Two-thirds of the approximately 700 software entrepreneurs who participated in the 2008 Berkeley Patent Survey report that they neither have nor are seeking patents for innovations embodied in their products and services. These entrepreneurs rate patents as the least important mechanism among seven options for attaining competitive advantage in the marketplace. Even software startups that hold patents regard them as providing only a slight incentive to invest in innovation.

These are three of the most striking findings from our recently published article, “High Technology Entrepreneurs and the Patent System: Results of the 2008 Berkeley Patent Survey.”

After providing some background about the survey, this column will discuss some key findings about how software startup firms perceive, use and are affected by the patent system.

While the three findings highlighted above might seem to support a software patent abolitionist position, it is significant that a third of the software entrepreneurs reported having or seeking patents, and that they perceive patents to be important to persons or firms from whom they hope to obtain financing.

Survey background

More than 1,300 high technology entrepreneurs in the software, biotechnology, medical devices, and computer hardware fields filled out the Berkeley Patent Survey. All of these firms had been started no more than ten years before the survey was conducted. We drew our sample from a general population of software firms registered with Dun & Bradstreet (D&B) and from the VentureXpert (VX) database that has a rich data set on venture-backed startups. (Just over 500 of the survey respondents were D&B firms; just under 200 were VX firms.)

Eighty percent of the software respondents were either the CEOs or CTOs of their firms, and most had experience in previous startups. The average software firm had 58 employees, half of whom were engineers. Between 10 and 15 percent of the software startup respondents among the D&B respondents were venture-backed firms. Among the software respondents, only 2 percent had experienced an initial public offering (IPO), while 9 percent had been acquired by another firm.

Our interest in conducting this survey arose because high technology entrepreneurs have contributed significantly to economic growth in recent decades. They build firms that create new products, services, organizations, and opportunities for complementary economic activities. We were curious to know the extent to which high tech startups were utilizing the patent system, as well as to learn their reasons for choosing to avail themselves of the patent system — or not.

The basic economic principle underlying the patent system is that technology innovations are often expensive, time-consuming, and risky to develop, although once developed, these innovations are often cheap and easy to copy. In the absence of intellectual property rights (IPRs), innovative high tech firms may have insufficient incentives to invest in innovation insofar as they cannot recoup their research and development (R&D) expenses and justify further investments in innovation because of cheap copies that undermine the firms’ recoupment strategy.

Although this economic principle applies to all companies, early-stage technology firms might, we conjectured, be more sensitive to IPRs than more mature firms. The former often lack various kinds of complementary assets (such as well-defined marketing channels and access to cheap credit) that the latter are more likely to enjoy. We decided it would be worthwhile to test this conjecture empirically. With generous funding from the Ewing Marion Kauffman Foundation, we and two other colleagues designed and carried out the survey and analyzed the results.

Why startups decide to patent — or not to

The most important reasons for seeking patents, as reported by the software executives who responded to the Berkeley Patent Survey, were these:

1. to prevent competitors from copying the innovation (2.3 on a 4 point scale, where 2 was moderately important)
2. to enhance the firms’ reputation (2.2)
3. and to secure investment and improve the likelihood of an IPO (1.96 and 1.97 respectively)

The importance of patents to investors was also evident from survey data showing striking differences in the rate of patenting among the VX and the D&B software companies.

Three-quarters of the D&B firms had no patents and were not seeking them. Because the D&B firms are, we believe, typical of the population of software startup firms in the U.S., their responses may be representative of patenting rates among software startups generally. It is, in fact, possible that the overall percentage of software startup patenting is lower than this, insofar as patent holders may have been more likely than other software entrepreneurs to take time to fill out a Berkeley Patent Survey.

In striking contrast to the D&B respondents, over two-thirds of the VX software startup respondents in the sample, all venture-backed, had or were seeking patents. We cannot say why these VC-backed firms were more likely to seek patents than other firms. Perhaps VCs are urging the firms they fund to seek patents; or VCs may be choosing to fund the development of software technologies that VCs think are more amenable to patenting.

Interestingly, the rate of patenting did not vary by the age of the firm (that is, older firms did not patent at rates statistically significant from younger firms).

Why forgo patenting?

The survey asked two sets of questions about decisions to forego patenting: For the last innovation for which the firm chose not to seek a patent, what factors influenced this decision, and then what was the most important factor in the decision?

The costs of obtaining and of enforcing patents emerged as the first and second most frequent explanation. Twenty-eight percent of the software startups reported that the costs of obtaining patents had been the most important factor in this decision, and 12 percent said that the costs of enforcing patents was the most important factor. (They reported that average cost of getting a software patent was just under $30,000.)

