3

Digital Products and Federal Policy for the Innovation Economy

Digital output from federal laboratories includes data, metadata, images, software, code, tools, databases, algorithms, and statistical models. These digital products can be copied at little or no cost and used by many without limit or additional cost. Given that digital products are produced by federal labs, are paid for with taxpayer dollars, and can generate large positive externalities, policies that foster their widespread use and rapid dissemination yield a large societal benefit. Society can often benefit most when technology is in the public domain freely, which in theory makes the assets freely available to all firms. However, small and minority-owned firms can be systemically excluded from accessing or exploiting what is in the public domain because of a lack of resources or the structure of the specific markets in which they operate. Moreover, when substantial investment capital and expenditures are necessary to bring a digital invention to market and otherwise promote its use by the public, and that investment would not be recoverable in a competitive market, some form of exclusivity may be required to create incentives for the investment.

DIGITAL PRODUCTS AND INNOVATION

Innovation has enabled capabilities in data collection, computation, simulation, and analysis unheard of even a decade ago. Numerous fields, from science to cybersecurity and city management to transportation, benefit from and are continuously challenged by rapidly evolving digital innovation.

Digital products power nearly every facet of modern life. Increased computational power, artificial intelligence (AI), and machine learning (ML) are driving pathbreaking progress in science and medicine. AI and ML, for example, are helping to make society “smarter,” whether by incorporating crash avoidance technology into automobiles, building more efficient energy management systems for homes and utilities, or using facial recognition technologies to combat terrorism. Further, open-source software (OSS) and open data have become

integral components of the modern technological infrastructure. Use of OSS is widespread among industry, academia, and the federal government, enabling a broad community of developers to collaborate on new, innovative technologies.

Looking beyond the information and communications technology (ICT) sector, digital technology also features prominently in the immense health care and biomedical research enterprise. Today, digital products anchor efforts to deploy health care, maintain medical records, conduct biomedical research, and discover new drugs and therapies. The Human Genome Project (HGP), for example, a massive international undertaking to sequence the entire genome of the human species, released all of the data it generated into the public domain, creating new industries virtually overnight. Bioinformatics, genetic diagnostics, pharmacogenomics, and precision medicine all have their roots in data produced and released by this and subsequent genomic data projects. As noted by the National Science Foundation (NSF), industries that produce products and services for the digital economy have some of the highest innovation rates in the country, ranging from software publishing (61 percent), to computer and electronic products manufacturing (53 percent), to medical equipment and supplies (44 percent), to scientific research and development (R&D) services (43 percent).¹

INTELLECTUAL PROPERTY POLICY AND INNOVATION IN THE FEDERAL LABORATORIES

When examining the laws and policies that specifically govern the development of digital products from the federal laboratories, it is useful first to take a step back and consider the overall framework of innovations arising from the federal lab system. The interaction of two foundational premises—the role of government in creating public resources and the role of intellectual property (IP) law in incentivizing innovation—anchors the function of the labs in the national economy and is essential to understanding the basis for IP regulation within the labs.

As discussed in Chapter 2, federal labs were established to conduct research in key sectors (for example, the agricultural experiment stations for agriculture), to assist in developing industry-wide standards, or to develop defense-related technologies (such as radar and the atomic bomb). These are classic public resources, designed to benefit the nation as a whole and funded by taxpayers. After World War II, the federal lab system expanded to encompass further civilian applications, such as nuclear power, space exploration, and medical research. In each of these cases, the role of the labs was to create some welfare-enhancing technology or to generate new knowledge, which would be deployed for the public benefit. As with other types of infrastructural assets, these public resources generated significant spillover effects in terms of commercial applications not pursued directly by government agencies.

___________________

¹ Percentages of U.S. companies in a particular industry reporting “product or process innovation” in the period 2014–2016 (NSB and NSF, 2020).

Yet by the 1970s, U.S. policy makers recognized that potentially useful technologies developed within government labs were not finding their way into commercial products and that the United States lagged behind its allies in transferring federally created technology to the private sector. For example, Japan’s Ministry of International Trade and Industry, which formulates Japanese industrial policy, directed significant resources to the successful development of such market sectors as consumer electronics and automobiles using technology initially developed in national labs. Thus the U.S. Congress recognized the need for a way to increase the commercialization of technologies developed by the federal government and began developing a series of policies for this purpose, including policies relating to IP for products developed with federal funds. Studies of the impact of the Bayh-Dole Act, which transferred ownership of inventions and IP arising from publicly funded research from federal agencies to universities showed that university ownership of IP rights did not have a detrimental impact on commercialization (NRC, 2011). See Box 3-1 for more information on the Bayh-Dole Act.

