Brains working towards machine consciousness

The 'soft' side of computer related research relates to people. There is not only a great diversity amongst people in general, but also many subtleties relating to the specific context in which they relate to technologies. Any computer system becomes part of a socio-technical system when in use, and this is most evident with tools which facilitate collaborative or social interaction. Trist (The evolution of socio-technical systems, 1981) identifies three levels at which socio-technical systems have to be investigated:

Primary work systems - these relate to systems which work in a bounded environment, for instance the people and tools involved in producing a particular academic paper or proposal. These have a specific purpose, which provides a focus for the people and activities.

Whole organization systems - these can be self-contained workplaces, such as a workgroup which deals with end to end production of a resource, or entire businesses or organisations. Trist describes them as persisting "by maintaining a steady state within their environment"

Macrosocial systems - these include systems in communities, and which operate at a societal level. Trist calls these 'domains'.

None of these are constant through time, and each has to be seen in the context of the other layers. Where knowledge and ideas are a constituent part of the work a system does, there is an additional dynamic concern. The components of a system use tools to achieve their goals. These components can be humans or other agents. Traditionally within the computer enabled knowledge industry, tools have been considered to comprise two main parts: hardware, and software. This is an abstract classification, with firmware being software which is semi-permanently linked to the hardware, and there are further abstractions of layers to provide, for instance, drivers for hardware and tool kits such as the DirectX graphics, sound and other human-computer interface elements. Whilst we are used to using a tool to manipulate something, such as data to help us turn it into information, we are not, in general, used to considering tools which use elements of their environment to provide us with a service. Generally, we would consider these to be services in their own right. However, with the increasing complexity and, in some cases, autonomy of systems in the knowledge industries, it behoves us to examine our use of the term 'tool' in connection with systems such as del.icio.us and Facebook.

Web 2.0 is a poorly defined term, but let us define it here as being those web based systems which allow the user to 'prosume' content. That is, they can access information, modify or aggregate it, and re-publish material for others to use. This is the same process as is involved in creating knowledge - building from experience to create and disseminate something new, whether genuinely novel or 'just' reflecting the ideas in the light of the individual's experience. del.icio.us and Facebook are examples of this type of technology.

Web 2.0 systems provide us with tools. We can use them to process information, and to produce new outputs. However, there is a question to be answered about what constitutes the 'tool'. Without the content in del.icio.us, for example, the system provides us with few affordances. Essentially, it just allows us to bookmark web content and apply tags to those bookmarks, keeping the resulting set on a web site so that we can use it on multiple machines. This is a tool in its own right, but if we consider the system as it operates today, it is much more powerful than that. Using the del.icio.us system as a whole, including the data contributed by others in the form of a folksonomy (Vander Wal), we can assess how popular a web page is, see how other people categorise the content of it, and follow links between web pages based on evaluations made by the wisdom of the crowd (Surowiecki). This whole system now becomes the tool. With the ability to 'follow' other contributors, the system also has primitive social networking facilities, which can serve to create sub-systems with tight feedback. Clipmarks, a system which allows for similar identification and tagging of material, but which also allows for a comment thread on each post, takes this a stage further. Once the system as a whole, including the data, is seen as a tool, the underlying technology involved becomes 'demoted' to the status of infrastructure - either that, or the system has to be regarded as a meta-tool (an equally valid view, but one more likely to produce quizical looks in conversation).

Any computer system which permits multiple users, and promotes comunication between them, is therefore a socio-technical system, and needs to be studied (if it is to be studied, that is) with regard to the whole system. Elements of this system include the people who use it and those who design and implement it. Each of these has their own experience and motivations, and whilst we might like to take a purely engineering approach, humans are not particularly susceptible to this kind of analysis. Therefore it is necessary to adopt techniques which have been developed, tried, and tested in the social sciences and in psychology in order to be able to conduct a reasonable analysis of the system.

Activity Theory (expanded by Engeström) provides a framework for analysis of systems which include the elements of community and tools. The tools is seen as being modified by the environment in which it is used, and this makes the theory a useful starting point for analysis of socio-technical systems.

There are certain practicalities to researching the human element of a system. Typically in computer science, as it is an engineering subject area, there is a tendency to seek metrics which can be used to measure a system. However, there is a lot of variation involved when people are included and the treatment of results needs to take into account the potentially widely varying contexts which apply to the users of a system. These contexts include the entire life experience of the individual, which makes it hard to produce a carefully controlled experiment. For this reason, if for no other, it is important to consider the use of qualitative methods to capture useful information from the users about their experiences, or their needs, rather than purely quantative ones.

Grounded Theory provides a method for extracting theory from data. It rejects the traditional scientific view that the study of a system should be rooted in the literature, and allows for the theory to be produced as a result of experimentation and observation of the system itself. Data can be gathered through interview, with the first stages being an open sampling to produce enough data to start coding the responses. This coding involves reading through the data and drawing out themes which emerge. This can take several iterations, as themes may become apparent as a result of identifying others later in the first pass. The themes which have been identified can then be categorized into broader concepts - a step towards the abstract. These concept-level categories are then analysed and theoretical memos written to describe how they have been generated, and how they link with one another. This stage can also involve breaking a category down into multiple parts. This stage is a synthesis of new knowledge, and it, too, can be iterative, allowing for the creation of further concepts and for additional layers of abstraction, if necessary. The final stage is to identify the key concept which forms the backbone of the theory being generated, known as the core category. This is also a synthesis of knowledge, and as such will be influenced by the experience and motivations of the researcher.

Other methods for researching these systems include Discourse Analysis, Thematic Analysis, Qualitative Content Analysis (which can be more, or less, rigorous), Ethnography, Action Research, Interpretative Phenomenological Analysis and various forms of Narrative Inquiry. These will be examined in greater detail in later posts.

As a side note, the techniques used in Qualitative research seem to form a good basis for requirements analysis and evaluation of systems.