The ‘MIT model’ and the history of hacker culture

Lisp Machines Inc. brochure
Lisp Machines Inc. brochure

In my previous post from this series, I outlined a period in the early 1980s when the work of hackers at MIT was commercialised through the use of venture capital and as a result, those hackers stopped sharing code. As a response to this, Richard Stallman left MIT to start the GNU project and within a few years had created the GPL ‘copyleft’ license, the most popular open source license in use today.

I concluded by pointing to the Bayh-Dole Act (1980) as an event that is worth understanding when examining the role of universities in the history of hacker culture. In this post, I want to outline what I mean by this and throw out a few ideas that, I admit, need to be more fully explored.

The Bayh-Dole Act

The Bayh-Dole Act was enacted by US Congress in December 1980 to clarify title of ownership and encourage universities to freely exploit the IP they generated through government funded research. Until that time IP arising from research that was federally funded was, by default, owned by the government and it was left to each federal agency to determine IP arrangements. Since WWII, the majority of research taking place in US universities was (and remains) government funded and in the midst of the industrial downturn of the 1970s, IP resulting from research was recognised as an under-exploited source of national economic potential. ((In 2004 it was 62% of all university research. See AAU (2006) University Research: The Role of Government Funding))

As an aside, in the UK, the Patent Act of 1977 gave employers the legal entitlement to employees’ inventions, but it wasn’t until the Department of Trade and Industry’s White Paper, Realising our Potential (1993), that universities were encouraged to pursue patents arising from their research. Similarly, other countries followed the Bayh-Dole Act in the 1990s and 2000s. ((Rigby and Ramlogan outline this in the 2012 report, Compendium of Evidence on the Effectiveness of Innovation Policy Intervention: Support Measures for Exploiting Intellectual Property.)) Until that time, the title to IP generated by publicly funded research in UK universities was usually controlled by the National Research Development Corporation (1948-1981), the National Enterprise Board (1975-1981) or the British Technology Group (under public ownership from 1981-1991).

Similarly, prior to the Bayh-Dole Act in the US, federally funded IP was first controlled by the National Defence Research Committee (NDRC) (1940-1947), then the Office for Scientific Research and Development (OSRD) (1941-1947), and then by several organisations: the NSF, NIH, DOD, NASA, DOE, USDA and others. ((See Mowery, (2004) Ivory Tower and Industrial Innovation: University-Industry Technology Transfer Before and After the Bayh-Dole Act.)) Etzkowitz has noted that despite significant levels of federal funding to US universities since WWII, in 1978, less than 4% of government-owned patents had been licensed. By creating a mechanism for universities to own the title to their research, the Bayh-Dole Act provided an incentive and driver for institutional change and whereas “previously only very few universities had the interest and capabilities to patent and license technology invented on campus”, during 1980-1990, the number of universities with technology transfer offices went from 25 to 200. Etzkowitz argues that this played a role in the development of Silicone Valley and made research universities “an explicit part of the US innovation system by restructuring the relationship among university, industry and government.” Out of the Bayh-Dole Act, arose incubators, research parks, technology transfer arrangements and other entrepreneurial features of modern universities.

When I began to think about the Bayh-Dole Act, I wondered what effect it might have had on Richard Stallman, who around the same time as the Act was in place, saw hackers leaving the AI Lab and joining two start-up companies (Lisp Machines Inc. and Symbolics Inc.) which were pursuing the exploitation of government-funded research and development that originated at MIT.  Was the Bayh-Dole Act a catalyst in the development of the GNU project? The answer was, no, not really. In pursuing this line of inquiry I contacted Stallman who said that he wasn’t aware of the Bayh-Dole Act at that time.

“In 1980 we all supported commercial fabrication of Lisp machines, because we wanted people to be able to buy them.  Thus, no pressure on us was needed on that particular point. Only the details were controversial.  And we did not foresee the consequences. It could be that MIT’s method of releasing the source code to the two competing companies, which both made it proprietary and set the stage for the software war of 82/83, was facilitated in some way by Bayh-Dole. But I don’t know whether that is so.” ((Email from Stallman, 19th December 2012))

Stallman’s reply is a consistent reminder that his work has never been in opposition to the commercial exploitation of software, but rather the prevention of restrictions on freedom. Although there were clearly differences of opinion among MIT hackers about the way in which Lisp Machines should be commercialised, with a minority opposed to VC funding, the culture of MIT in 1980 was such that the Bayh-Dole Act was following MIT, rather than imposing anything significantly different onto its technology transfer processes. The Act was certainly one of several instruments that provided clarity around the ownership and commercial potential of software and other IP during that period (others were the Copyright Act of 1976 and two amendments in 1980), and by requiring all US universities to consider ways in which government funded research could be commercially exploited, research was to some extent marketised, creating a more intensive environment for technology transfer in which MIT and other universities found themselves competing. Etzkowitz summarises this as follows:

