Open Source Biotechnology Project

Copyright (c) 2003 Janet Hope, Research School of Social Sciences, Australian National University, Canberra, ACT 0200. Verbatim copying and distribution of this site is permitted in any medium, provided this notice is preserved.

Open source biotechnology?

Ultimately, the view I am taking for the purposes of this project is that the business case for open source software is not in question - it has already proved itself. The crucial question is whether the differences between software development and biotechnology R&D doom the open source biotechnology approach to failure. The following sections offer some preliminary thoughts on this question.

A plausible scenario?
Lessons from open hardware

What is open hardware?

A sceptical viewpoint

Other issues
Comments on open source and drug development
Conclusion
Post Script: A speculative future for open source biotechnology

A plausible scenario?

Imagine a small biotechnology company that generates revenue by exploiting an in-house proprietary technology platform centred around a patented technique widely used in both animal and plant research. Some of the company's income comes from licensing fees derived from granting an exclusive licence to another company to produce a range of laboratory kits incorporating the technique; the licensee chooses to produce kits only for animal research. The bulk of the company's revenue comes not from licence fees, however, but from performing contract research and data analysis services, mainly for the meat industry, and by selling software for analysing data generated by means of the technique.

As well as direct revenue generating activities, this imaginary company engages in research and development activities. The company devotes a substantial proportion of its resources each year to research, which is of economic importance to the company in two ways: first, it maintains and improves the technology platform on which the company's revenue-generating activities are based, and second, it attracts customers for services and software by enhancing a reputation for high-quality work and expertise in the field. Because our imaginary company is small, its stand-alone research capacity is limited. In response, the company makes a point of pooling research resources in corporate partnerships with other research organisations. The results of these collaborations are then privately owned.

Is this company's patented technique a good candidate for open source licensing? It could be, for the following reasons. First, the company obtains only a fraction of its total revenue from licensing fees; it already operates largely in the secondary markets of research and software sales. If the number of users of the patented technique were to increase, even if those users did not pay licence fees, the market for the company's research and software offerings would expand. Second, according to our imagined scenario, the company devotes significant resources to R&D and would therefore probably benefit from sharing those costs with a group of external user-developers. Any improvements in the use value of the patented technique would directly benefit the company because it doesn't just sell the technology, it is also a major user in its contract research and data analysis activities. Building a user community would also help build the company's valuable reputation for high-quality research. Third, suppose that the patented technique is no longer at the very cutting edge of research tools in its area. This would mean that the competitive edge in research services that the company gains by exclusive possession of the technique is starting to trail off, and user loyalty and overall satisfaction are becoming more important.

Finally, there is a whole set of potential users - and therefore potential customers for closed source technologies built on the platform of the patented technique - who are being prevented by the company's existing licensing arrangements from gaining access to modified versions of the technology that would be useful in their research (i.e., plant researchers). This is a lost opportunity that opening up the exclusive licence could help recover.

This hypothetical example demonstrates that there may well be biotechnology companies for whom a proprietary technology currently represents "the crown jewels" that could benefit economically from converting to an open source approach to intellectual property management. It also provides a concrete setting for a discussion (in the next section) of the ways in which the open source/closed source trade-off might differ between the software and biotechnology contexts.

Lessons from open hardware

Benkler (Benkler. 2002. Coase's Penguin, or, Linux and The Nature of the Firm. 112 Yale L.J. (Winter 2002-03). Available at http://www.law.nyu.edu/benklery/.) has suggested that the success of commons-based peer production depends on relativly low physical capital costs and information transaction costs. One obvious difference between computer software and biotechnology research tools is that computer software is digital information, while many biotechnology tools are physical objects. The open hardware movement is of interest because, like the concept of open source biotechnology, it extends open source principles into a context where physical capital costs are non-negligible.

What is open hardware?

The open hardware movement has its roots in the radical technology movement of the 1960s, inspired by Ivan Illich's "Tools for a Convivial Society", and the subsequent development of bazaar-style chip manufacture. It is currently enjoying a renaissance for a number of reasons, including the success of the free software movement and the advent of the Internet.

"Open source hardware" refers to computer hardware for which all the design information is made available to the general public. A "free" or "libre" hardware design is a design which can be freely copied, distributed, modified and manufactured; the fact are design is free does not imply that it cannot be sold or that any hardware implementation of the design will be free of cost. Open source hardware may or may not be based on a free hardware design.

The term "open hardware" has a more precise meaning than either of these terms. For hardware to be open hardware:

1. Its design must be publicly accessible in a form that enables implementation and full understanding.

2. The software tools used to create the design should be free so that others can develop and improve the design.

3. The software interface to the hardware must be publicly accessible and free to use.

A sceptical viewpoint

Open hardware, like open source, has its share of ideologues and sceptics; here we are interested in the sceptics' view. Sceptics of open hardware have made a number of points that could also potentially be applied to open source biotechnology. These are listed here in order to highlight some of the differences between software and biotechnology research tools with respect to the feasibility of open source development that may be associated with the non-digital nature of many biotechnology research tools.

The first point is that no generally accepted open hardware licences yet exist. As noted in an earlier section, copyleft licences (and other forms of open source software licence) rely on copyright: the copyright owner uses his or her exclusive rights to guarantee certain freedoms for users of the software program covered by the licence. However, as with biotechnology research tools, computer hardware is mostly protected by patent. In contrast to copyright protection, which is quick, cheap and simple, obtaining a patent is a costly, time-consuming process; moreover, maintaining a patent requires the payment of substantial renewal fees. (There may be other relevant differences between patent protection and copyright protection besides cost -- if so, they will be explored as part of this project.) The costs of obtaining patent protection may make patent owners less willing than copyright owners to give up the income stream associated with standard proprietary licensing, and/or it may discourage otherwise willing contributors to an open source biotechnology project who are not in a position to obtain patent protection for their contributions. The question of how to translate open source licences into the context of patent protection (and other types of protection that may apply to biotechnology research tools) is an important one for open hardware, and also for open source biotechnology. Licensing experiments in the open hardware context continue: for example, the Indian Simputer is subject to a lawyer-drafted GPL-like licence.

