Information-Centric Computing
Modern operating systems are basically resource managers whose main assets are hardware, applications and data. They provide hardware and data-access to applications and application-access to users. By themselves, they provide very little functionality in terms of information handling except for generic file operations. They can assist you copying a file from the cloud to the device and help you to open that file with an application that can handle it. In some cases, they can even show specific details about the file contents, e.g. the date a photograph was taken or the length of a video. But they cannot suggest things you could do with e.g. a shopping list. Nor do they guide you through a work process, e.g. booking a hotel room using data that you have available from managing a project. The problem of course is, operating system generally do not understand the data they handle, for them it is just files, they know very little about the contents.
Most applications on the other hand are dedicated to one task that they can perform on a limited set of data types. Sharing data between different applications is often limited.
The usual process when doing something with computers is (1) open an application, (2) load the data/information and (3) process that data in some way. This process is application-centric and so are our operating systems. For many domains, this is a good design. However, for many others it is not very efficient, it is not intuitive; it is superficial and is far from being cognitive.
Information-centric computing (ICC) has a different approach from our current, application-centric information systems: Information is at the center, the system understands the information it is dealing with and suggests tasks that can be performed depending on information type and the context in which the information is handled; a much more cognitive approach. Context is defined through different mechanisms, for example through user configuration, device and location. For ICC to be universal, some form of standardized protocol or data format capable of expressing (or at least describing) arbitrary information is required. Havel is such a format.
The ICC approach is distinctly different from e.g. using a file manager to select files and then an app to manipulate them. In ICC, the underlying system understands the information it is dealing with much deeper than operating systems or file managers understand the contents of files. In addition, the base for operations is not monolithic files but pure, relational information, allowing changing focus seamlessly in any direction or even dimension.
A user can connect to a universal repository using an IME or NME. Because there is no underlying order in a universal repository, the selection of data to work with can be erratic. A user can create ad-hoc order by dynamically structuring a set of information or simply search and select arbitrary expressions. He can also create dynamic expressions from parametrized mergers of static data. This process is technically comparable with executing a query against a database; practically it is much easier, most users will use tools and intuitive UIs instead of having to write queries.
Unlike a traditional database system, an IME/NME understands the data that has been selected; it therefore can suggest what to do with it. Not just generic tasks but also tasks that are specific for the selected information in combination with the current user/device context. The context can be user settings, the user’s current role, the task the user is currently working on, hardware, geographical location and so on.
Tools and functionality can be freely added to the system or created by the users themselves. When searching for suitable functionality for a specific situation, the system can use the user’s entire infoverse, which can be rather extended. If, for example, his infoverse includes a sharing platform or marketplace the system can automatically suggest functionality even if the user has never before executed the task he is currently working on before. It can even suggest a workflow including explanations on how to execute that specific task. This approach eliminates the border between local device and the cloud in terms of available functionality. Any functionality that is available in a user’s infoverse is available to the user at any time. Even if the user is not aware, that such functionality exists.
An ICC can guide a user through work she has never done before, basically teaching users how to do things. That is not limited to suggesting available functionality but can also include guidance. A user can select a task and let the system guide him through the whole process from ground up. Furthermore, this technology is not limited to classic computer applications. Any kind of real-world or academic process or workflow can be abstracted and modelled. An ICC can be a production tool as well as an educational tool.
This post is part of a comprehensive introduction into SRDM, graph languages and Havel.
Continue reading: Havel And Artificial Intelligence
Comments