Ease of inventing around the innovation and satisfaction with trade secrecy also influenced software startup decisions not to seek patents, although only rarely were these factors considered the most important.

Intriguingly, more than 40 percent of the software executive respondents cited the unpatentability of the invention as a factor in decisions to forego patenting, and almost a quarter of them rated this as the most important factor. Indeed, unpatentability ranked just behind costs of obtaining patents as the most frequently cited “most important factor” for not seeking patents.

It is difficult to know what to make of the unpatentability finding. One explanation might be that the software entrepreneur respondents believed that patent standards of novelty, non-obviousness, and the like are so rigorous that their innovation might not have satisfied patent requirements. Yet, because the patentability of software innovations has been contentious for decades, it may also be that a significant number of these entrepreneurs have philosophical or practical objections to patents in their field.

How important are patents to competitive advantage?

One of the most striking findings of our study is that software firms ranked patents dead last among seven strategies for attaining competitive advantage identified by the survey, as Figure 1 below shows. (The relative unimportance of patents for competitive advantage in the software field contrasts sharply with the perceived importance of patents in the biotech industry, where patents are ranked the most important means of attaining such advantage.)

Figure 1: Measures of Capturing “Competitive Advantage” from Inventions

As Figure 1 shows, software startups regard first-mover advantage as the single most important strategy for attaining competitive advantage. Next most important was complementary assets (e.g., providing services for licensed software or offering a proprietary complement to an open source program).

Interestingly, these two strategies for getting ahead in the market outstrip the IPRs about which we inquired for software firms. Among IPRs, though, copyrights and trademarks, closely followed by secrecy and difficulties of reverse engineering, outranked patents as means of attaining competitive advantage among software respondents by a statistically significant margin.

What incentive effects do patents have?

The Berkeley Patent survey asked startup executives to rate the incentive effects of patents on a scale, where 0 = no incentive, 1 = weak incentive, 2 = moderate incentive, and 3 = strong incentive, for engaging in four types of innovation: (1) inventing new products, processes, or services, (2) conducting initial R&D, (3) creating internal tools or processes, and (4) undertaking the risks and costs of commercializing the innovation.

We were surprised to discover that the software respondents reported that patents provide only weak incentives for engaging in core activities, such as invention of new products (.96) and commercialization (.93). By contrast, biotech and medical device firms reported just above 2 (moderate incentives) for these same questions.

Interestingly, the results did not change significantly even when focusing only on responses from software entrepreneurs whose firms hold at least one patent or application. Even patent-holding software entrepreneurs reported that patents provide just above a weak incentive for engaging in these innovation-related activities.

Resolving a paradox

If patents provide only weak incentives for investing in innovation among software startups, why are two-thirds of the VX firms and at least one-quarter of the D&B firms seeking patents?

The answer may lie in the perception among software entrepreneurs that patents may be important to potential funders, such as venture capitalists (VCs), angel investors, other firms, commercial banks, and friends and family. Sixty percent of software startups that had negotiated with VCs reported that that they perceived patents to be an important factor in VC decisions about whether to make the investments. Between 40 and 50 percent of the software respondents reported that patents were important to other types of investors, such as angels, investment banks, and other companies.

How well is the patent system working?

While most of the Berkeley Patent Survey questions focused on what firms had actually been doing vis-à-vis patents, we decided to ask a few questions to gauge the perception of high tech entrepreneurs about the patent system. We asked, for example, how well the entrepreneurs perceive the patent system to be working for them and for their industry. The scale for responses ranged from 0 = very poorly to 4 = very well, and 2 = neither poorly or well.

The software entrepreneurs’ for-my-industry rating was 1.6 and their for-my-firm rating was 1.7. Both results tend toward the poorly end of the scale (in contrast to the biotech and medical device firms that reported above 2 ratings on both questions).

It is interesting is that the VX firms were slightly less positive about the patent system than the D&B firms, although the difference was not statistically significant. We also tested to see if the responses were bipolar (that is, did some software firms rate the patent system very poorly and their ratings canceled out by some positive responses?), but discovered that the ratings fell into a normal distribution, suggesting that we had drawn a sample from a cross-section of the population.

Conclusion

Over the next several years, we expect to engage in further analysis of the results of the 2008 Berkeley Patent Survey and to report new findings about the roles that patents play in the software industry. The initial findings reported here and in the larger article suggest that software entrepreneurs do not find persuasive the canonical story that patents provide strong incentives to invest in technology innovation. These executives regard first-mover advantage and complementary assets as more important than IPRs in conferring competitive advantage upon their firms. Moreover, among IPRs, copyrights and trademarks are perceived to be more important than patents. Still, about one-third of our software entrepreneur respondents reported having or seeking patents, and their perception that their investors care about patents seems to be a key factor in decisions to obtain patents.