Federal IP policy also plays a role in incentivizing innovation in the private sector. Article I, Section 8, Clause 8 of the U.S. Constitution states that Congress is authorized to enact patent and copyright laws “to promote the progress of science and useful arts”—an unbridled instrumentalist justification for IP. It authorizes Congress to grant limited periods of exclusivity to “authors and inventors” to incentivize “progress”—the creation of new works. In other words,

exclusivity may be granted to the extent that it promotes innovation. Without exclusive rights, incentives for firms to invest in innovation tend to be considerably reduced because of the possibility of free rider effects (Heller and Eisenberg, 1998; Buchanan and Yoon, 2000). Conversely, if granting exclusivity would not promote innovation, it should not be granted. This is the case because works the public can freely utilize and adapt also lead to innovation, as demonstrated by the significant advances achieved through OSS and works dedicated largely to the public domain, such as the Internet and the Global Positioning System (GPS) signal.

At the same time, the government may need to assert some control over the use of digital products generated by federal labs. For example, free distribution of digital technologies might encourage and enhance the ability of non-U.S. entities to invest in and develop those same technologies, in competition with U.S. firms. Free distribution might also hinder the government from exerting downstream controls over innovations produced from the digital technologies, including providing transparency of product or pricing information to the public or restricting anticompetitive behaviors (Okediji, 2016). Finally, there is evidence that such factors as gender, race, and appearance affect the ability of a scientist to commercialize new inventions (Dolmans et al., 2016; Shane et al., 2015). And there is evidence that social dimensions, including differences in resources, economic disparities, and gender or racial discrimination, affect an individual’s ability to take advantage of resources, such as digital technologies, that are placed in the public domain (Chander and Sunder, 2004; Veletsianos and Kimmons, 2012; Willinsky, 2006).

In short, acting in the public interest will require balancing a variety of factors to determine when government stewardship is best accomplished by allowing exclusive use by a firm, by the adoption of open-access licensing, or by dedication to the public domain (with or without a license) to advance scientific progress and innovation.

Bringing Federal Technologies into the Marketplace

The private sector plays an essential role in bringing federal technologies into the marketplace. Federal laboratories are neither well suited nor chartered to commercialize their work products directly, being driven by the missions of the agencies with which they are associated rather than market forces. Therefore, technology transfer from federal labs to the commercial marketplace is possible only through partnerships with private actors and measures to meet the need for economic incentives that make such partnerships feasible.

Clearly, some products of federal labs, such as U.S. Geological Survey (USGS) data, the GPS signal, and the human genome sequence, can be used by the private sector with minimal adaptation. Commercial firms, each of which has access to the same federally provisioned no-cost inputs, can then compete on the basis of their proprietary improvements on and applications using those inputs (e.g., navigation software, weather prediction systems, diagnostic kits, and

genetically tailored drugs). However, not all products of federal labs can be translated directly, and without significant additional investment, into commercial products; some may require augmentation, adaptation, or modification to become commercially viable. Successfully commercialized software also can require significant investment in user support, upgrades, and maintenance.

Intellectual Property Rights Considerations

Government decisions related to IP rights (IPRs) in federal technology need to take into account the likelihood that the technology can be commercialized absent exclusivity, or costly adaptations may be required that would justify granting an IPR. For example, if the total expected market for the adapted product (i.e., when offered by a single source without competition) would yield net profit in excess of the cost of making the necessary adaptations, the grant of exclusivity may be justified for that product. This is the case, however, only if commercialization of the federal product would not otherwise occur in a market under normal competitive conditions.² Otherwise, granting that private actor the exclusive right to exploit the federal product would operate simply as a wealth transfer to the private actor facilitated by the federal government, with no offsetting welfare benefit. Such wealth transfers, absent other welfare gains, are to be avoided by government agencies. Admittedly, moreover, it is difficult to predict in any given instance how much it will cost to adapt a particular federal technology for commercial use, whether that product would be commercialized without exclusivity, whether it will be successful in the marketplace, what costs will be involved in maintaining and supporting it, and whether patent or copyright rights for that product would eventually be recognized by a court.

The National Institutes of Health (NIH) has developed best practices for its laboratories with regard to the patenting and licensing of genomic inventions. These practices include avoiding exclusivity for “broad enabling technologies” that would be most beneficial if “widely available and accessible to the scientific community” (NIH, 2005). By the same token, if a federal technology would be broadly beneficial to society, granting exclusive rights to a single firm should be avoided unless it is necessary to ensure that the technology is commercialized.