“Starting from practices developed early in the twentieth century at MIT, university technology transfer had become [with the Bayh-Dole Act] a significant input into industrial development. William Barton Roger’s mid-nineteenth-century vision of a university that would infuse industry with new technology has become universalied from a single school to the entire US academic research system. Greatly expanded with federal research funding, the US academic enterprise has become a key element of an indirect US industrial policy, involving university, industry and government. The origins and effects of the Bayh-Dole Act are a significant chapter in the spread of the MIT model and the rise of entrepreneurial science.” (Etzkowitz, 113)

The question then, is not about what effect the Bayh-Dole Act had on Stallman and his fellow hackers in 1980, but rather what was the ‘MIT model’, which was later universalised by the Bayh-Dole Act, and what unique role did the ‘MIT model’ play in the history of hacker culture?

The MIT model

MIT began as a ‘Land Grant’ university, partially funded by a government grant to establish science-based universities which would “promote the liberal and practical education of the industrial classes” ((Read a transcript of the Morrill Act)) while undertaking research to improve agriculture. Land Grants were provided under the Morrill Act of 1862 and were a response to many years of campaigning by farmers and agriculturalists for research institutions that would contribute to the improvement of US farming. The Act led to States being allocated federal land which was to be sold or donated in order to establish such universities. The European Polytechnic movement was also gaining popularity in the US and seen as a model for new applied science universities in contrast to the largely teaching universities that existed at that time. Following the Morrill Act, the Hatch Act (1867) and Smith Lever Act (1918) again encouraged applied research in US universities as well as building capacity for technology transfers, again with a specific focus on the needs of agriculture.

Until the Land Grant universities of the late 19th century, there were no ‘research universities’ in the US and even academic staff dedicated to research were rare. ((Richard C. Atkinson, William A. Blanpied (2008), Research Universities: Core of the US science and technology system)) Founded as both a teaching and research university with a remit to undertake applied science that could be transferred to industry, MIT has always had close contact with private enterprise; from early on in its history MIT had employed engineers from industry as members of its academic faculty. By the 1920s, these academics were noted for their consulting activities to the extent that there was tension within MIT between academics who felt that it was their job to focus on teaching and the needs of students, and those who spent a significant portion of their time focusing on the needs of industry. During the Great Depression of the 1930s, MIT was forced to confront this issue as it was accused by private consultants of effectively subsidizing academics to consult, amounting to unfair competition. As a result, a policy was established called the ‘one fifth rule’, whereby MIT academics could spend a day a week using MIT resources to undertake consulting services. “Such activities as providing advice, testing materials in university laboratories and solving problems at company sites had become so much a part of the work of academic engineering professors that it proved impossible to disentangle them from the academic role. Prominent professors felt that their connection to the industrial world through consultation was essential to their research and teaching.” ((Etzkowitz, 37)) The role of academics acting as consultants is now commonplace in universities but in the US, it was at MIT where the practice was first formalised.

To protect and further exploit the industrial research undertaken, the first patent policy at MIT was developed in 1932.  Such a policy allowed MIT to assert ownership of its research, which at that time was mostly internally funded, rather than it being freely exploited by industry partners. With a patent policy in place, MIT could license its research to industry and control its intellectual property.

To support this academic-industrial partnership, MIT was one of the first universities to establish a department to handle its commercial contracts and its Division of Industrial Cooperation (DIC) later became the model by which the government provided research funding for all other universities. Being unique among universities in having such a department, MIT was in an advantageous position when the US entered WWII. Between 1940 and 1942, MIT’s research funding increased from $105,000 to $5.2M (c. x50!) thanks to its foresight in starting internally-funded military research early and having bureaucratic processes in place to handle the large increase in research contracts. Prior to WWII, there was little government funding of research outside of the land-grant interests of agriculture with around 5% of university funding coming from government and 40% of that relating to agriculture. Since 1946 federal funding to US universities has risen to between 12% and 26% of income, settling, on average, at around 15% in the 1980s; this funding is dispersed very unevenly and MIT is always one of the top recipients. ((See Lewontin in The Cold War and the University))

There is much to say (in a later post) about what gave MIT the privileged position as a centre for government research during WWII, which led to the formation of military-funded Labs such as that which Stallman worked in during the 1970s.  By 1940, through its entrepreneurial efforts, which I have just skimmed over here, MIT had become the model for the military-industrial-academic relationship which has continued to this day.

