clear that individ-

of loss. Users who freely reveal what they have done often find that

ual users and user firms---and sometimes manufacturers---often

others then improve or suggest improvements to the innovation, to

freely reveal detailed information about their innovations.

mutual benefit (Raymond 1999). Freely revealing users also may

benefit from enhancement of reputation, from positive network ef-

40

The practices visible in “open source” software development were

fects due to increased diffusion of their innovation, and from other

important in bringing this phenomenon to general awareness. In

factors. Being the first to freely reveal a particular innovation can

these projects it was clear policy that project contributors would

also enhance the benefits received, and so there can actually be a

routinely and systematically freely reveal code they had developed

rush to reveal, much as scientists rush to publish in order to gain

at private expense (Raymond 1999). However, free revealing of

the benefits associated with being the first to have made a partic-

product innovations has a history that began long before the advent

ular advancement.

of open source software. Allen, in his 1983 study of the eighteenth-

century iron industry, was probably the first to consider the phe-

Innovation Communities (Chapter 7)

43

nomon systematically. Later, Nuvolari (2004) discussed free re-

Innovation by users tends to be widely distributed rather than con-

vealing in the early history of mine pumping engines. Contem-

44

centrated among just a very few very innovative users. As a result,

porary free revealing by users has been documented by von Hip-

it is important for user-innovators to find ways to combine and lever-

pel and Finkelstein (1979) for medical equipment, by Lim (2000)

age their efforts. Users achieve this by engaging in many forms of

for semiconductor process equipment, by Morrison, Roberts, and

cooperation. Direct, informal user-to-user cooperation (assisting

von Hippel (2000) for library information systems, and by Franke

others to innovate, answering questions, and so on) is common.

and Shah (2003) for sporting equipment. Henkel (2003) has doc-

Organized cooperation is also common, with users joining together

umented free revealing among manufacturers in the case of em-

in networks and communities that provide useful structures and

bedded Linux software.

tools for their interactions and for the distribution of innovations.

41

Innovators often freely reveal because it is often the best or the

Innovation communities can increase the speed and effectiveness

only practical option available to them. Hiding an innovation as a

with which users and also manufacturers can develop and test and

trade secret is unlikely to be successful for long: too many generally

diffuse their innovations. They also can greatly increase the ease

know similar things, and some holders of the “secret” information

with which innovators can build larger systems from interlinkable

stand to lose little or nothing by freely revealing what they know.

modules created by community participants.

Studies find that innovators in many fields view patents as having

Free and open source software projects are a relatively well-

only limited value. Copyright protection and copyright licensing are

45

developed and very successful form of Internet-based innovation

applicable only to “writings,” such as books, graphic images, and

community. However, innovation communities are by no means

computer software.

restricted to software or even to information products, and they

42

Active efforts by innovators to freely reveal---as opposed to sullen

can play a major role in the development of physical products.

acceptance---are explicable because free revealing can provide in-

Franke and Shah (2003) have documented the value that user

SiSU

web.mit.edu/evhippel/www/

9

Democratizing Innovation

innovation communities can provide to user-innovators developing

of past and current policy decisions. Research done in the past

physical products in the field of sporting equipment. The analogy

30 years has convinced many academics that intellectual property

to open source innovation communities is clear.

law is sometimes or often not having its intended effect. Intellec-

tual property law was intended to increase the amount of innova-

46

The collective or community effort to provide a public good---which

tion investment. Instead, it now appears that there are economies

is what freely revealed innovations are---has traditionally been ex-

of scope in both patenting and copyright that allow firms to use

plored in the literature on “collective action.” However, behaviors

these forms of intellectual property law in ways that are directly

seen in extant innovation communities fail to correspond to that

opposed to the intent of policy makers and to the public welfare.