The second point made by open hardware sceptics that may also apply to open source biotechnology is that hardware is not as modular and compartmentalised as software. Benkler has emphasised modularity as an important feature of successful open source projects, in particular in connection with contributors' motivation. However, open hardware sceptics have raised a different point, which is that unless the technology itself is highly modular and compartmentalised, small changes to one part are likely to interact in unforeseen ways with the rest. Whether this is in fact a difference between software and hardware is not clear. However, the general point may be particularly relevant to biotechnology research tools, as many are living organisms or components of living organisms: unpredictable, delayed side-effects in response to apparently small changes are established characteristics of living organisms as a class of complex systems.

The third point is that capital costs associated with hardware manufacture are higher than for software, so that human creativity costs are a smaller proportion of the total costs in the hardware context. (As we saw earlier, Benkler suggests that the advantages of the peer production mode relative to other modes relate to information about and allocation of human creativity, and become salient when human creativity is a salient component of production.) Capital costs for hardware manufacture are higher than for software manufacture in relation to both development (for example, tools for developing, testing and debugging software are much cheaper and more easily made accessible -- e.g. by Internet and open source software licensing, as many such tools are themselves software -- than tools for developing hardware) and production (for example, silicon for making chips costs money). Open hardware sceptics have suggested that there are therefore minimal start-up costs for software programmers but not for hardware developers, and further, that resulting reliance on institutional funding for hardware manufacture makes the process more vulnerable to conservative institutional attitudes and employment-related legal constraints. (As noted in an earlier section, the latter problems may be partially solved by top-down influence on institutional thinking by, for example, funding agencies such as the NIH, and by direct participation in open source biotechnology projects by institutions or companies, as distinct from individual employees.)

As we have seen in earlier sections, these issues also arise in relation to many biotechnology research tools (although the ability of living organisms to self-replicate means that for some, production costs at least may be negligible). It is important to realise that there are likely to be exceptions to every generalisation about biotechnology research tools, as the term is being used here in its broadest possible sense; part of the aim of this study is to identify particular research tools, or classes of research tools, that may be more amenable to open source licensing than others. As noted earlier, discussions of open source software often invoke the stereotype of the hobbyist hacker alone at a home computer needing nothing but an Internet connection and packet of Doritos, whereas in fact a lot of open source software development and production is carried out by professional programmers employed in various public and private sector institutions and may involve non-negligible costs: in other words, the differences between costs of software development and biotechnology research and development may often be exaggerated. In addition, although biotech research probably will remain more expensive on average than software development, some biotechnology research and development projects may be less capital intensive than others. The bioinformatics revolution in biology is already making it possible for some researchers to work from secondary data instead of doing everything in the lab. Even for wet lab inventory, some costs are going down rapidly; for example, is now possible to get the parts for a DNA sequence synthesiser that would have cost US $100,000 several years ago for about US $10,000. The point is that the question whether differences in capital costs are likely to determine the success or failure of open source biotechnology is a matter for empirical study and will depend on the circumstances of each case -- there can be no foregone conclusion.

A final point arising from skeptical writings on open hardware relates to motivation -- not to contribute to a project, but to explore the concept itself and find ways to overcome obstacles. Richard Stallman has asserted that freedom to copy hardware is not as important as freedom to copy software because copying software is easy to do, whereas copying hardware is difficult. Whether or not this argument has any merit in its original context, the factual premise -- that copying is difficult -- is not true of many biotechnology research tools, especially those that are able to self-replicate. In fact, this ability raises a different kind of potential obstacle to open source development in the biotechnology context: because modifications to living organisms can spread without human intervention, the origin of modifications to a living research tool such as a cell line are very difficult to trace, and it is also very difficult to define such modifications in the definition section of a license (the NIH uniform biological material transfer agreement has provisions that attempt to deal with this problem). This difficulty, of course, applies generally to licensing of biotechnology research tools, not just to open source licensing, and in fact may create more of a problem in the drafting of standard proprietary licences (this is a matter for further consideration). Indeed, critics of genetic engineering might see this problem of definitions as a symptom of a larger problem that is pervasive in biotechnology research and development, that of philosophical reductionism. [Not clear how to integrate writings on this topic -- leave for a later revision.]

A similar point to Stallman's, made in response to Dan Burk's paper on open source genomics, is that in the biotechnology community the history of commercialisation may have generated a greater commitment to the standard proprietary business and licensing model than in the software community, resulting in more cultural resistance to the concept of open source. This appears to be a fairly drastic generalisation; presumably it matters which sector of the "community" one is thinking of. Some parts of the software development community are notoriously committed to the standard proprietary model and resistant to the concept of open source, but this has not prevented open source from gaining a foothold elsewhere. Some players in the pharmaceutical and agricultural biotechnology industries have a long tradition of commitment to aggressively closed intellectual property management practices, but others have an equally long tradition of commitment to open scientific communication and the public interest, while still others are simply trying to make a living and may well be willing to consider even a radical departure from the standard business model if an alternative model promises lower costs or increased profits. It may be that the most sympathetic audience for ideas about open source biotechnology will be researchers and institutions in developing countries and institutions that have their interests at heart. Interestingly, the renaissance of the open hardware movement has been partly attributed to the efforts of developing countries to develop "convivial tools" as an alternative technology strategy in response to global expansion of intellectual property protection; one of the most successful open hardware projects to date is the Simputer. ("The Simputer is not a Personal Computer in the conventional `PC' sense. The `Win-tel' architecture of the de facto standard PC is quite unsuitable for deployment in the low-cost mass-market of any developing country... [the Simputer] is targeted as a shared computing device for a local community of users. A local community such as the village panchayat or the village school, or a kiosk, should be able to give this device out to individuals for a specific period of time and then pass it on to others in the community. Designed with this application in mind, the Simputer is heavily based on personalization through the use of smart cards. Other related designs are the MIT Pengachu (designed for minimal power consumption, perhaps even wind-up)." - http://opencollector.org/Whyfree/freedesign.html) [See also Bangalore Declaration on Information Technology for Developing Countries.]