Related:

• How do we measure innovation?

References:

Stuart J.H. Graham, Robert P. Merges, Pam Samuelson, & Ted Sichelman, High Technology Entrepreneurs and the Patent System: Results of the 2008 Berkeley Patent Survey, Berkeley Technology Law Journal, 25:4, pp. 1255-1327 (2010), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1429049.

Google has scanned the texts of more than seven million books from major university research libraries for its Book Search initiative and processed the digitized copies to index their contents. Google allows users to download the entirety of these books if they are in the public domain (about 1 million of them are), but at this point makes available only “snippets” of relevant texts when the books are still in copyright unless the copyright owner has agreed to allow more to be displayed.

In the fall of 2005, the Authors Guild, which then had about 8000 members, and five publishers sued Google for copyright infringement. Google argued that its scanning, indexing, and snippet-providing was a fair and non-infringing use because it promoted wider public access to books and because Google would take out of the Book Search corpus any digitized books whose rights holders objected to their inclusion. Many copyright professionals expected the Authors Guild v. Google case to be the most important fair use case of the 21st century.

This column argues that the proposed settlement of this
lawsuit is a privately negotiated compulsory license primarily designed to
monetize millions of orphan works. It will benefit Google and certain authors and publishers, but it is questionable whether the authors of most books in the corpus (the “dead souls” to which the title refers) would agree that the settling authors and publishers will truly represent their interests when setting terms for access to the Book Search corpus.

Orphan Works

An estimated 70 per cent of the books in the Book Search
repository are in-copyright, but out of print. Most of them are, for all practical purposes, “orphan works,” that is, works for which it is virtually impossible to locate the appropriate rights holders to ask for permission to digitize them.

A broad consensus exists about the desirability of making
orphan works more widely available. Yet, without a safe harbor against possible infringement lawsuits, digitization projects pose significant copyright risks. Congress is considering legislation to lessen the risks of using orphan works, but it has yet to pass.

The proposed Book Search settlement agreement will solve the
orphan works problem for books—at least for Google. Under this agreement, which must be approved by a federal court judge to become final, Google would get, among other things, a license to display up to 20 per cent of the contents of in-copyright
out-of-print books, to run ads alongside these displays, and to sell access to the
full texts of these books to institutional subscribers and to individual
purchasers.

The Book Rights Registry

Approval of this settlement would establish a new collecting society, the Book Rights Registry (BRR), initially funded by Google with $34.5 million. The BRR will be responsible for allocating $45 million in settlement funds that Google is providing to compensate copyright owners for past uses of their books.

More important is Google’s commitment to pay the BRR 63 per
cent of the revenues it makes from Book Search that are subject to sharing
provisions. The revenue streams will come from ads appearing next to displays of in-copyright books in response to user queries and from individual purchases of and institutional subscriptions to some or all of the books in the corpus. Google and the BRR may also develop new business models over time that will be subject to similar sharing.

One of the main jobs of the BRR will be to distribute the settlement revenues. The money will go, less BRR’s costs, to authors and publishers who have registered their copyright claims with BRR. Although the settlement agreement extends only to books published prior to January 5, 2009, BRR is expected to attract authors and publishers of later-published books to participate in the revenue sharing arrangement that Google has negotiated with BRR.

Class Action Settlement

By now, readers may be a bit puzzled. How can Google be getting a license to make millions of in-copyright books available through Book Search just by
settling a lawsuit brought by a small fraction of authors and publishers?

U.S. law allows the filing of “class action” lawsuits whose
named plaintiffs claim they represent a class of persons who have suffered the
same kind of harm from the defendant’s wrongful conduct as long as there are
common issues of fact and law that make it desirable to adjudicate the claims
in one lawsuit instead of many.   

The Authors Guild and three of its members sued Google, claiming to represent a class of similarly situated authors whose books Google was scanning and whose copyrights Google was violating. By bringing a class action, the Authors Guild put considerable financial pressure on Google because the winner of a class action lawsuit is entitled to compensation that equals all of the monies owed to the class, which may be exponentially higher than awards to individual
plaintiffs.

In the absence of the proposed settlement, Google would almost certainly have vigorously fought against certification of the class in the Authors Guild case. After all, the guild has only a few thousand members and most of them do not write the kinds of scholarly works that are typically found in major university research libraries. Many scholars would want their books to be scanned by the Book Search project so they would be more accessible to potential readers.

