Overview
From Meta Desktop
Contents |
[edit] Enhanced Desktop Usability
[edit] Movements towards enhanced usability
The guiding principle of the Meta Desktop project is simple: It is to provide an improved and enhanced desktop user experience. To achieve this, there are three core elements of the current way of doing things that might be changed. They are adding the ability to more efficiently and pleasurably access one's information; providing the means of doing so in an updated set of GUI, language-controlled or command line tools; and improving the computer's own ability to intelligently sift and present information, both to human users and other computers.
The Meta Desktop project firstly proposes a fundamental shift in approach to organising the file system of a desktop computer. It does this without throwing away the current method and with respect to the desktop shifting from a local machine towards network interoperability. Secondly, the project looks in detail at what applications are required to realise this shift in thinking about data organisation. Further, it proposes a re-think about desktop UI presentation (or at least pokes holes in the current interfaces). It also proposes a system whereby a user can linguistically interrogate the system. Finally it makes murmurs about enhancing the machine readability of data in order to better bridge the semantic gap. It should be noted that this third goal is subordinate to the enhancement of human interaction, though some might argue that it is fundamentally connected to the ability to realise that goal.
It should be noted that there is limited metadata support in both Mac OS X (Tiger) and Windows Vista and some support for saved searches in both. OS X even goes so far as to add what they call 'smart folders' based on Spotlight searching. Support is very much secondary and ad-hoc in both systems and there is currently no integration into the core file saving or browsing workflow.
[edit] How would all this work?
The system would achieve these goals by providing the means to write, edit, suggest and interrogate enhanced metadata associated with the system's files. Both the system and user would be able to write appropriate metadata to files. The tools to achieve this should be part of a highly usable interface.
[edit] Core principles
This diagram describes the key project principles. A smart filesystem, working by both enhanced and traditional methods, plus an enhanced UI incorporating tools to manipulate files in the enhanced system join to bring about an enhanced user experience.
fig. 1 Image:Core principle brief.png
[edit] A further principle - Machine readability
We can further add one more element to the core project: Machine readability.
Looking towards greater network interoperability and more intelligent machine-layer presentation of data, the system, through its use of enhanced metadata, should attempt to bridge the human-machine semantic gap. This would offer a user benefit through the more intelligent and accurate presentation of information.
A quick look at the principles of the Semantic Web should offer an explanation about what could be seen as bringing those principles to the computer desktop.
On our second diagram shows, 1, the core principles of fig. 1 and shows, 2, the smart filesystem and UI sitting on top of machine readability.
fig. 2 Image:Core principle extended.png
[edit] A calm, flat pool of data
Go to: Main discussion
The data on any system should be presentable in any way the system's creator designs. Until now we have been used to a strictly hierarchical system of directories with each file stored in a single location. While it might sound counterintuitive, there is no reason why this data can't be presented in multiple places at once where it it contextually useful. For example, it might be just as meaningful to store the letter I wrote to my boss asking for a raise in my 'work' directory as well as my 'accounts' directory.
While most filesystems organise files in a hierarchical manner, there is no real reason why this presentation need be applied to the user's view. Short of writing a new filesystem, Meta Desktop would be formed from a suite of applications sitting on-top of the existing structure, in userland, presenting a new view of the data. Users could then choose to see files as a pool of data which is essentially 'shelfless' or use the traditional system.
In essence, this model thinks of a pool of data and various ways of ordering files and presenting the relationship between them.
[edit] Faceted and folksonomic
This flat way of presenting data requires meaningful metadata to be added to files by the system and the user. This system would add initial metadata to a file based on context or task. For example all downloaded files might be labelled, 'download' all photos imported form a camera, 'photo'; all word processing documents 'words' etc. the system could even analyse a file and add metadata accordingly. The standard set of user, time, permissions etc would be appended to a file. With enough thought a useful set of machine-created metadata would set the system up to be useful. This would create an a usable, filterable faceted system with no user interaction.
By adding user-entered metadata in the form of tags or labels the system becomes more personalised and and the information capable of advanced filtering. The majority of real-world examples are online and current principles on the web are well known. Essentially adding user metadata to our system would give us a folksonomic system.
Together the principle of machine and user written file metadata will create a highly usable and relevant desktop environment.
[edit] Tempering the pool
One might think that such a pool of tagged data could run the risk of becoming chaotic. However the system would provide the tools to find and store files and a default 'structure'.
The core tools would be:
- A metadata-aware file browser incorporating a filtering application
- Metadata aware save and open dialogs
- User-definable and system default 'smart' directories
The applications would present and seek information intelligently and by adding default locations for certain core 'smart' directories, like 'pictures' and 'words', people should be able to find things without much prior knowledge of the system and without relying wholly on search.
[edit] Using language
Meta Desktop also proposes a system whereby a user can linguistically interrogate the system, at least on the file-browsing/access level. By this, it is meant that certain, simple instructions could be interpreted by the application and return appropriate results. Such a system would be 'forgiving' rather than require a set of commands or controls be learnt.
Similar systems have been written for some online applications. Stikkit, for example, interprets a form of simplified natural language to mark-up reminders and notes. It can take a sentence like "Meet Marie at The Effra, 8pm next Wednesday", parse it and return a note linked to a calendar reminder next Wednesday, link it to my contact Marie and create a new contact for "The Effra", my local pub. Stikkit has a complimentary and interesting new sister project called 'Sandy', which uses natural language emailed to a specific address to create event reminders and task lists. Sandy presents herself not as an email driven text-parser, but as an intelligent, personal PA.
Meta Desktop would include simple tools within the system (in particular the file browser) to enable a simple-language query like "All files worked on the week before last with the tag 'meta desktop'", to return a meaningful set of files. This free-language input would work as an alternative method of interrogating and filtering the metadata at the heart of Meta Desktop.
An interesting discussion about such interface devices forms part of Don Gentner and Jakob Nielsen's well-known 1995 paper, The Anti-Mac Interface.
Next: The System