The next section of the website discusses other issues and potential obstacles that have been raised in connection with the concept of open source biotechnology that do not have any strong parallel with respect to open hardware.

Other issues

A number of other differences between software and biotechnology research tools may affect the likely outcome of the trade-off between open source and closed source business strategies in the biotechnology context compared with the software context.

First, the costs of effective communication and coordination among contributors to a distributed development effort may be much higher in the biotechnology context, due to the less codified nature of much of the relevant information (for example experimental protocols that may be difficult to reproduce reliably outside the laboratory where they originated) and the need to exchange biological materials such as vectors, cell lines, knockout mice etc. Recall that for Mandeville, codification represents formalised learning -- that is, learning arranged and organised into a pattern -- and the costs of exchanging highly codified information, of which software code is an excellent example, are lower than for uncodified information. (With regard to biological materials, although the information contained in such materials is "embodied in a tangible object", the complexity of the object and its overall information content -- e.g. a living organism -- means that this type of information belongs at the "uncodified" end of Mandeville's information spectrum.) This is true not just for the giver of information but also for the receiver. While the Internet can certainly lower costs of some types of communication that would be necessary among contributors to an open source biotechnology project, it is unlikely that such a project would be able to dispense entirely with face-to-face meetings (e.g. to transfer technical skills) and physical networks (i.e. postal and courier services).

There are two possible replies to this point. First, some proponents of the application of open source principles to biology have predicted that new technologies such as rapid DNA synthesis machines may eliminate the need for physical networks in some cases: they suggest that instead of genetic samples being sent in the mail, it might become possible to send sequence information via telephone lines in electronic form to be received by DNA synthesis machines that would replicate the physical sample. The obvious concern with this kind of solution to the costs of exchanging biological information is that it may exacerbate distortions associated with the already highly reductionist character of biotechnology research. Second, there already exist distributed development projects in the biotechnology field that, one way and another, have succeeded in overcoming the difficulties of establishing communication and structuring and effective community. The existence of such initiatives indicates that the real question for the purposes of this project is not so much whether biotechnology research and development can be successfully conducted in a distributed fashion -- after all, this is the traditional academic mode of production -- but whether it can sustain itself without enormous injections of external funding (bearing in mind, of course, that there is no reason to impose more stringent self-sufficiency requirements on open source biotechnology than on "closed" biotechnology, which does rely to some degree on public funding [see pie chart p7 BIO Report 2003]). This question is discussed in more detail below, but finding even a tentative answer will not be easy -- it is a major objective of this project.

A second issue that has been raised as a possible obstacle to establishing a successful open source biotechnology project is the need to comply with biosafety and biosecurity regulations. This is also a requirement for non-open source biotechnology research and development, and it is not clear that the costs of compliance would necessarily be higher for an open source project than for other projects. The existence of regulations also does not of itself constitute a difference between biotechnology and software: for example, in the US software is subject to export controls for national security reasons, and this has not prevented open source development (although it is an issue that has had to be dealt with: open source licences are not allowed to discriminate among users on the ground of nationality or any other ground). However, this does not mean there are no differences in the costs of compliance between the software and biotechnology contexts; one obvious point is that food and especially medicine are two key areas of application for biotechnology research tools, and these areas are both subject to regulatory processes that are both expensive and time-consuming. It has been suggested that these costs may "raise the bar" in terms of motivation to contribute to an open source project so high that open source development is simply not feasible. For example, we saw earlier that experience in the software context suggests contributors should be able to see their contributions incorporated into the technology as quickly as possible. One preliminary response to this concern is that the costs of obtaining approval for a new food or medicine will generally be incurred after the process of development has been completed, so that they need not necessarily affect that process. (This answer may seem surprising given that strong intellectual property protection for pharmaceutical products has long been justified by reference to a supposed lack of incentives to invest in research and development in the face of high regulatory costs, but it is important to realise that this justification presupposes that the dominant incentive to invest in research and development is economic, whereas an open source development project, even if it is initiated by a business for economic gain, relies on contributors being motivated by incentives other than economic incentives.) If this is true then in Benkler's terms the problem of compliance costs is an integration problem that might be solved by the intervention of a firm in the closing stages of the project.

Of course, the costs of complying with regulations are not the only issue being raised here. The real concern may be that if open source biotechnology eventually allows people to hack new viruses in their garage, it could create a serious public health and/or security risk. In response, some would argue that the widespread release of GM crops into the environment without long-range environmental impact studies over the last few years shows that industry and governments cannot be trusted with biosafety issues either. In fact, the fact that an open source approach to biotechnology research and development may have the capacity to weaken government (in particular, US government) and industry control over the rate and especially the direction of scientific progress in this field is part of its appeal. Acknowledging, however, that this argument does not resolve the issue completely, it is worth noting that when similar concerns have arisen in the software context, one counterargument has been that no matter how tightly controlled the technology may be, people who wish to use it for (e.g.) terrorist purposes will always be able to obtain access somehow through sheer commitment to the goal, and when that happens, the safety of the majority lies in open access to the technology so that there may be as many technically competent defenders as possible: in biotechnology terms, the more people who are equipped to look for a vaccine to a newly created virus, the more quickly a vaccine will be found. The "safety in numbers" argument also applies to unintentionally created threats. Ultimately, however, these are broad public policy issues that cannot and perhaps need not be resolved in the context of this project.

A third issue relating to the difference in subject matter between the software and biotechnology contexts that may affect the overall trade-off between open source and more traditional business approaches is that biological functions are highly conserved at the molecular level, so that in relation to many biotechnology research tools there will be very limited scope for substitution or imitation or even further innovation. [Charles Lawson has made this point in relation to Taq polymerase and erythropoetin.] Exactly how the difficulty (or even impossibility) of "inventing around" some biotechnology research tools will affect the open source-closed source trade-off is a matter for further consideration. Clearly, this characteristic of biotechnology research tools is relevant to the question whether they should be protected by patent at all, but this is not a concern of the present study.