literature at major points. In essence, innovation communities ap-

Major firms can invest to develop large portfolios of patents. They

pear to be more robust with respect to recruiting and rewarding

can then use these to create “patent thickets”---dense networks

members than the literature would predict. Georg von Krogh and I

of patent claims that give them plausible grounds for threatening

attribute this to innovation contributors' obtaining some private re-

to sue across a wide range of intellectual property. They may do

wards that are not shared equally by free riders (those who take

this to prevent others from introducing a superior innovation and/or

without contributing). For example, a product that a user-innovator

to demand licenses from weaker competitors on favorable terms

develops and freely reveals might be perfectly suited to that user-

(Shapiro 2001). Movie, publishing, and software firms can use

innovator's requirements but less well suited to the requirements

large collections of copyrighted work to a similar purpose (Benkler

of free riders. Innovation communities thus illustrate a “private-

2002). In view of the distributed nature of innovation by users, with

collective” model of innovation incentive (von Hippel and von Krogh

each tending to create a relatively small amount of intellectual prop-

2003).

erty, users are likely to be disadvantaged by such strategies.

47

Adapting Policy to User Innovation (Chapter 8)

It is also important to note that users (and manufacturers) tend

50

48

Is innovation by users a “good thing?” Welfare economists answer

to build prototypes of their innovations economically by modifying

such a question by studying how a phenomenon or a change af-

products already available on the market to serve a new purpose.

fects social welfare. Henkel and von Hippel (2005) explored the so-

Laws such as the (US) Digital Millennium Copyright Act, intended

cial welfare implications of user innovation. They found that, rela-

to prevent consumers from illegally copying protected works, also

tive to a world in which only manufacturers innovate, social welfare

can have the unintended side effect of preventing users from modi-

is very probably increased by the presence of innovations freely

fying products that they purchase (Varian 2002). Both fairness and

revealed by users. This finding implies that policy making should

social welfare considerations suggest that innovation-related poli-

support user innovation, or at least should ensure that legislation

cies should be made neutral with respect to the sources of innova-

and regulations do not favor manufacturers at the expense of user-

tion.

innovators.

49

The transitions required of policy making to achieve neutrality with

It may be that current impediments to user innovation will be solved

51

respect to user innovation vs. manufacturer innovation are sig-

by legislation or by policy making. However, beneficiaries of exist-

nificant. Consider the impact on open and distributed innovation

ing law and policy will predictably resist change. Fortunately, a

SiSU

web.mit.edu/evhippel/www/

10

Democratizing Innovation

way to get around some of these problems is in the hands of inno-

need and willingness to invest in obtaining a precisely right product

vators themselves. Suppose many innovators in a particular field

remains constant.

decide to freely reveal what they have developed, as they often

Equivalents of the innovation resources described above have long

have reason to do. In that case, users can collectively create an

55

been available within corporations to a few. Senior designers at

information commons (a collection of information freely available

firms have long been supplied with engineers and designers un-

to all) containing substitutes for some or a great deal of informa-

der their direct control, and with the resources needed to quickly

tion now held as private intellectual property. Then user-innovators

construct and test prototype designs. The same is true in other

can work around the strictures of intellectual property law by sim-

fields, including automotive design and clothing design: just think

ply using these freely revealed substitutes (Lessig 2001). This is

of the staffs of engineers and modelmakers supplied so that top

essentially what is happening in the field of software. For many

auto designers can quickly realize and test their designs.

problems, user-innovators in that field now have a choice between

proprietary, closed software provided by Microsoft and other firms

But if, as we have seen, the information needed to innovate in im-

56

and open source software that they can legally download from the

portant ways is widely distributed, the traditional pattern of concen-

Internet and legally modify to serve their own specific needs.

trating innovation-support resources on a few individuals is hugely

inefficient. High-cost resources for innovation support cannot effi-

52

Policy making that levels the playing field between users and man-

ciently be allocated to “the right people with the right information:”

ufacturers will force more rapid change onto manufacturers but will

it is very difficult to know who these people may be before they

by no means destroy them. Experience in fields where open and

develop an innovation that turns out to have general value. When

distributed innovation processes are far advanced show how man-

the cost of high-quality resources for design and prototyping be-

ufacturers can and do adapt. Some, for example, learn to supply

comes very low (the trend we have described), these resources

proprietary platform products that offer user-innovators a frame-

can be diffused very widely, and the allocation problem diminishes

work upon which to develop and use their improvements.

in significance. The net result is and will be to democratize the

53

Democratizing Innovation (Chapter 9)

opportunity to create.

