The prehistory of open source: episodes in collective invention

I found this April 2003 paper, which details historical examples of shared design in the industrial era:

* Essay: Episodes of Collective Invention. Peter Meyer. U.S. Bureau of Labor Statistics Office of Productivity and Technology, 4 August 2003

The author seems to suggest that in the past, such open periods were temporary, and of course they were, but this could lead to the erroneous conclusion that the same is true today. We would rather argue today that it is the collective invention that is the norm, which subsumes proprietary development as just a moment in its development.

The Examples:

Peter Meyer:

Blast furnaces in Britain’s Cleveland district

Allen (1983) found that from the 1850s through the 1870s iron-making companies in northeast England’s Cleveland district allowed visitors and consultants to see the insides of plants and to write about the way their blast furnaces made usable iron from ore. Wellknown researchers at the time such as Isaac Lowthian Bell, Thomas Whitwell, and J.G. Beckton published information about the designs, size, temperature, and contents of blast furnaces. The information came from observing production, not from formal research efforts. Publications and well-informed consultants helped establish which blast furnace designs used fuel most efficiently. Plant designs evolved to have taller furnace stacks (filled with the input materials) and toward hotter and hotter furnaces. The design changes were not generally patentable because of the nature of the technology. Since plant design was a natural area of competition one might have thought the owners of blast furnaces would each prefer to keep this information secret. But Allen concluded that through the sharing process firms could reasonably expect to learn more valuable information than they gave up, and therefore each firm preferred such information be made public over taking the risk of shutting it down by withdrawing. The collective invention regime substituted for research and development spending. Furnace efficiency improved over time. Allen found that little of this improvement was caused by private research and development efforts. He attributed most of the productivity improvement in blast furnace practice over time to the sharing of information which he called collective invention.

Steam engines discussed in Lean’s Engine Reporter (1811-1904)

Steam engines had been around since 1712 but in 1769 James Watt patented a new, much more efficient design for them. Despite legal attacks on his patent, it was upheld until 1800. Mine owners in the Cornwall region of England used steam engines to pump water out of mines, sometimes using illegal copies of Watt’s design. Mine owners and steam engine makers resented Watt’s unwillingness to license the invention cheaply. After Watt’s patent expired they could legally make modifications to the design. There was an explicit debate on alternative forms of intellectual property rights among the steam engine engineers. Few actually filed patents. Rather, there was a collective invention environment, as shown by Nuvolari (2002).

Starting in 1811 there was a publication read by the Cornish steam engine makers, called Lean’s Engine Reporter for its editor, Joel Lean. Its contents were technical comparisons of operating steam engines. Nuvolari (2001) establishes that the efficiency of steam engines improved substantially in Cornwall through this period, probably through many minor or unattributed innovations and discoveries by the steam engine engineers. Collective invention sustained by the Reporter thus supported useful engineering improvements.

Bessemer steel in the U.S. (1866-1885 and beyond)

British inventor Henry Bessemer announced a new steel making process in 1856. He correctly foresaw it would be a quick, high volume, fuel efficient approach and trod lightly on the fact that it hadn’t actually worked yet. After some further innovations with others, it developed into a thriving business in the 1860s. It took longer to transplant the technology to the U.S., which took another wave of innovations.

Several institutions served to aid collective invention in the U.S. technology. First, several new industrial journals and organizations arose during this period when mass production steel technologies were being adopted. The open-discussion environment in the British iron and steel industry may have been a model for its extension in the U.S., leading to collective invention on both sides of the Atlantic. Second, famous engineer Alexander Holley ran a consulting practice which pooled the key patents for Bessemer steel manufacture so members of one licensing organization would have access to them all. Holley himself designed most of the first fifteen Bessemer steel plants though they had different owners. Third, job turnover was high in the industry, so many employees had diverse experiences drawn from previous employers.

Huge demand for steel rails for railroad construction sustained the industry through a depression that started in 1873. Production quantities rose dramatically, as shown in Figure 1. The technologies in use improved quickly, and the price of Bessemer steel fell from over $100 per gross ton of rails in 1870 to about $60 in 1880.”

Discussion by Peter Meyer:

“As societies have adapted to waves of new technology there have been phases where technological information was openly and enthusiastically shared. We will look here at certain cases where technological developments were published openly to some community, and we consider why it happened and what institutions made it work.

Robert C. Allen (1983, p. 2) used the term collective invention to describe “the free exchange of information about new techniques and plant designs among firms in an industry.” Allen saw that this had happened among iron producers in Britain’s Cleveland district. Richard Nelson (1982, p. 468) encouraged research into the subject.

Schrader (1991) and von Hippel (1987) documented explicit, informal “know-how trading” among mini-mill steel makers in the U.S. in the 1980s. Nuvolari (2002) showed that steam engine engineers in Cornwall maintained a collective invention approach. Harhoff, Henkel, and von Hippel (2002) collected several recent examples of “freely revealed” innovations in recent years and modeled a game between innovators in which revealing could be an optimal choice by an innovator because it helps diffuse the technology. A number of investigators have described open-source software projects as examples of collective invention; relevant papers are at the web site

One might take the view that a technology is truly invented only once, and subsequent applications of it are innovations which are part of a separate process of diffusion. If so, collective invention is not the right term – collective innovation or user innovation would be better. But in the cases we will consider, the first versions of the technology are not applied for long. The invention as generally perceived afterward includes the many improvements from the early discussion. In these cases the form of the technology that turns out to matter is the one slowly invented over the course of years, so for this paper’s purposes we will say it was collectively invented.

There are several differences between collective invention environments and those depending on legal enforcement through patents. A patent system tolerates bogus submissions or unclearly described inventions, since an inventor benefits from users who pay for a license or are otherwise dependent. On the other hand, in a collective information environment, the innovator does not have an incentive to hide essential details. Patenting environments encourage the inventor to invest in and develop inventions, whereas collective information environments do not give an inventor a direct financial incentive to develop it further. Either environment could give innovators the satisfaction of seeing their innovations widely used. Which one will generate more technological advance depends on context. Below I argue that when the technology’s future is uncertain, collective invention does better at advancing the technology.

As we shall see, restricting the technical community to for-profit firms makes it hard to use the term in other episodes.

Therefore let us enlarge the definition for the purposes of historical comparison:

• Include hobbyists and workers in not-for-profit organizations such as universities and government agencies in the community of technologists.

• Include job changes by people working with the technology as one of the mechanisms of interchange of technical information.

• Include cases in which collective invention practices co-exist with some firms who do secret research and development.

Even if some firms keep their findings secret, collective invention may still occur in some subset of an industry. So collective invention is defined here to be a process in which improvements or experimental findings about a production process or tool are regularly shared. Put this way, collective invention seems to be an important and regular feature of the historical process by which societies adapt to radically new technologies. It is part of a larger picture in which the new technology turns into new products and the producers, consumers, and markets are jointly developed.”

1 Comment The prehistory of open source: episodes in collective invention

  1. AvatarFrances Bell

    Thanks for this interesting post. Ironically the last Cleveland Steel plant is currently being mothballed.

Leave A Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.