Another potential obstacle to open source biotechnology alleged in response to Dan Burk's paper titled "Open Source Genomics" is that the research and development process in biotechnology is characterised by more obvious endpoints than is the process of software development. It is not clear whether this is true either in general or for any given project, but in any case, it has been argued that these obvious endpoints would lead to a collective action problem for cooperative projects because as the project nears its endpoint, the incentive for each participant to defect from cooperation and reap the benefits on its own increases; if the endpoint can be seen from the start, cooperation tends to unravel by a process of induction backwards from the endpoint. This argument is based on the experience among cartel members of difficulties in enforcing member loyalty to the scheme (Peter Drahos has argued that increasing reliance on intellectual property by multinational firms is in fact a response to this problem). The major flaw in this argument is that the very ability to "defect" from an open source project would require the legal failure of the licence scheme. Assuming it is possible to come up with a legally enforcable open source biotechnology licence, such a licence would provide security against defection: that would be its purpose.

A further perceived difference between software development and biotechnology R&D is cultural. It is argued that the history of commercialisation in biotechnology is such that members of the biotechnology community are more committed to making large profits from research and development than are members of the software community. This seems a rather drastic generalisation, that has not (as far as I know) been backed by empirical evidence. It may be true that investor expectations are higher in the biotechnology context (see next paragraph). However, in relation to the general point about culture, it can equally be argued that there are significant cultural similarities between the biotech and software development communities. Some of these are a matter of divergent evolution -- both communities originally came from, and remain substantially rooted in, an academic environment. Other similarities may be a result of the bioinformatics revolution, which is closing the disciplinary gap between molecular biology and computer science; as genomic sequence information becomes available for more and more organisms, the cultural impact of this disciplinary overlap is likely to spread. (In fact, bioinformatics is one field in which nearly all the relevant software is open source). As in the software context, where programmers are often said to be motivated to participate in open source projects by a desire to enhance their reputations, reputation is a strong motivator in the biotechnology research community.

Another issue to be considered when comparing the likely outcome of the open source/closed source trade-off in the software and biotechnology context is whether in the biotechnology context there exist viable equivalents to the secondary markets companies are starting to exploit in the software context (discussed above in the section on open source business models). As noted earlier, from a commercial perspective this consideration boils down to what specific value the company would be adding, what it would be worth from other customers' point of view, and whether that would be enough to support a business (or in the case of non-profit institutions, to make up for a funding shortfall). The question then becomes what is "enough"? The answer depends, among other things, on investors expectations, i.e. what return is regarded as sufficient over what timeframe. In the pharmaceutical industry, some investors may be used to demanding 25% profit per annum, which is quite an ask for any business, open source or not, but whether this expectation is usual I do not know, and whether it is much higher than in analogous circumstances in the software industry is not clear. Certainly, no business model is bombproof, and recently there has been some commercial fallout in the open source software context. The point is that, as the first section under this heading shows, it is possible to construct a plausible hypothetical scenario in which it may be possible for a biotechnology company to improve its overall profitability by adopting an open source strategy. Whether there is any real company for which this is true comes down to specifics and is the question this project will attempt to answer.

Finally, there is the question of demand for open source products in biotechnology. This is related to the subject matter of the previous paragraph, in that it essentially deals with the question whether there is sufficient demand for such products to sustain secondary markets that can support commercial enterprise. Some users of biotechnology research tools may value open source products because of their mode of production, just as many consumerse prefer to buy from companies that respect the rights of their employees. However, it has been pointed out in the software context that ultimately, open source itself is merely a feature of production; the benefit for users from using open source tools is control. Thus, we must ask whether choice and control are qualities that are valued by users/consumers of biotechnology research tools. I would argue the answer is yes. For example, although the consumer backlash against genetically modified foods has been complex, one of its most striking themes has been resentment at having the decision whether to eat or grow genetically engineered products.

Comments on open source and drug development

In this section, "Maurer's proposal" refers to S. Maurer, New Institutions for Doing Science: From Databases to Open Source Biology, conference paper, European Policy for Intellectual Property, University of Maastricht, November 2003, available at http://ist-socrates.berkeley.edu/~scotch/maurer.htm.

Maurer's proposal

As I understand it, Maurer's proposal emphasises the idea of harnessing the input of a large number of unpaid volunteers to scan biological information (eg genomic sequence information) in order to identify potential drug targets. The motivation is that this will be cheaper than paying full-time researchers and also more effective because, as the software people say, "given enough eyes, all bugs are shallow". It seems like a good idea, assuming it solves a relevant technical problem in identifying drug targets (ie lack of computational and human "scanning" resources). How much difference it will make to drug development costs overall depends on whether that phase is really the bottleneck in terms of expense.

Obviously, the proposal is "open source" in the sense of adopting a relatively non-hierachical contribution and management structure. It's not entirely clear whether it is also intended to be "open source" with regard to IP. The connection between the development methodology and IP aspects of open source is essentially that an open source approach to IP helps to motivate volunteer contributors by assuring them that they will have continuing access to the fruits of their collective efforts. (Even here we have a definitional problem: most open source licences have this effect, but not all - the ones that do are the GPL-style "copyleft" licences that require derivative works to be licensed on the same terms as the original works.) In the software context, this assurance comes at a cost in terms of tracking the ownership of each module that is contributed to a software program. In the biotech/pharma context, the cost is higher because you aren't just dealing with IP pedigree problems, you're also dealing with the costs of obtaining patent protection, which are considerable (unlike with copyright in software, which arises automatically and without cost to the author). So you have to ask yourself whether in this particular case the extra motivating effect of assuring contributors that their collective efforts won't be "hijacked" by some corporation for its own profit is worth the extra transaction costs associated with going open source instead of just having the contributors donate their work to the public domain. In this particular case you might be able to create sufficient motivation for people to contribute in other ways.