54

Users' ability to innovate is improving radically and rapidly as a

On a level playing field, users will be an increasingly important

57

result of the steadily improving quality of computer software and

source of innovation and will increasingly substitute for or comple-

hardware, improved access to easy-to-use tools and components

ment manufacturers' innovation-related activities. In the case of

for innovation, and access to a steadily richer innovation commons.

information products, users have the possibility of largely or com-

Today, user firms and even individual hobbyists have access to

pletely doing without the services of manufacturers. Open source

sophisticated programming tools for software and sophisticated

software projects are object lessons that teach us that users can

CAD design tools for hardware and electronics. These information-

create, produce, diffuse, provide user field support for, update, and

based tools can be run on a personal computer, and they are

use complex products by and for themselves in the context of user

rapidly coming down in price. As a consequence, innovation by

innovation communities. In physical product fields, product devel-

users will continue to grow even if the degree of heterogeneity of

opment by users can evolve to the point of largely or totally sup-

SiSU

web.mit.edu/evhippel/www/

11

Democratizing Innovation

planting product development---but not product manufacturing---by

plore the needs of users in the target market, product-development

manufacturers. (The economies of scale associated with manu-

groups to think up suitable products to address those needs, and so

facturing and distributing physical products give manufacturers an

forth. The needs and prototype solutions of lead users---if encoun-

advantage over “do-it-yourself” users in those activities.)

tered at all---are typically rejected as outliers of no interest. Indeed,

when lead users' innovations do enter a firm's product line---and

58

The evolving pattern of the locus of product development in

they have been shown to be the actual source of many major inno-

kitesurfing illustrates how users can displace manufacturers from

vations for many firms--- they typically arrive with a lag and by an

the role of product developer.

In that industry, the collective

unconventional and unsystematic route. For example, a manufac-

product-design and testing work of a user innovation community

turer may “discover” a lead user innovation only when the innovat-

has clearly become superior in both quality and quantity relative

ing user firm contacts the manufacturer with a proposal to produce

to the levels of in-house development effort that manufacturers

its design in volume to supply its own in-house needs. Or sales or

of kitesurfing equipment can justify. Accordingly, manufacturers

service people employed by a manufacturer may spot a promising

of such equipment are increasingly shifting away from product

prototype during a visit to a customer's site.

design and focusing on producing product designs first developed

and tested by user innovation communities.

Modification of firms' innovation processes to systematically search

62

for and further develop innovations created by lead users can pro-

59

How can or should manufacturers adapt to users' encroachment on

elements of their traditional business activities? There are three

vide manufacturers with a better interface to the innovation process

general possibilities: (1) Produce user-developed innovations for

as it actually works, and so provide better performance. A natu-

general commercial sale and/or offer custom manufacturing to spe-

ral experiment conducted at 3M illustrates this possibility. Annual

cific users. (2) Sell kits of product-design tools and/or “product plat-

sales of lead user product ideas generated by the average lead

forms” to ease users' innovation-related tasks. (3) Sell products or

user project at 3M were conservatively forecast by management

services that are complementary to user-developed innovations.

to be more than 8 times the sales forecast for new products devel-

Firms in fields where users are already very active in product de-

oped in the traditional manner---$146 million versus $18 million per

sign are experimenting with all these possibilities.