An entrepreneurial environment

In 1978, the AI Lab received additional DARPA funding for Lisp Machines and in 1979 discussions began taking place within the Lab about the commercialisation of Lisp Machines. Differences emerged between Noftsker and Greenblatt around the form that commercialisation should take and this delayed the enterprises until 1981. It would be interesting to know whether the Bayh-Dole Act did help crystallise decision-making among AI Lab management at that time, although more likely it was the knowledge that Xerox had developed a Lisp Machine in 1979 that caused Noftsker and Greenblatt to consider commercialising their work. Tom Knight, who with Greenblatt was the original designer of MIT’s Lisp Machine, has said that in the late 1970s, “MIT was not in the business of making computers for other labs, and was not about to build them” ((If It Works, It’s Not AI: A Commercial Look at Artificial Intelligence Startups)) but as external demand for them grew, Noftsker and Greenblatt developed their own responses in the form of Symbolics Inc. and Lisp Machines Inc. and Stallman’s hacker community began to crumble.

Stallman worked to keep the Lisp Machine source code free to share until 1983, when, presumably, he saw little hope of reforming the community of hackers that he longed for and could see that academic culture and the intellectual property regime had changed in ways that were no longer compatible with sharing software. MIT’s long history of entrepreneurialism and the more recent obligation to commercialise government-funded research suggests to me that the writing was on the wall for Stallman and the free sharing of source code within academia until the late 1980s when responses from within academia to this new regime were created: The GPL (GNU), BSD (University of California, Berkeley) and MIT licenses.

The ‘MIT model’, later universalised within the US as the Bayh-Dole Act, provided universities with the legal means and obligation to exploit federally-funded research. In doing so, it established not just a mechanism for patenting but an academic environment that was overall more entrepreneurial as it allowed universities to create partnerships with individual academics who would themselves profit from their research. Researchers working at MIT in the 1970s, were encouraged and well-supported to look for commercial opportunities deriving from their work. What the Bayh-Dole Act did for MIT and other universities, was provide clarification and incentives for exploiting government-funded R&D, which were previously absent. As Etzkowitz states: “In addition to rationalising and legitimsing university patenting and licensing, the law induced a phsychological change in attitudes towards technology transfer as well as an organisational change in encouraging the creation of offices for this purpose.” (p.114)

In my next post, I shall address the role of military funding in the development of hacker culture and the Labs which became playgrounds for hacking.

The role of the university in the development of hacker culture

PDP-10
A PDP-10 computer from the 1970s.

The picture above is of a PDP-10 computer similar to that found in universities during the 1970s. The PDP-10 was developed between 1968-1983 by Digital Equipment Corporation (DEC) and is a useful point of reference for looking backwards and forwards through the history of hacking.  The PDP-10 (and its predecessor the PDP-6) were two of the first ‘time-sharing‘ computers, which among other significant technological developments, increased access to computers at MIT. Hackers working in the MIT Artifical Intelligence Lab (AI Lab) wrote their own operating system for the PDP-6 and PDP-10 called ITS, the Incompatible Timesharing System, to replace the Compatible Time Sharing System (CTSS) developed by MIT’s Computation Centre. Richard Stallman, who Levy describes as “the last true hacker”, was a graduate student and then AI Lab staff system hacker who devoted his time to improving ITS and writing applications to run on the computer. Stallman describes the Lab during the 13 years he worked there as “like the Garden of Eden”, a paradise where a community of hackers shared their work.

However, this period came to a bitter end in 1981, when most of the hackers working with Stallman left to join two companies spun off from the Lab. Four left to join Lisp Machines, Inc. (LMI), led by Stallman’s mentor, the ‘hacker’s hacker’, Richard Greenblatt, while 14 of the AI Lab staff left to join Symbolics, Inc. a company led by Russell Noftsker, who was Head of the AI Lab for eight years and had hired Stallman. (Noftsker had taken over from the original Director, Marvin Minsky, who worked on MIT’s DARPA-funded Project MAC, which later became the AI Lab). For a while in 1979, Noftsker and Greenblatt discussed setting up a company together that sold Lisp Machines, but they disagreed on how to initially fund the business. Greenblatt wanted to rely on reinvesting early customer orders and retain full control over the company while Noftsker was keen to use a larger amount of venture capital, accepting that some control of the company would be given up to the investors. Greenblatt and Noftsker couldn’t agree and so set up companies independent of each other, attracting most of the ITS hackers in the AI Lab to the extent that Stallman’s beloved community collapsed. With maintenance and development of ITS decimated, administrators of the AI Lab decided to switch to TOPS-20, DEC’s proprietary operating system when a new PDP-10 was purchased in 1982. A year later, DEC ceased production of the PDP-10 which Stallman described as “the last nail in the coffin of ITS; 15 years of work went up in smoke.”