In this connection, note that some people seem to use the terms "open source" and "public domain" pretty much interchangeably (ie they completely ignore the IP aspect of open source). I think this blurring of ideas is particularly important to avoid when you are talking about drug development because it obscures the fundamental policy choice that society makes when it relies on private corporations to conduct drug development instead of funding it directly from the public purse. In a sense, open source is tinkering around the edges of the status quo, and so talking about open source as if it were the same thing as public domain immediately removes from consideration a whole swathe of other options, like nationalised health provision, that might make better sense from a social welfare perspective. In my experience it is difficult to get people in the US to actually consider the full range of options because there is such a strong commitment to the private investment model of funding R&D that open source looks really radical. It isn't radical at all when you look at the full spectrum of possibilities. A major issue with drug development, from this point of view, is simply that market forces will naturally tend to direct research efforts by the private sector to where there is the most substantial potential return. The higher the costs of doing the research, the higher the return must be before it will justify private investment: so lowering R&D costs by means of a proposal like Maurer's may help, but if you're still relying on private investment in the later stages of development, it won't actually shift the focus away from profitability in the marketplace. All it does is lower the threshold (somewhat) for profitability - which is great except that if other opportunities are still more profitable, these much-needed drugs will still be neglected.

To illustrate the point, look at the Salk vaccine story. Maurer's proposal goes against the conventional wisdom that you will encounter everywhere, that drugs themselves could never be open sourced because no pharmaceutical company would ever make the necessary investment in developing the drug if all its competitors, with roughly equal resources, can do the same and only one drug will result. Why go into a competitive environment if you can avoid it? In this regard the experience of pharmaceutical companies' involvement in the development of the Salk polio vaccine in the early 1950s is instructive. The university lab in which the vaccine was created did not have the capacity to generate enough vaccine for a large-scale field trial, so there was a recognised need for the involvement of pharmaceutical firms, which had the appropriate infrastructure in place. Getting the vaccine through the field trial stage was going to require substantial investment, was going to be difficult and complex and entailed substantial risk - before a field trial it was uncertain whether the vaccine would ever be approved. However, it was clear that if it could get past the field trial stage, the vaccine would be very profitable because the degree of public fear of polio would ensure all parents would want their children vaccinated. This situation, where you have a medicine which would have a market but needs substantial further development with no guarantee of success (ie a high risk, high payoff investment), is the classic scenario where there is assumed to be an incentive problem that should be solved by patent monopoly. (With low risk, there's no incentive problem, and with low payoff nobody argues that patent rights are a solution as there is only value in patent rights if there is a market for the product.) In this case there was no patent, because the vaccine was in the public domain, so if the patent proponents' argument is correct you would expect insufficient incentive. But companies were interested without patents and even in the end without any exclusive contract to produce field trial vaccine, because the current pie (government procurement for the field trial) was some guaranteed income to cover at least part of the investment, the eventual pie (if approved) was going to be big enough and the costs of imitation were high enough (ie there would be a significant lead-time advantage in being ready to roll if the vaccine was approved) that they felt the risk was worth taking on commercial grounds.

Now this story is often told to illustrate the fact that you don't necessarily need a strong exclusive patent position to induce pharmaceutical companies to participate in drug development (Maurer's point). BUT I would point out that it also illustrates the fact that the key inducement is the size of the eventual market for the product. If there's no big market - and this includes the case of malaria drugs, where even though there are millions of people who would benefit from access to the drug, those people don't have any money to pay for them - then you won't get drug development with or without a patent. It's not obvious how the open source approach that Maurer is putting forward will solve this problem. Ultimately you still get down to the question of whether society as a whole thinks it is worth paying for the development of a particular drug, and if so, whether it would be more efficient to pay for it directly or indirectly.

My views on open source and pharmaceutical companies

My own emphasis re the concept of open source is a little different from the one underpinning Maurer's proposal. In adopting the open source analogy, he explicity takes the Linux project as his model and emphasises the social/development methodology aspects of open source. When I talk about open source in biology, I am specifically interested in whether private companies in the biotech and pharmaceutical sector might find it commercially advantageous to make some of their IP publicly available without seeking direct compensation in the form of licence fees or other pecuniary consideration. (That is, I use the term "open source" to evoke the reasons why some software developers chose to talk about "open source" rather than "free" software, ie as a strategy to promote the idea of free software to the commercial world on pragmatic rather than ideological grounds.)

The obvious question for a for-profit company (eg a pharmaceutical company) is why give away the crown jewels? Why would I donate free access to my intellectual property, that I have generated through my private investment, to everyone -- including my competitors? And the answer is that there are other ways for a firm to benefit from the adoption and use of its IP, eg:
1. free revealing of IP may establish an advantageous industry standard which may be useful to a company that hopes to commercialise complementary goods and services that sit on top of that technology platform. In other words, free revealing can lead to growth in a secondary market that may bring in more revenue than it loses.
2. free revealing may allow or facilitate improvements to the core technology. If the original innovator then gains access to those improvements, this represents a cost saving in R&D for that company.
3. by revealing its IP, a company may generate a favourable reputation that is useful in selling associated offerings, by enhancing brand value or enhancing the company's ability to attract and keep high quality employees.

As with any other strategy, there are costs as well as benefits associated with an open source approach. Opportunity costs are the gains that an innovator could have made by adopting an exclusive proprietary approach according to the traditional model in biotech and elsewhere. Actual costs include the costs of producing and then diffusing an innovation (if you choose to actively build a user community around your open source product, obviously maintaining and supporting that community will entail extra costs).