year. In addition, lead user projects were found to generate ideas

for new product lines, while traditional market-research methods

60

Application: Searching for Lead User Innovations (Chapter

were found to produce ideas for incremental improvements to ex-

10)

isting product lines. As a consequence, 3M divisions funding lead

61

Manufacturers design their innovation processes around the way

user project ideas experienced their highest rate of major product

they think the process works. The vast majority of manufacturers

line generation in the past 50 years (Lilien et al. 2002).

still think that product development and service development are

Application: Toolkits for User Innovation and Custom Design

63

always done by manufacturers, and that their job is always to find a

(Chapter 11)

need and fill it rather than to sometimes find and commercialize an

innovation that lead users have already developed. Accordingly,

Firms that understand the distributed innovation process and

64

manufacturers have set up market-research departments to ex-

users' roles in it can change factors affecting lead user innovation

SiSU

web.mit.edu/evhippel/www/

12

Democratizing Innovation

and so affect its rate and direction in ways they value. Toolkits for

In chapter 12 I discuss links between user innovation and some

67

user innovation custom design offer one way of doing this. This

related phenomena and literatures. With respect to phenomena,

approach involves partitioning product-development and service-

I point out the relationship of user innovation to information com-

development projects into solution-information-intensive subtasks

munities, of which user innovation communities are a subset. One

and need-information-intensive subtasks.

Need-intensive sub-

open information community is the online encyclopedia Wikipedia

tasks are then assigned to users along with a kit of tools that

(www.wikipedia.org). Other such communities include the many

enable them to effectively execute the tasks assigned to them.

specialized Internet sites where individuals with both common and

The resulting co-location of sticky information and problem-solving

rare medical conditions can find one another and can find special-

activity makes innovation within the solution space offered by a

ists in those conditions. Many of the advantages associated with

particular toolkit cheaper for users. It accordingly attracts them to

user innovation communities also apply to open information net-

the toolkit and so influences what they develop and how they de-

works and communities. Analyses appropriate to information com-

velop it. The custom semiconductor industry was an early adopter

munities follow the same overall pattern as the analyses provided

of toolkits. In 2003, more than $15 billion worth of semiconductors

in this book for innovation communities. However, they are also

were produced that had been designed using this approach.

simpler, because in open information communities there may be

little or no proprietary information being transacted and thus little

65

Manufacturers that adopt the toolkit approach to supporting and

or no risk of related losses for participants.

channeling user innovation typically face major changes in their

Next I discuss links between user-centric innovation phenomena

business models, and important changes in industry structure may

68

and the literature on the economics of knowledge that have been

also follow. For example, as a result of the introduction of toolkits

forged by Foray (2004) and Weber (2004). I also discuss how

to the field of semiconductor manufacture, custom semiconductor

Porter's 1991 work on the competitive advantage of nations can

manufacturers---formerly providers of both design and manufactur-

be extended to incorporate findings on nations' lead users as prod-

ing services to customers---lost much of the work of custom product

uct developers. Finally, I point out how findings explained in this

design to customers. Many of these manufacturers then became

book link to and complement research on the Social Construction

specialist silicon foundries, supplying production services primarily.

of Technology (Pinch and Bijker 1987).

Manufacturers may or may not wish to make such changes. How-

ever, experience in fields where toolkits have been deployed shows

I conclude this introductory chapter by reemphasizing that user

69

that customers tend to prefer designing their own custom products

innovation, free revealing, and user innovation communities will

with the aid of a toolkit over traditional manufacturer-centric devel-

flourish under many but not all conditions. What we know about

opment practices. As a consequence, the only real choice for man-

manufacturer-centered innovation is still valid; however, lead-user-

ufacturers in a field appropriate to the deployment of toolkits may

centered innovation patterns are increasingly important, and they

be whether to lead or to follow in the transition to toolkits.

present major new opportunities and challenges for us all.

