Future Directions: Maintainability
We have a long way to go before these ideas form a workable benchmark
- completing a standard task taxonomy
- automating and simplifying measurements of task cost
- built-in hooks for system-wide fault injection and user response monitoring
- can we eventually get the human out of the loop?
- developing site profiling techniques to get task freqs
- developing useful cost functions
Better human studies technology needed
- collaborate with UI or social science groups
- larger-scale experiments for statistical significance
- collaborate with sysadmin training schools?