Heuristic Analysis of Improvement 1

The changes made in Improvement 1 are beneficial to the overall task model in numerous ways. The following section lists improvements that have been made in to the eight interface rules:

  1. Consistency - The first improvement to consistency was made by constraining nearly all action to the same window. Using monotony in this sense, not continuously switching between domains (windows) and regearing for a new set of interaction rules, improves the user's mental model. It has also been benefited by giving access to the action (rule-making) directly from the e-mail message window.
    Impact assessment: This improves time-on-task by decreasing the amount of thinking that the user needs to do in figuring out what action to take for all but the most advanced users (those who have made so many rules that each step is like second nature).
  2. Universal usability - By cutting the number of steps, it would make basic rule creation easier for any number of users: beginners and experts, people with motor differences (who would have difficulty with the number of clicks the old method takes) and the able-bodied, people with superb memories and the forgetful.
    Impact assessment: Again improves efficiency and also makes the interface less frustrating to use.
  3. Informative feedback - The main improvements that have been made in regards to feedback are that:
    a) The choices that have been made by the user are displayed in the same form in which they were entered - currently, the rule criteria are entered in one window in one form (a text box) and then displayed in another (hyper linked text)
    b) A verification window is displayed as the last step that shows the new rule in natural language form.
    Impact assessment: Both of these improvements give the user the confidence to know that she is doing the right thing and that the outcome will be as desired, which can decrease hesitation, thus improving efficiency and experience.
  4. Design dialogs to yield closure - See Informative feedback.
  5. Error Prevention - Error messages are still returned if all of the required information isn't filled in correctly. Also, when user begins task, for instance by clicking on subject line, the rule's target has already been chosen and the condition field filled out (but still editable)
    Impact assessment: Saves a lot of time by automatically completing parts of task.
  6. Reversal of errors - It is still possible to edit any part of a rule in progress and, as mentioned, the verification step should prevent faulty rules from being created. Rule deletion, however has not been addressed in this version.
    Impact assessment: For benefits, see Feedback above. However, having a main list of rules in a different location does not help if one wants to delete an old rule. This is not part of the rule creation task but could be disruptive in the Message Rules domain.
  7. Locus of control - As shown in previous examples, Improvement 1 attempts to remove any obfuscation of the goal from the creation process. By limiting the number of actions that the task is made up of, it is hoped that the goal and means to reach it will remain clear, thus keeping control in the hands of the user.
    Impact assessment: Removing multiple windows is a major improvement because it wrests most control away from the computer. The task window can now be repositioned to reveal the message if model consultation is necessary. However, the negative side to automatic rule-target generation is that if it were chosen by mistake, there is no obvious way to change it.
  8. Reliance on Memory - As stated in Locus of Control, by consolidating the rule-creation steps (thus removing most of the additional windows) and by making the task's starting point be the rule target itself (thus allowing for automatic generation of rule criteria (target and condition)), a large burden is taken off of the users thought-process.
    Impact assessment: Moving some memory requirements and actions off of the user Improves efficiency and ease of task greatly.

While this new version does make improvements on many parts of the task design, some of these changes create other concerns. The most noticeable flaws include:

With all of these judgments made after predicting the positive and negative effects that Improvement 1 can have, a second version was created to address some of the faults of the first. next>>





Current version

screen   0a - 0b - 1 - 2 - 3 - 4a - 4b - 5 - 6 - 7a - 7b - GOMS - heuristics

Improvement 1

screen   1a - 1b - 2a - 2b - 2c - 3a - GOMS - heuristics

Improvement 2

screen   1a - 1b - 2a - 2b - 2c - GOMS - heuristics