66

Linking User Innovation to Other Phenomena and Fields

(Chapter 12)

SiSU

web.mit.edu/evhippel/www/

13

Democratizing Innovation

whether you innovated or not, you might be more inclined to re-

spond if your answer is “Yes.”). Also, each of the studies looked

at innovation rates affecting a particular product type among users

who care a great deal about that product type. Thus, university sur-

geons (study 4 in table 2.1) care a great deal about having just-right

surgical equipment, just as serious mountain bikers (study 8) care

a great deal about having just-right equipment for their sport. As

the intensity of interest goes down, it is likely that rates of user inno-

vation drop too. This is probably what is going on in the case of the

study of purchasers of outdoor consumer products (study 6). All we

are told about that sample of users of outdoor consumer products

is that they are recipients of one or more mail order catalogs from

suppliers of relatively general outdoor items---winter jackets, sleep-

ing bags, and so on. Despite the fact that these users were asked

if they have developed or modified any item in this broad category

of goods (rather than a very specific one such as a mountain bike),

just 10 percent answered in the affirmative. Of course, 10 percent

or even 5 percent of a user population numbering in the tens of

millions worldwide is still a very large number---so we again realize

that many users are developing and modifying products.

Table 2.1 Many respondents reported developing or modifying

76

products for their own use in the eight product areas listed

here.

77

Number and type of Users Sampled

Percentage

Source

developing

and

building

product

for

own use

Industrial products

1.

Printed circuit

136 user firm attendees at PC-CAD confer-

24.3%

Urban and von Hip-

CAD software

ence

pel 1988

2. Pipe hanger hard-

Employees in 74 pipe hanger installation firms

36%

Herstatt

and

von

ware

Hippel 1992

SiSU

web.mit.edu/evhippel/www/

14

Democratizing Innovation

Number and type of Users Sampled

Percentage

Source

70

2 Development of Products by Lead Users

developing

and

building

product

for

own use

71

The idea that novel products and services are developed by man-

3.

Library informa-

Employees in 102 Australian libraries using

26%

Morrison et al. 2000

ufacturers is deeply ingrained in both traditional expectations and

tion systems

computerized OPAC library information sys-

tems

scholarship. When we as users of products complain about the

4.

Surgical equip-

261 surgeons working in university clinics in

22%

Lüthje 2003

shortcomings of an existing product or wish for a new one, we

ment

Germany

commonly think that “they” should develop it---not us. Even the

5.

Apache

OS

131 technically sophisticated Apache fea-

19.1%

Franke and von Hip-

server

software

tures users (webmasters)

pel 2003

conventional term for an individual end user, “consumer,” implicitly

security features

suggests that users are not active in product and service develop-

Consumer products

ment. Nonetheless, there is now very strong empirical evidence

6.

Outdoor con-

153 recipients of mail order catalogs for out-

9.8%

Lüthje 2004

sumer products

door activity products for consumers

that product development and modification by both user firms and

7. "Extreme" sport-

197 members of 4 specialized sporting clubs

37.8%

Franke

and

Shah

users as individual consumers is frequent, pervasive, and impor-

ing equipment

in 4 "extreme" sports

2003

tant.

8.

Mountain biking

291 mountain bikers in a geographic region

19.2%

Lüthje et al.

equipment

72

I begin this chapter by reviewing the evidence that many users in-

deed do develop and modify products for their own use in many

fields. I then show that innovation is concentrated among lead

The cited studies also do not set an upper or a lower bound on

78

users, and that lead users' innovations often become commercial

the commercial or technical importance of user-developed prod-

products.

ucts and product modifications that they report, and it is likely that

most are of minor significance. However, most innovations from

73

Many Users Innovate

any source are minor, so user-innovators are no exception in this

74

The evidence on user innovation frequency and pervasiveness is

regard. Further, to say an innovation is minor is not the same as

summarized in table 2.1. We see here that the frequency with

saying it is trivial: minor innovations are cumulatively responsible

which user firms and individual consumers develop or modify prod-

for much or most technical progress. Hollander (1965) found that

ucts for their own use range from 10 percent to nearly 40 percent

about 80 percent of unit cost reductions in Rayon manufacture were

in fields studied to date. The matter has been studied across a

the cumulative result of minor technical changes. Knight (1963, VII,

wide range of industrial product types where innovating users are

pp. 2--3) measured performance advances in general-purpose dig-

user firms, and also in various types of sporting equipment, where

ital computers and found, similarly, that “these advances occur as

innovating users are individual consumers.