The publisher lawsuit did not start out as a class action, perhaps in part because McGraw-Hill, et al., recognized how difficult it would be for them to prove they adequately represented a class of all book publishers whose books Google had scanned.

However, the settlement agreement that Google has negotiated
with the Authors Guild and the Association of American Publishers would, if
approved, be settled as a class action on behalf of all book authors and publishers, with the Guild and AAP claiming to represent their entire respective classes. By acceding to the certification of these classes through
this settlement, Google will get a license from all authors and publishers of books
covered by the agreement (which is to say nearly every in-copyright book ever
published in the U.S.) so that it can commercialize them though Book Search.

Google’s New Monopoly

The proposed settlement agreement would give Google a
monopoly on the largest digital library of books in the world. It and BRR, which will also be a monopoly, will have considerable freedom to set prices and terms and conditions for Book Search’s commercial services. BRR is unlikely to complain that the price is too high, the digital rights management technology is too restrictive, or the terms are too onerous.

Google will also be the only service lawfully able to sell orphan books and monetize them through subscriptions. BRR will get 63 per cent of these revenues which it will pay out to authors and publishers registered with it, even as to books in which they hold no rights. (Some unclaimed orphan book funds may go to charities that promote literacy.) No author whose books are in the corpus can get paid by the BRR unless he/she has registered with it.

Virtually the only way that Amazon.com, Microsoft, Yahoo!, or the Open Content Alliance could get a comparably broad license as the settlement would give Google would be by starting its own project to scan books. The scanner might then be sued for copyright infringement, as Google was. It would be very costly and very risky to litigate a fair use claim to final judgment given how high copyright damages can be (up to $150,000 per infringed work). Chances are also slim that the plaintiffs in such a lawsuit would be willing or able to settle on equivalent or even similar terms.

Dead Souls

The Book Search settlement brings to mind Nikolai Gogol’s story, Dead Souls. Chichikov, its main character, travels around the Russian countryside to buy “dead souls” so that he can become a wealthy and influential man. In
the early 19th century, you see, Russian landowners had to pay annual
taxes on the number of serfs (counted as “souls”) they owned as of the last census.

Chichikov offered to buy “dead souls” (i.e., serfs who had died since the last census) from the landowners. His plan was to acquire enough of these souls so that he could take out a large loan secured by his portfolio, and thereby to become a
wealthy man.

In Gogol’s story, Chichikov’s scheme falls apart. Rumors fly that the souls he owns are all dead and he flees the town in disgrace.

However, Google’s “dead souls” scheme may pay off handsomely, as the settlement would, in effect, give Google the exclusive right to
commercially exploit millions of orphan books.

Representativeness?

As galling as it is to realize that the BRR and its registered authors and publishers will derive income from millions of books they didn’t write or publish, it is even more galling that copyright maximalists will almost certain dominate the BRR governing board.

(The Authors Guild president, for example, recently complained about the “read aloud” feature of Kindle, denoting it a “swindle,” and a copyright infringement. The AAP is supporting legislation to forbid the National Institutes of Health from promoting “open access” policies for articles written under NIH grants. And of course, the Authors Guild and AAP characterized Google as a thief for scanning books from research libraries.)

If asked, the authors of orphan books in major research libraries might well prefer for their books to be available under Creative Commons licenses or put in the public domain so that fellow researchers could have greater access to them. The BRR will have an institutional bias against encouraging this or considering what terms of access most authors of books in the corpus would want.

In reviewing the settlement, the judge who is supposed to consider whether the settlement is “fair” to the classes on whose behalf the lawsuits were brought. He may assume the settlement is fair because money will flow to authors and publishers. But importantly absent from the courtroom will be the orphan book authors who might have qualms about the Authors Guild and AAP as their representatives.

Conclusion

In the short run, the Google Book Search settlement will unquestionably bring about greater access to books collected by major research libraries over the years. But it is very worrisome that this agreement, which was negotiated in secret by Google and a few lawyers working for the Authors Guild and AAP (who will, by the way, get up to $45.5 million in fees for their work on the settlement—more than all of the authors combined!), will create two complementary monopolies with exclusive rights over a research corpus of this magnitude. Monopolies are prone to engage in many abuses.

The Book Search agreement is not really a settlement of a dispute over whether scanning books to index them is fair use. It is a major restructuring of the book industry’s future without meaningful government oversight. The market for digitized orphan books could be competitive, but will not be if this settlement is approved as is.