Whether the balance of costs and benefits of an open source approach make it more attractive to an IP owner than the traditional proprietary approach will depend on the circumstances. From what I've seen, the likelihood of pharmaceutical companies "open sourcing" their drugs - the actual therapeutic molecules - seems very low. There are two main reasons for this:
1. Opportunity costs associated with open source. The major opportunity cost of adopting an open source exploitation strategy is the loss of profits you might have made if you had taken an exclusive proprietary approach. In most fields that opportunity cost is actually not all that high, because it turns out there are serious caps on the return you can realistically get from a limited number of licensees. The most important reason for this is that because the patent grant technically covers the means of achieving an end, not the end in itself, patents in most fields are quite easy to invent around, so in effect, the upper limit of the licensing fee you can charge is the estimated cost to potential licensee of coming up with an alternative way of achieving the same goal. Interestingly for our purposes, pharmaceuticals are one of only two exceptions to the rule that patent ownership is not all that profitable in practice (the other, not co-incidentally, is chemicals). The main reason for this is that over the years the pharmaceutical industry has successfully pushed for patent grants that are broad enough to effectively cover not just a particular molecule that happens to have value as a drug, but all the variations of that molecule that might be effective, and this means that pharmaceutical patents are actually almost impossible to invent around. So the opportunity costs for a pharmaceutical company in giving up an exclusive proprietary approach to drugs in favour of an open source approach are likely to be too high for pharma to be interested.
2. Established business practice: Pharmaceutical companies rely heavily on patent positions - whether or not they have a patent on a particular pharmaceutical product makes a big difference both to the results of their operations in terms of the economics of selling drugs, and to the company's valuation on the stock market; and this translates into a big emphasis on IP, to a degree even in areas that aren't really related to their proprietary position on their drug products. In other words, not sharing IP is deeply ingrained. To a large degree the biotech industry has inherited that culture of not sharing. The costs of changing established business practice are very real in terms of organisational structure and providing incentives for your employees to shift the way they look at things - and so there is plenty of inertia for large, established companies like the big pharma companies when it comes to adopting fundamentally new business models like open source.

But that doesn't mean open source has no application in relation to the pharmaceutical industry. For example:
1. One established use of open source business strategies in the software context is to pre-empt the establishment of proprietary technological standards owned by your rivals. Even though pharmaceutical companies hate sharing, one thing they hate even more is being beholden to a single supplier for some critical value driver. This means they might be prepared to put money into an open source biotech company that planned to come up with a really critical tool -- say a toxicology tool that would help predict R&D failures before a drug hit the expensive clinical phase of development. This is a significant example because currently, exclusive patent positions are important to biotech start-ups largely as a way to attract capital. So this kind of support would provide a credible alternative story for biotech start-ups to tell potential investors and thus could promote the development of open source strategies in the biotechnology sector.
2. Using open source as a form of leverage to gain access to improvements to its technology that other people have made for their own purposes: this is of course a form of pre-competitive collaboration, well known in other industries despite high capital costs and time investments. Ironically, the closest to a true open source arrangement that I've come across in interviews was in the heart of exclusive IP territory - in the business model of a successful biotech firm that provides data to pharmaceutical companies. This biotech firm had a licensing arrangement where if any of their customers discovers and characterises a full length gene using the information in their database, the customer has to grant back to the biotech company AND to all of its other customers non-exclusive freedom to operate in drug discovery. Initially this clause was inserted in order to allow customers to use the data without fear of infringement suits from other customers, but it came to be seen as an additional source of value - customers aren't just getting access to the biotech firm's data, they're also getting access to information from all of their rivals. The most interesting part of this story is that this has been happening since the biotech firm set up shop in the early 1990s, some time before Linux took off.

It's also important to note that even if only one or two companies or institutions in a given industry sector choose to go open source, the actions of those companies have a big impact. By undermining customers’ willingness to pay for access to tools from one company that they can get at a lower cost or even for free elsewhere, a couple of open source players can shift the basis of competition in the sector away from proprietary technologies. So in this sense, open source really has the power to transform industries. This is one reason why Maurer's proposal is quite exciting - but it also means you'd expect resistance from established firms to anything that looked like open source. We shouldn't underestimate the power of big pharma to influence how open source is perceived if they felt it would be to their disadvantage. Also note that it wouldn't have to be a company that provided open source technologies in order to have this industry-wide effect: it could be a non-profit organisation.

Applicability of open source in non-IT intensive areas

There are a few things to say here.

First, one objection to the idea of open source in non-software biotech applications relates to the cost of conducting distributed R&D where the innovation in question is not just a string of software code that can be cheaply and instantaneously sent from one collaborator to another via the internet, but may also have a material element and in any case is not likely to be perfectly well defined in an information sense (eg a GM mouse or a cell line). If you talk to people who are already running large scale, geographically dispersed biological research collaborations, it turns out these costs are in fact not a significant drain on resources or an obstacle to progress.

Second, another objection relates to the supposed difficulty of building a user community in biotech due to the greater skill needed by biotech researchers relative to software programmers. The idea is that the pool of potential contributors to a project would be too small to get it off the ground. I don’t think this is a serious issue. It’s true that Linux harnesses input from thousands of contributors, but in this respect it’s actually atypical: most open source software projects have fewer than twenty contributors. It’s also true that although some biological research requires high skill levels, a lot of it is basically gruntwork.

Third, another common objection in relation to open source biotech is that the cost of actually doing the research was significantly higher than in software development.
Speaking to people running research groups in both software and biotech, I've found that labour costs were by far the largest cost for both types of research, with the cost of supplies and other wet lab inventory being relatively small and generally on the decline. Of course in really “big science” areas like the human genome project, the work is on such a huge scale that these sorts of costs start to really add up. But in general, the ability to do more and more research using computers instead of chemicals is making biological research more and more accessible to people without huge resources. In that sense there isn't really any stage of the drug development process that is not in some sense "IT intensive" these days - or more accurately, in areas where the process is still largely uncomputerised (eg clinical trials), there is nevertheless plenty of scope for computer applications that could improve the process.

As you can see from my earlier comments, I don't actually think that whether open source is relevant in a particular context really depends on absolute costs of resources or the absolute amount of investment required. Instead, it's a trade-off between exclusive and non-exclusive exploitation strategies for a given innovation -- a matter of relative costs and benefits, not absolutes. For many software applications, the exclusive approach is rational; conversely, for many biotech applications, a non-exclusive/open source approach will be rational. The innovation literature is full of examples of essentially open source strategies that pre-date Linux and the free software movement (going back to the 19th century and even further) that brought plenty of benefits to commercial players in industries that are much more capital intensive than software - including textiles and steel manufacturing.