the result of equipment designers using their knowledge of elec-

tronics technology to produce a multitude of small improvements

75

The studies cited in table 2.1 clearly show that a lot of product de-

that together produce significant performance advances.”

velopment and modification by users is going on. However, these

findings should not be taken to reflect innovation rates in overall

Although most products and product modifications that users or

79

populations of users. All of the studies probably were affected by

others develop will be minor, users are by no means restricted to

a response bias. (That is, if someone sends a questionnaire about

developing minor or incremental innovations. Qualitative obser-

SiSU

web.mit.edu/evhippel/www/

15

Democratizing Innovation

vations have long indicated that important process improvements

tions (and the most commercially attractive ones) are developed

are developed by users. Smith (1776, pp. 11--13) pointed out

by users with “lead user” characteristics. Recall from chapter 1

the importance of “the invention of a great number of machines

that lead users are defined as members of a user population hav-

which facilitate and abridge labor, and enable one man to do the

ing two distinguishing characteristics: (1) They are at the leading

work of many.” He also noted that “a great part of the machines

edge of an important market trend(s), and so are currently experi-

made use of in those manufactures in which labor is most sub-

encing needs that will later be experienced by many users in that

divided, were originally the invention of common workmen, who,

market. (2) They anticipate relatively high benefits from obtaining

being each of them employed in some very simple operation, nat-

a solution to their needs, and so may innovate.

urally turned their thoughts towards finding out easier and readier

methods of performing it.” Rosenberg (1976) studied the history of

The theory that led to defining “lead users” in terms of these two

82

the US machine tool industry and found that important and basic

characteristics was derived as follows (von Hippel 1986). First, the

machine types like lathes and milling machines were first devel-

“ahead on an important market trend” variable was included be-

oped and built by user firms having a strong need for them. Tex-

cause of its assumed effect on the commercial attractiveness of in-

tile manufacturing firms, gun manufacturers and sewing machine

novations developed by users residing at a leading-edge position in

manufacturers were important early user-developers of machine

a market. Market needs are not static---they evolve, and often they

tools. Other studies show quantitatively that some of the most im-

are driven by important underlying trends. If people are distributed

portant and novel products and processes have been developed

with respect to such trends as diffusion theory indicates, then peo-

by user firms and by individual users. Enos (1962) reported that

ple at the leading edges of important trends will be experiencing

nearly all the most important innovations in oil refining were devel-

needs today (or this year) that the bulk of the market will experi-

oped by user firms. Freeman (1968) found that the most widely

ence tomorrow (or next year). And, if users develop and modify

licensed chemical production processes were developed by user

products to satisfy their own needs, then the innovations that lead

firms. Von Hippel (1988) found that users were the developers of

users develop should later be attractive to many. The expected

about 80 percent of the most important scientific instrument inno-

benefits variable and its link to innovation likelihood was derived

vations, and also the developers of most of the major innovations

from studies of industrial product and process innovations. These

in semiconductor processing. Pavitt (1984) found that a consid-

showed that the greater the benefit an entity expects to obtain from

erable fraction of invention by British firms was for in-house use.

a needed innovation, the greater will be that entity's investment in

Shah (2000) found that the most commercially important equip-

obtaining a solution, where a solution is an innovation either devel-

ment innovations in four sporting fields tended to be developed by

oped or purchased (Schmookler 1966; Mansfield 1968).

individual users.