Lisp Machines went bankrupt in 1985 while Symbolics remained active until the end of the Cold War when the military’s appetite for AI technologies slowed down and subsequently the company declined. One more thing worth noting about these two AI Lab spin-offs is that within a year of doing business, Stallman and Symbolics clashed over matters of sharing code. Having been deserted by his fellow hackers, Stallman made efforts to ensure that everyone continued to benefit from Symbolics enhancements to the Lisp Machine code, regularly merging Symbolics code with MIT’s version, which Greenblatt’s company used. Stallman was acting as a middle-man between the two code bases and the split community of hackers. Like other MIT customers, Symbolics licensed the Lisp Machine code from MIT and began to insist that their changes to the source code could not be redistributed beyond MIT, thereby cutting off Greenblatt’s Lisp Machines, Inc. and other MIT customers. Stallman’s efforts to keep the old AI Lab hacker community together through the sharing of distributed code came to an end.

In an essay by Stallman, he writes about how this was a defining moment in his life from which he resolved to start the GNU Project and write his own ‘free’ operating system. In 1984, Stallman left his job at MIT so as to ensure that the university didn’t have any claim to the copyright of his work, however he remained as a guest of the AI Lab at the invitation of the Director, Patrick Winston, and still does so today. If you are at all familiar with the history of free software and the open source movement, you will know that Stallman went on to develop the General Public License in the late 1980s, which has become the most popular open source license in use today. Advocates of open education will know that the GPL was the inspiration for the development of Creative Commons licenses in 2001. Arguably, the impact of spinning off Lisp Machines and Symbolics from the AI Lab in 1981 is still being felt and the 18 hackers that left to join those divergent startups can be considered as paradigmatic for many hackers since, conscious of whether they are working on shared, open source software or proprietary software.

Everything I have described above can be easily pieced together in a few hours from existing sources. What is never discussed in the literature of hacking is the institutional, political and legal climate during the late 1970s and early 1980s, and indeed the decades prior to this period that led to that moment for Stallman in 1984. In fact, most histories of hacking begin at MIT in 1961 with the Tech Model Railroad Club and, understandably, concentrate on the personalities and development of an ethic within the hacker community. What is never mentioned is what led to Greenblatt and Noftsker deciding to leave that ‘Garden of Eden’ and establish firms. What instruments at that time encouraged and permitted these men to commercialise their work at MIT? Much of what I have written above can be unravelled several decades to show how instrumental the development of higher education in the USA during the 20th century was to the creation of a hacker culture. The commercialisation of applied research; the development of Cybernetic theory and its influence on systems thinking, information theory and Artificial Intelligence; the vast sums of government defense funding poured into institutions such as MIT since WWII; the creation of the first venture capital firm by Harvard and MIT; and most recently, the success of Y-Combinator, the seed investment firm that initially sought to fund student hackers during their summer break, are all part of the historiography of hacking and the university.

Over the next few blog posts I will attempt to critically develop this narrative in more detail, starting with a discussion of the Bayh-Dole Act, introduced in 1980.

References

I’ve linked almost exclusively to Wikipedia articles in this blog post. It’s a convenient source that allows one to quickly sketch an idea. Much needs to be done to verify that information. There are a few books worth pointing out at this introductory stage of the narrative I’m trying to develop.

The classic journalistic account of the history of hacking is by Stephen Levy (1984) Hackers. Heroes of the Computer Revolution. I found this book fascinating, but it begins in 1958 with the Tech Model Railroad Club (chapter 1) and doesn’t offer any real discussion about the institutional and political cultures of the time which allowed a ‘hacker ethic’ (chapter 2) to emerge.

Eric Raymond’s writings are also worth reading. Raymond is a programmer and as a member of the hacker tradition has made several attempts to document it, including the classic account of the Linux open source project, The Cathedral and the Bazaar, and as Editor of the Jargon File, a glossary of hacker slang. Again, Raymond’s Brief History of Hackerdom, begins in the early 1960s with the Tech Model Railroad Club and does not reflect on the events leading up to that moment in history.

Another useful and influential book on hackers and hacking is Himanen (2001) The Hacker Ethic. Himanen is a sociologist and examines the meaning of the work of hackers and their values in light of the Protestant work ethic.

Tim Jordan’s 2008 book, Hacking, is a general introduction to hacker and cracker culture and provides an insightful and useful discussion around hacking and technological determinism. Like Himanen, Tim Jordan is also a sociologist.

Stallman’s book, Free Software Free Society (2002), offers a useful direct account of his time at MIT in chapter 1.

Sam Williams’ biography of Stallman, Free as In Freedom (2002), later revised by Stallman in collaboration with Williams (2010), is essential reading. Chapter 7 ‘A Stark Moral Choice’, offers a good account of the break up of Stallman’s hacker paradise in the early 1980s.

E. Gabriella Coleman’s book, Coding Freedom. The Ethics and Aesthetics of Hacking (2012) is an anthropological study of hackers, in particular the free software hackers of the Debian Linux/GNU operating system. Coleman’s book is especially useful as she identifies hackers and hacking as a liberal critique of liberalism. This might then be usefully extended to other movements that hackers have influenced such as Creative Commons.