I'm at the What's up Doc Research Blogging training event today and very much enjoying it - http://www1.vitae.ac.uk/researchers/346441/Whats-up-doc-blog-for-postgraduate-researchers.html.
I wrote this post on the train from Bristol to Birmingham, I've just been working on my PhD changes and on an NVQ 2 in business admin. I often undertake vocational training to keep my research focused. The combination of the PhD changes, Communications Unit I just finished, and the What's up Doc? event hashelped me generate this post.
Communication is the key to successful computing, so this involves person to person, person to computer, and computer to person communication.
For some tasks the person does not need to know how the computer is doing the tasks. Feedback is still very important. For example when applying formatting to a document a person does not need to know how the computer processes this, only how it's applied. Just like in person to person communication what is needed is trust. The software provides feedback illustrating what has been done. The person then trusts that what he or she has intended has been done. A failure of trust may occur when formatting on screen and in print preview looks a certain way and the printout looks different. That loss of trust affects the quality of user interaction for the task, the application such as the word processor, and the persons' trust in computers and computer applications.
For modelling and calculation, the user seeing the results is not enough to establish trust. When taught maths at school we were taught 'show your working'. We could get follow through marks even if our result was wrong, because exactly where the mistake happened could be seen. The same is true for computer systems, they must show the whole calculation chain. The more complex the calculation, then the more effort is needed to ensure good representation, visualisation, and interaction with the calculation including the 'workings'. This must be done in an audit trail that is developed with the users in mind. Then as users have different needs it should be possible to choose different visualisations. More advanced users may want to create and edit their own model and visualisation. This is where the end-user programming/modelling comes in.
The user needs to be able to interact with the way models/calculations are performed to create and edit them, and choose how they display. The more this can be achieved via diagrammatic programming skills such as drag and drop, then the wider the range of people that can model/program this way.
For written information also, trust is required, this is often called 'provenance' in science for example, who is the source? and can they be trusted. Wikipedia documents show an audit trail of changes. A more structured and visualised view of sources and provenance helps a user track these sources and establishes trust in the document writers (if the trust is deserved), the document, and the document management system. Semantic Web based systems based on RDF (Resource Description Framework) can track these relationships of provenance/trust. Visualising and enabling of user interaction with this would enable trust built on greater user interaction.
Users aren't stupid, so developers should always enable the maximum interaction that is necessary and useful, then trust the user to get it right, but leave them a way of backing out when they make a mistake.
No comments:
Post a Comment