Empirical studies to date have confirmed lead user theory. Mor-

83

rison, Roberts, and Midgely (2004) studied the characteristics of

80

Lead User Theory

innovating and non-innovating users of computerized library infor-

81

A second major finding of empirical research into innovation by

mation systems in a sample of Australian libraries. They found

users is that most user-developed products and product modifica-

that the two lead user characteristics were distributed in a contin-

SiSU

web.mit.edu/evhippel/www/

16

Democratizing Innovation

uous, unimodal manner in that sample. They also found that the

ing edge of the density trend, Urban and von Hippel collected

two characteristics of lead users and the actual development of

a sample of 138 user-firm employees who had attended a trade

innovations by users were highly correlated. Franke and von Hip-

show on the topic of PC-CAD. To learn the position of each firm

pel (2003b) confirmed these findings in a study of innovating and

on the density trend, they asked questions about the density of

non-innovating users of Apache web server software. They also

the boards that each PC-CAD user firm was currently producing.

found that the commercial attractiveness of innovations developed

To learn about each user's likely expected benefits from improve-

by users increased along with the strength of those users' lead user

ments to PC-CAD, they asked questions about how satisfied each

characteristics.

respondent was with their firm's present PC-CAD capabilities. To

learn about users' innovation activities, they asked questions about

84

Evidence of Innovation by Lead Users

whether each firm had modified or built its own PC-CAD software

85

Several studies have found that user innovation is largely the

for its own in-house use.

province of users that have lead user characteristics, and that

products lead users develop often form the basis for commercial

Users' responses were cluster analyzed, and clear lead user (n =

89

products. These general findings appear robust: the studies have

38) and non-lead-user (n = 98) clusters were found. Users in the

used a variety of techniques and have addressed a variety of

lead user cluster were those that made the densest boards on aver-

markets and innovator types. Brief reviews of four studies will

age and that also were dissatisfied with their PC-CAD capabilities.

convey the essence of what has been found.

In other words, they were at the leading edge of an important mar-

ket trend, and they had a high incentive to innovate to improve their

86

Innovation in Industrial Product User Firms

capabilities. Strikingly, 87 percent of users in the lead user clus-

ter reported either developing or modifying the PC-CAD software

87

In the first empirical study of lead users' role in innovation, Urban

and von Hippel (1988) studied user innovation activity related to a

that they used. In contrast, only 1 percent of non-lead users re-

type of software used to design printed circuit boards. A major mar-

ported this type of innovation. Clearly, in this case user innovation

ket trend to which printed circuit computer-aided design software

was very strongly concentrated in the lead user segment of the

(PC-CAD) must respond is the steady movement toward packing

user population. A discriminant analysis on indicated that “build

electronic circuitry more densely onto circuit boards. Higher density

own system” was the most important indicator of membership in

means one that can shrink boards in overall size and that enables

the lead user cluster. The discriminant analysis had 95.6 percent

the circuits they contain to operate faster---both strongly desired

correct classification of cluster membership.

attributes. Designing a board at the leading edge of what is techni-

The commercial attractiveness of PC-CAD solutions developed by

cally attainable in density at any particular time is a very demanding

90

lead users was high. This was tested by determining whether lead

task. It involves some combination of learning to make the printed

users and more ordinary users preferred a new PC-CAD system

circuit wires narrower, learning how to add more layers of circuitry

concept containing features developed by lead users over the best

to a board, and using smaller electronic components.

commercial PC-CAD system available at the time of the study (as

88

To explore the link between user innovation and needs at the lead-

determined by a large PC-CAD system manufacturer's competitive

SiSU

web.mit.edu/evhippel/www/

17

Democratizing Innovation

analysis) and two additional concepts. The concept containing lead

making it difficult for staff and patrons to find books without pre-

user features was significantly preferred at even twice the price (p

cise directions. There was little duplication of innovations except

< 0.01).

in the case of adding Internet search capabilities to OPACs. In that

unusual case, nine libraries went ahead and did the programming

91

Innovation in Libraries

needed to add this important feature in advance of its being offered