Another relevant factor to consider is the ultimate cost that will be borne by members of the public to access and use the federal product. According to the so-called “double subsidy” argument, U.S. taxpayers should have free access to work funded by tax dollars; allowing a private actor to privatize a public good results in a double charge to the taxpayers. But in contrast, some critics have observed that releasing federal products to the public gives access not only to the U.S. taxpayers who funded that work but also to foreign users, which raises two separate issues. First, these foreign users did not contribute to the creation of U.S.

___________________

² As discussed in Chapter 6, the Department of Energy, the National Institutes of Health, and the National Aeronautics and Space Administration have introduced programs to improve technology transfer of inventions created at the federal labs.

taxpayer investments. Second, giving foreign firms access to U.S.-created digital products enables them to compete with American businesses, an outcome that appears to be at odds with the goals of the Stevenson-Wydler Act.

Even so, as discussed further below, federal policy favors disseminating the results of scientific endeavors as widely as possible. Accordingly, scientists in most federal labs are encouraged to publish their research in scientific journals and other publications, which many do. The publications of scientists employed by government-owned, government-operated (GOGO) federal labs are a significant form of knowledge transfer and are generally not subject to copyright restrictions. However, the underlying invention, if any, described in the published article may be the subject of a patent or copyright. Chapters 4 and 5 describe the IP issues related to federal labs in more detail.

Table 3-1 summarizes the analysis outlined in this section at a high level using paradigmatic examples of each technology type.³ When an unclassified federal technology is readily adaptable to commercial use without significant expenditure, as in the case of genomic data, that federal technology should achieve the broadest dissemination and yield the greatest welfare gains when it is treated as a public resource accessible to and usable by all. Likewise, when a federal technology is adaptable to commercial use with only modest expenditures, as in the case of a software program that was designed for use in a government setting but can be adapted to a commercial setting with modest programming modifications, treating the technology as a public resource (e.g., through OSS release or dedication to the public domain) is likely to lead to its broadest dissemination and use. In contrast, a federal technology, such as certain types of health care software that require approval from the Food and Drug Administration

TABLE 3-1 Nature, Recommended Treatment, and Examples of Federal Laboratory Technologies

___________________

³ A more detailed set of considerations used by federal labs in deciding whether to release a particular software product on an OSS basis is described in Chapter 6.

that would require substantial expenditure before commercialization may require some degree of exclusivity if it is to be commercialized successfully.

In general, the application of IP law to the federal labs has, over the years, been inconsistent and lacking in a single set of guiding principles. This situation likely reflects a combination of historical accident, uncoordinated decision making across a range of agencies and legal domains, and the accumulation of special-purpose exceptions (some directed by Congress) that have persisted over the years. As described in Chapters 4 and 5, the result is that the landscape of IP for federal labs today is difficult to catalog and address through any single set of policy prescriptions.

THE ECONOMICS OF DIGITAL PRODUCTS

The framework described above is generally agnostic to the underlying technology in question and relies instead on the technology’s “usability” factor. It is important to note, however, that digital products differ in important ways from their tangible counterparts, and it is worth examining the economic theory behind them to see whether the above framework still applies. Digital products, unlike physical products, are distinguished by their “nonrival” nature, which means they have nearly zero costs of reproduction, and multiple users can share them at no additional cost. As with all products that fall within the IP domain, digital products typically have substantial fixed or “first-copy” costs. Once they have been created, however, the cost of making them available to another user (the “marginal cost”) is often negligible.

The near-zero reproduction cost or nonrival nature of digital products means that maximizing their availability can yield large economic and societal benefits. When it costs virtually nothing to serve another consumer, charging for a product can result in inefficiency. Of course, it is possible to generate revenue by selling (or otherwise monetizing, such as via advertising) digital products, as commercial firms do.

Unlike firms, which generate revenue from digital products in order to survive, federal laboratories do not need to generate royalties and licensing revenue to fund their operations.⁴ As discussed above, most commercially relevant digital products generated by federal labs are by-products of the labs’ mission-driven research activity. Moreover, these digital products are financed by taxpayers, which constitutes an additional rationale for making them openly and freely available. Indeed, many digital products developed by the federal labs are freely distributed via repositories of open data, open code, and open publications (see Chapter 7 for more information). At the same time, as observed earlier, evidence suggests that some users are not in a position (because of their limited capabilities or the competitive structure of the specific industry) to easily

___________________

⁴ As discussed Chapter 6, a small share of federal labs fund their technology transfer offices out of licensing and royalty revenue.

determine how to access and use what is in the public domain (Chander and Sunder, 2004; Veletsianos and Kimmons, 2012; Willinsky, 2006).