Conclusion

The preceding discussion raises many issues that cannot yet be resolved, but does not appear to indicate there is any fundamental difference between the software and biotechnology context which renders the open source approach generally irrelevant to biotechnology research and development. Open source licensing will not be suitable to every biotechnology enterprise, just as it is not always the best scheme for developing software; on the other hand, charging for access to intellectual property is also no guarantee of business success.

Key issues for advancing the open source biotechnology analysis will be developing open source patent licences and other licences appropriate for biotechnological subject matter, assessing the importance of higher capital costs in biotechnology development and establishing whether or not there exist secondary markets for biotechnology services or other commercial offerings that might support business models along the lines that have proved successful in the software context. Other important issues include the costs of exchanging uncodified information and the amenability of biotechnology research to being broken up into relatively fine-grained modules that can be developed in parallel by contributors possessing varying degrees of motivation (refer to Benkler).

The purpose of this study is to examine these and other issues through interviews, case studies and literature analysis. The ultimate goal is to either demonstrate the commercial feasibility or otherwise of an open source approach to biotechnology research and development. Preliminary interviews suggest that many biotechnology researchers are interested in this question (though opinions are heavily polarised), but that advancing the analysis is secondary to meeting research deadlines and existing funding requirements. It is hoped that this project will facilitate discussion of issues relating to open source biotechnology; in adopting this objective the project is intended to supplement rather than supplant other similar efforts, including the Open Source Biotechnology initiative at the Molecular Sciences Institue, the open source bioinformatics and neuroinformatics movements, Tom Knight's (MIT) standard biobricks project, Tom Michaels' (OAC) General Public Licence for Plant Germplasm, the open hardware movement, Al Gilman's Alliance for Cellular Signalling, and the Human Genome Project's Bermuda Rules (no doubt there are many others).

Post Script: A speculative future for open source biotechnology

The following extracts are taken from verbatim transcripts of seminars I delivered during 2002 on the topic of open source biotechnnology. Some of the material has also been used elsewhere in this website. This page is very much under construction! The point to keep in mind is that nobody can predict what will happen with respect to open source biotechnology. But it's tempting to think about what could happen....

Seminar 1:

"What would happen if open source biotech schemes starting taking off? Besides just being a superior business model in some cases, open source approach could be used not just for tactical advantage but also as a strategic weapon, to reshape markets by establishing open standards, preventing other players from getting a proprietary choke hold on important tools and resetting the competition in favour of weaker players. Rebecca Eisenberg has suggested that the similarities between software development and genomics in connection with the open source approach may be more ideological and rhetorical than economic. If Eisenberg says this is true in biomedical biotechnology, then I think we can believe her, but it seems to me there are strong parallels between the software industry, dominated by Microsoft, and the agbiotech industry, dominated by a handful of powerful oligopolists. In both cases, smaller organisations that try to compete with the big players for their own turf are sensibly crushed by the monopoly. The imbalance of power and resources is just too great. But monopolies will never offer their customers choice and control over the product itself -- that is the source of their power. However, this is something that customers really want, in biotech as in the computer world. The consumer backlash against genetically modified foods has been complex, but one of its most striking themes has been resentment at having the decision whether to eat this stuff or grow this stuff taken out of one's hands.

"One important question, supposing open source biotech did get off the ground, would be how the big players would be likely to react. What they perceive open source as a threat and take action to destroy it? Or would they see some benefits in it for themselves and take a more benevolent line? In the software world, Microsoft's attempts to remove their main open source competitor, Linux, have failed. It might be fair say that open source licences are popular precisely because they are a "monopoly resistant" business scheme in an industry which has otherwise been very friendly to the dominant players.

"One optimistic vision that has been put to me is that if an open source initiative began developing the next revolutionary tool in agbiotech, then instead of trying to block this initiative and racing to get there first, the big players might calculate that the value of having freedom to operate in relation to such a tool would outweigh the fact that everyone else would have access to it as well. My first reaction to this vision, if indeed I understood it rightly, was that if an open source initiative could envisage such a revolutionary new tool, surely so could the oligopolists, and instead of supporting as open initiative, they would rush to gain monopoly control over it. But perhaps in the light of public-private sector initiatives like the SNP consortium, I am underestimating the value of freedom to operate for these big firms. A more pessimistic vision would be that the big players would take action to undermine the use of open source as a competitive weapon.

"These speculations relate to the medium-term future, and of course they are highly uncertain. Perhaps the most likely long-term outcome of a move towards open source biotech would be neither be completely open source industry nor total destruction of the open source impulse by big players. Instead we might end up with a kind of ecosystem, incorporating both open source and proprietary elements. We might expect that scientific resources that represent non-substitutable standards -- for example, the sequence of the human genome -- would be open source, while standalone vertical markets where network effects are weak would tend to be populated mainly by proprietary tools. In such an ecosystem, there would be a role for competition law to intervene to maintain a productive equilibrium between proprietary and open source elements. Competition law could also, of course, play an important role in getting us there from here."

Seminar 2:

"Okay, so I've spent quite a lot of time describing the individual organisation's perspective on going open source, and so far that's the perspective to which I've devoted the most thought. Now I just want to make a couple of brief comments about the broader industry or community perspective.

"The image on the slide is of David and Goliath, which is to remind us that there are substantial imbalances of power and resources in the agbiotech industry and the industry as a whole. In the last few years ag biotech has come to be dominated by an oligopoly of only about 5 large firms, and one reason that many research organisations outside that oligopoly are interested in initiatives like the IP clearinghouse is that they are looking for ways to counter the effects of oligopoly ownership of enabling technologies - that is, critical research tools.

"If the open source business model were to take hold in the ag biotech industry and then perhaps in biomedical biotech, how would that affect industry structure? Would it in fact favour the smaller players, as the David and Goliath analogy suggests, or would it favour those who already control most of the resources?

"It's interesting to note that in the software context, leaked documents have revealed that Microsoft's attitude to Linux is one of fear and hostility, and it would not be surprising if we were to see a serious backlash against open source. Would larger players in the agbiotech industry react any differently to the growth of a successful open source community? What about the biotech industry generally - pharmaceutical firms are notorious for playing hard ball over IP issues.

