The Sociology of Information Infrastructure Geoffrey C. Bowker Informatics
Introduction
Designing Cyberinfrastructure
Understanding Infrastructure Report In this report, we tried to find the common factors which affect all kinds of infrastructure development. We drew from the history of electricity and phone infrastructures among others to draw lessons for the understanding of the development of information infrastructures in use in both science and business.
In this report, we advised the National Science Foundation and the Sloan Foundation about new ways of creating knowledge (not just information) through infrastructural design. http://knowledgeinfrastructures.org/
Ways of Designing Infrastructure Top-Down Middle-Out Historically there have been three ways of building infrastructures. These are not mutually exclusive, but each type has its own advantages and disadvantages. We shall look briefly at the first two in turn. In the interests of time I will not discuss ‘bottom-up’ infrastructures (also known as ‘inverse infrastructures’) but am happy to take questions on these.
All infrastructures are ‘sociotechnical’ – they all involve at the same time configurations of technology and of organizations working together to a common end. There can be no infrastructure which cannot adapt to local social and political considerations (infrastructures are never purely technical) and that no infrastructure can be designed purely by either engineers on the one hand or administrators on the other.
Top-Down Design Example: the first electricity networks This is also the mode used most by local and national governments to this day The first electricity networks were built by system builders such as Thomas Alva Edison. Edison built his system AC system in competition with Tesla’s DC system. Only one could win –the two engaged in a huge struggle. Top-down systems have the advantage of promoting a single vision –they are internally consistent, and can work very efficiently as far as they go. Every electronic device needed a single standard plug to use the system.
Lessons from Top-Down Designs Understanding reverse salients is core to understanding infrastructure development. In this image, we see a variety of possible reasons for a scientific information infrastructure not to work. There might be technical issues, such as dealing with different data formats or standardizing metadata so that databases can work together. There might also be social issues such as disciplinary or organizational culture. Or economic issues such as incentives or sustainability over the long term. At each point, the manager must pay attention to the possibility of ‘reverse’ salient occurring in a number of different ways.
Problems with Top-Down Designs They are hard to adapt to other infrastructures/networks – and it is often unclear who has the responsibility to make them work. This can heavily disadvantage peripheral groups, who must adapt to the standards and practices at the center. Often, standards developed at hi-tech centers do not work well in other areas. They can be very hard to maintain, since they frequently lack adaptability to new social and political settings. To use the same device in multiple power networks as we travel the world, plugs and sockets must be adapted to each other. This is an instance of the gateway problem. Ordinarily we solve it with a plug adapter, i.e. a special-purpose gateway that connects one kind of plug with a different kind of socket. Other solutions could exist, however, and all of them are simultaneously technical, political and economic. Other adaptations come with other locations of design, cost, and responsibility. In this example, adaptation could also occur on the device plug (a plug with multiple pin configurations to fit multiple sockets). This would shift design, cost, and responsibility to the device designer. Or it could happen on the socket (a reconfigurable socket to fit multiple plugs). This solution would shift cost and responsibility to builders, building codes, socket manufacturers, and potentially to electric power providers
Middle-Out Design Example: the Internet This is also a more flexible mode which can work over the long term The Internet is best understood as a ‘network of networks’. When it was first constructed messages went over phone lines. Now they can be sent by wireless, cable, satellite, DSL as well. There is not central designer – the core design choice is to make one’s infrastructure flexible and adaptable to new conditions.
EarthCube Building Blocks
Lessons from Middle-Out Designs 1 Designing a sustainable infrastructure: You cannot declare that you will be fully sustainable You need to continually forge new alliances
Lessons from Middle-Out Designs 2 When planning for infrastructural design, we often assume that the present form of the infrastructure will persist forever. This is the central line in this diagram. This leads to the issue of ‘path dependency’ – because we have done things in this way in the past, we assume that this is the way that it will operate in the future. There is a tight coupling in this case of software, hardware, people and practices and technology. However, this can lead to a blindness to discontinuities (for example, disruptive technologies) which can change the game. Libraries have over the past 40 years been struggling to adapt to the prevalence of electronic storage of information – they find it hard to think beyond the model of book storage. Infrastructure design must be sensitive to discontinuity.
Which came first?
Conclusion