5. Direct Manipulation ("point and select")
  - visual representation of domain objects
  - direct action on objects
  - immediate and visible effects
  - reversible actions

Discount usability engineering
  - a quick-and-dirty approach
  - basic premise: designs change substantially in early development phases
  - make a prototype that shows UI to get feedback early on
  - perfection is not cost effective
  - test early and often
  - how to do
    1. scenarios
      - task analysis
      - prototyping - partial
        - use paper & pencil or simple tools
    2. simplified thinking out loud
      - use 3-6 real users
      - ask user to think out loud while performing given tasks
      - collect data by note-taking

  User Analysis
    - learn by recognition, not recall
    - remember things in related groups
    - have different ways of learning and communicating
    - like to be in control

    pre-experience, such as KLM (Keyboard Level Model)
      provides numerical prediction of user performance
      primitive operations
        - physical actions
        - mental preparation
        K = 0.35 sec (hit key on keyboard, press button on mouse)
        P = 1.1 sec  (point - move mouse to position)
        H = 0.4 sec  (homing - move hand b/w kbd & mouse)
        M = 1.35     (mental preparation time)
        R =          (responding - time for computer to respond)
      rules for using the M operator
        1. point and click (MPMK) => MPK (it is one cognitive unit)
        2. menu selection (MPK [File] MPK [Save]) => MPK PK (File on way 2 goal)
        3. typing (MK 'n' MK 'o' MK 't') => MKKK
        4. typing a terminator (MK 'a' K '.' K 'enter') => MKKK (bring closure)
      what KLM doesn't do
        1. errors
        2. learning time
        3. recall

Usability:
the extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction  in a specified context of use

Visibility:
Making the user aware of the system’s components and processes, including all possible functionality and feedback from user actions.
Reduce memory load
Recognition/Recall
the principle of visibility is based on the fact that we are better at recognition than we are at recall

User testing
Users were videotaped and timed when performing increasingly complex tasks
Captured data including time taken, errors, help accessed, and task steps missed

Heuristic evaluation
Identify an interface error by predicting user problems it will cause
Good at finding poor terminology and lack of clarity
Ten heuristics are inadequate as a guide
A subjective process

End user testing
Identify the symptom and infer its cause
Good at finding problems while performing real tasks
Task-based 
May miss features not encountered in tasks 
Users tend to blame themselves rather than the interface

Both techniques share the same goals, but the actual results are quite different 
End user testing indicates the symptom of a problem
Heuristic evaluation identifies its cause
Heuristic evaluation helps analyze observed problems. But observation of novices is still vital as many problems are a consequence of the user’s knowledge, or lack of it.