"At this stage of my research I'm not ready to propose answers to these questions, but I have thought a bit about how to carry out the analysis. The difficulty is that like any complex system, technology innovation systems will react unpredictably to change. This means that to study them effectively, you have to apply the principles of system dynamics. Unfortunately, these principles are currently fairly inaccessible because they are generally expressed in purely mathematical terms. At present, I'm in the early stages of a side project aiming to articulate system dynamics principles in a rigorous but non mathematical way so they can be applied to this particular problem.

From a more traditional legal perspective, of course, systems theory is a branch of regulatory theory, so if you look at the question of the likely industry effects of an open source biotechnology movement, in legal terms you are really asking whether and how governments -- or other players -- should intervene to bring about desired policy effects in what is otherwise a self-regulating system. So to summarise with respect to industry effects, I will be drawing on both system dynamics and legal regulatory theory to examine these issues."

References

Alliance for Cellular Signaling 2003. [Internet]. Alliance for Cellular Signaling 2002 [cited 3 January 2003]. Available from http://www.cellularsignaling.org/.

HTTP://www.technologyreview.com/articles/innovation60502.asp;

Campbell, Eric G., Brian R. Clarridge, Manjusha Gokhale, Lauren Birenbaum, Stephen Hilgartner, Neil A. Holtzman, and David Blumenthal. 2002. Data Withholding in Academic Genetics: Evidence from a National Survey. Journal of the American Medical Association 287 (4):473 - 479.

Benkler. 2002. Coase's Penguin, or, Linux and The Nature of the Firm. 112 Yale L.J. (Winter 2002-03). Available at http://www.law.nyu.edu/benklery/.

Seaman, Graham. 1998. Open Collector. http://opencollector.org/Whyfree/

Sulston, John. (forthcoming, 2002). Intellectual Property and the Human Genome. In Global Intellectual Property Rights: Knowledge, Access and Development, edited by P. Drahos and R. Mayne: Palgrave.

Knight, Tom. Idempotent Vector Design for Standard Assembly of Biobricks. MIT Artificial Intelligence Laboratory, www.syntheticbiology.org/docs/sa3.pdf http://www.syntheticbiology.org/docs/sa3.pdf

Thompson, Nicholas. 2002. May the Source Be With You. The Washington Monthly Online, 1-5.

Michaels, Tom. 1999. General Public Release for Plant Germplasm: A proposal by Tom Michaels, Professor of Plant Agriculture, University of Guelph, v1.1, 26, February 1999, http://www.oac.uoguelph.ca/www/CRSC/pltag/1998-99/gnucrop2.htm

von Gavel, Stephanie. 2001. Biotechnology Licensing. Paper read at Licensing Course 2001, at Leuven-Belgium, 13-14 December 2001.

Eckersley, Peter, et. al. 2002. Neuroscience Data and Tool Sharing: a Legal & Policy Framework for Neuroinformatics. Melbourne.

Michael. 2002. OSI Approves Two New Licences [internet discussion group posting]. Slashdot 2002 [cited September 29 2002].

Shewmaker, Mark. Open Patent Licence 2002 [cited. Available from http://www.openpatents.org/.

Wellcome Trust. 1997. Summary of the Report of the Second International Strategy Meeting on Human Genome Sequencing. Paper read at Second International Strategy Meeting on Human Genome Sequencing, at Bermuda, 27 February-2 March 1997.

HUGO Statement on Patenting of Gene Sequences 2003. Human Genome Organisation 2000 [cited 21 Jan 2003]. Available from http://www.hugo-international.org/hugo/patent2000.html.

Wellcome Trust. 1996. Summary of Principles Agreed at the International Strategy Meeting on Human Genome Sequencing. Paper read at International Strategy Meeting on Human Genome Sequencing, at Bermuda, 25 - 28 February 1996.

Marshall, Eliot. 2001. Bermuda Rules: Community Spirit, With Teeth. Science 291 (5507) (16 February 2001):1192.

McNaughton, Noel. 2002. Sharing Germplasm [Internet-- collection of articles and commentaries]. Raven's Eye Consulting 1999 [cited 24 July 2002].

Lawson, Charles. 2002. Patenting genes and gene sequences and competition: patenting at the expense of competition. Federal Law Review 30:97-133.

Cringely, Robert X. 2002. The True Believer: Can Mike Doyle do to Microsoft what the rest of the computer industry and the Department of Justice couldn't do? The Pulpit, November 7 2002.

Perens, Bruce. 1999. The Open Source Definition. In Open Sources: Voices from the Open Source Revolution, edited by C. DiBona, S. Ockman and M. Stone. Online version: O'Reilly. Available from :

Young, Robert. 1999. Giving It Away: How Red Hat Software Stumbled Across a New Economic Model and Helped Improve an Industry. In Open Sources: Voices from the Open Source Revolution, edited by C. DiBona, S. Ockman and M. Stone. Online version: O'Reilly. Available from :

http://www.dodds.net/~hozer/opensource.html, Troy Benjegerdes, Industry Analysis Paper, linked from http://www.opensource.org/advocacy/case_for_business.phpOSI:http://www.opensource.org/advocacy/faq.php, Frequently Asked Questions, OSI

Burk, Dan. 2002. Open Source Genomics. Boston University Journal of Science and Technology Law 8 (Symposium on Bioinformatics and Intellectual Property Law, April 27, 2001, Boston, Mass. (Winter 2002)):254-.

The Magic Cauldron, in The Cathedral and the Bazaar, in Raymond, Eric S. 2001. The Cathedral and the Bazaar: Musings on Linux and Open Source by an Accidental Revolutionary. 2001. Online version: O'Reilly.

HTTP://www.egenetics.com, Electric Genetics

OpenInformatics.org (http://www.openinformatics.org/)

Open-Bio (the Open Bioinformatics Foundation) (HTTP://open-bio.org/)

Bioinformatics.org (HTTP://bioinformatic.org/about/)

Please Check EGO at Door. 2001. Molecular Interventions 1:14-21.

Return to Home