Although the nonrivalrous nature of government digital products suggests that charging for them can create inefficiency, there can be situations in which alternatives to open availability can better advance development. For example, it is possible that some digital products created in federal labs are not directly usable off the shelf and require such substantial additional commercial investment that only a sole licensee would garner commercial returns adequate to cover the cost of that investment.

It also is possible that market actors will not partner with the federal labs without securing exclusive rights to the products created through the partnership. If multiple competitors have access to the same underlying federal technology and the ultimate product is readily imitable, none of them may be able to capture sufficient revenue when the product is sold to justify the initial expenditure required to develop and commercialize the technology. As a result, no private actor may have a sufficient economic incentive to commercialize the federal product, and it will remain uncommercialized, yielding no social benefits whatsoever. Researchers have shown, at least in the case of university technology transfer, that exclusive licenses may be required to induce firms to invest in the development of embryonic inventions (Buchanan and Yoon, 2000; Colyvas et al., 2002; Heller and Eisenberg, 1998; Mowery et al., 2001; Thursby and Thursby, 2007). A variety of examples illustrating this effect have been reported to the Government Accountability Office and repeated at various Congressional hearings, although the extent of these circumstances is unknown. These examples included “a computer program that would assist dermatologists in prescribing medications and other treatments for medical problems, such as acne” (GAO, 1991, p. 4); software that allows researchers to capture vehicle performance data during an emissions and fuel economy test (U.S. House, 1992, p. 33); and software “to maximize cotton yields in the southern United States by assisting farmers in deciding, for example, when to irrigate, fertilize, and defoliate their cotton crops” (GAO, 1990, p. 32). The committee received evidence of the need for IP protection for certain products, such as software for instrumentation, to aid in commercialization.⁵

Finally, while federal labs may develop software that offers the potential for commercial application, that potential may not be realized without significant codevelopment effort on the part of both lab and industry personnel, necessitating investment on the part of industry partners. In such a circumstance, it is no longer the government that bears the full cost of technology development. One example of this situation presented to the committee was a cooperative research agreement between Los Alamos National Laboratory (LANL) and the consumer goods company Procter & Gamble (P&G).⁶ In that case, P&G asked for help to improve product reliability, and LANL responded by developing a new (and unproven)

___________________

⁵ Presentation to the committee by Mark Rohrbaugh, National Institutes of Health, December 5, 2019.

⁶ Presentation to committee by Art Koehler, Procter & Gamble, December 6, 2019.

statistical model that could be used to simulate changes to production lines. To test this model, P&G had to collect and feed to the model real-world data, and once proven, the model had to be translated into a user-friendly “toolbox” that could be used across the company. The use of collaborative efforts with industry, discussed in Chapter 6, suggests that the need for such codevelopment is common. The general principle guiding the conferring of control rights from the government initially is whether, absent such control rights, a private actor would have sufficient incentive to build upon and develop the data or software in question for societal benefit.

CONCLUSION

In conclusion, there can be important exceptions to the argument that digital products produced in federal laboratories should be made available to private firms with no exclusivity restrictions and no charge. The criteria for making those exceptions are discussed further in Chapters 4 and 5.

FINDINGS AND RECOMMENDATIONS

Finding 3-1: Making government data freely and openly available maximizes the use, reuse, and therefore the value of these data for commercial and noncommercial entities.

Finding 3-2: Federally produced digital products often yield large societal benefits when widely distributed, although federal laboratories may need to restrict access to those products when significant and costly follow-on development by firms is needed to commercialize them.

Finding 3-3: While placing digital products in the public domain may reduce obstacles to their use, reliance on the public domain alone will not enable the participation of small firms, minority-owned firms, woman-owned firms, and members of society that lack the market networks, resources, and tools to discover and exploit what is available in the public domain.

Recommendation 3-1: Federal laboratory directors should ensure that data and associated metadata produced by their labs are freely and openly available for use by individuals, researchers, and firms to the fullest extent possible under existing statutes and policies.

Recommendation 3-2: When additional substantial investment by the private sector is necessary to commercialize a digital innovation and the ultimate product is readily imitable, a federal laboratory director should allow exclusive access to that innovation to incentivize such investment.

Recommendation 3-3: Federal laboratory directors should consider ways to identify and support small, minority-owned, and woman-owned firms and members of society that may not be in a position to identify freely available digital goods, may not know how to interface with the federal labs on such matters, and/or may not understand how the use of these digital products may be relevant to their interests. Federal lab directors should proactively address such disparities in access.