A Diary Study of Task Switching and Interruptions

1
A Diary Study of Task Switching and Interruptions
Toward Understanding Supporting Multitasking
Mary Czerwinski Eric Horvitz Susan
Wilhite Microsoft Research
2
Motivation and Goals

• Hypothesis Current MS software does not support
  multitasking well
• How bad/universal is the problem?
• Seek SW design ideas
• Research shows users developing workaround
  strategies
• Interruptions research shows harmful effects of
  incoming notifications on current task
• Memory for To Dos notoriously poor,
  undersupported

3
Approach

• Earlier Work
• Characterize memory for important computing
  events
• Memory augmentation systemswill they work?
• Current Study
• Capture nature of interruptions in typical users
  day
• Characterize frequency of task switch/returns
• Gather data on what constitutes a task
• Long-term Goal Design new UIs for
  semi-automated, efficient access to prior task
  contexts

4
Memory Research

• ? Retrospective Memory
• Memory for past eventsorganized into episodes
• Strong Cues people, place, key events, (time)
• Not much computer support traditionally
  (calendar)
• Prospective Memory ?
• Memory for future actions or events
• Retrieval cues often internally generated (must
  remember to remember), but people, places prompt
  too
• Relies some on retrospective memory for cuing
  event
• More computer support via reminders
• Users still use all kinds of alternative methods
  to remember stuff

5
Previous Work

• Prospective memory failures are a serious problem
  for information workers (Eldridge, Sellen
  Bekerian, 1992 Terry, 1988)
• Users come up with unique ways of remembering
  (Gwizdka, 2000 Jones, Bruce Dumais, 2001)
• Users memories for computing events erode
  quickly (Czerwinski Horvitz, 2002)
• Still little is known about the mechanisms for
  bringing intentions to mind (Sellen et al., 1992)
• Or how technology could be used to reduce
  forgetting after a task switch or interruption

6
Memory Prostheses

• Video diaries (Lamming et al., 1994)
• Useful, but require real time for review
• Personal stills (Czerwinski et al., 2002) work
  well
• Still require efficient browsing mechanism
• Hard to automatically detect key stills
• Action history explanations for programming
  error recovery (Renaud, 2000)
• Forget Me Not (Lamming and Flyn, 1994)first
  wearable
• Continuously logged users physical location,
  workstation activities, file exchange and
  printing, phone calls, email, and other users
  present
• Filter based on person, place, object, time, etc.

7
Memory Prostheses (Continued)

• Remembrance Agent (Rhodes Starner,
  1996)automatic text retrieval based on users
  current location
• Time Machine Computing (Rekimoto, 1999)access to
  desktop contents in past or future
• Cyberminder (Dey Abowd, 2000)
• Memory Glasses (Pentland et al., 2000)
• Lifestreams (Gelernter et al. 96), etc.
• These techniques seem usefulwould they help the
  average information worker?

8
Diary Study

• What are the important issues related to novel sw
  design for multitasking?
• Does it seem memory prosthesis would help?
• Wanted to characterize task switching
• Over multiple days
• In situ
• Users own descriptions of how they multitask

9
Participants

• 11 Participants recruited from a wide variety of
  work backgrounds (1 dropped)
• Professor, CAD programmer, web page designer,
  boat salesman, graphics artist, developer, stock
  broker, real estate agent, etc.
• All participants were experienced computer users,
  as per standardized screener
• All reported they were responsible for multiple
  projects at once, and proud of it
• Good mix of deadline-driven v. self-directed

10
Methods

• All users completed a baseline survey once
  recruited
• An excel spreadsheet was used as a diary
  template to be filled out each day
• Diaries emailed back to me each evening
• Participants instructed to write down every task
  switch
• how hard to switch, of docs required, of
  interrupts experienced, task time, anything
  forgotten, notes, etc.

11
Partial diary for MS (6 hours)
12
Task Descriptions for BH (6 hours)

• Daily Schedule Preparation
• Synch PocketPC
• Check Internet Email
• Check and respond to email
• Matlab coding
• Create Charts for Meeting
• Edit Word documents for meeting
• Meeting
• Matlab coding

13
Overall Diary Results

• Threw out 1 participant (little task switching)
• Diaries were coded for Task Type, Switch Type
• Codes were validated between two researchers _at_
  97 agreement
• In addition, task times, difficulty ratings, of
  interruptions and documents analyzed
• Analyzed difference between task returns and
  other task types
• Used linear regression to identify key variables

14
Task Breakdown
Indicative of Difficulty Tracking Tasks
Returned to Tasks from this group
15
Number of Documents per Task (Conservative)
16
Frequency of Interrupts (Conservative)
17
Task Priority
18
Task Times
19
Frequency of Task Shift Initiators
20
Overall Difficulty Switching Tasks
21
Difficulty Switching by Type
22
Difficulty Returning to Task by Elapsed Time
23
Task Length by Type
24
Document Requirements by Task Type
25
Interruptions by Task Type
26
Discussion of Findings

• During a given week, KWs task shift an awful lot
  (avg. 50 task shifts weekly)
• Long-term projects are more complex shifts
• Lengthier (11.25 of the week), more documents,
  interrupts, returns
• Rated significantly harder to return to
• Negative influence of interrupts on multitask
  performance and memory well known
• Passage of time also takes its toll
• What designs will help?

27
General Design Ideas from Participants

• Smarter, adjustable To Do list tracking
  alarming
• In the projects versus just in Calendar
• Consider sticky notes for partial / future tasks
• Auto-categorization of email and files
• Better reminders for things forgotten
• Track events we know about and visualize them, or
  rely on user manual tagging
• Better user adaptivity
• e.g., knowing what kinds of paste operations a
  user typically performs and automating them

28
Focus on Returned to Tasks

• Elapsed time spanned hours to days
• Maintaining desktop state isnt always the answer
• Often, users said they were waiting on info from
  other people or places (web, server)prospective
  reminders needed here
• Info came in via phone, email, web, or personal
  contacts (better app integration needed here)
• But reminding about task context and info
  assembly / layout was a key problem identified

29
Tools for Task Management

• GroupBar joins related items in the taskbar,
  remembers spatial layouts of tasks (Smith et al.,
  2003)
• Desktop snapshots
• Can rehydrate tasks with the press of a button
• Scalable Fabric and VibeLog to be presented at
  AVI 2004
• Visualizations of task activity a continued focus
• Future designs will merge all apps, across devices

30
Timelines

• Task Timeline Visualizations Should Help
  Information Workers
• Rekimotos Time Machine Computing
• Gelertners Lifestreams,
• On the right track!

31
Thanks for Your Attention

• Contact/Further Info
• marycz_at_microsoft.com
• http//research.microsoft.com/research/vibe

32
Many Assume Disruptions are Unavoidable

• Recent focus on memory augmentation systems
• Reminder systems using automatically recognized
  indices of computing events
• But need to know what key events users
  remember/forget and why
• What external cues can be useful to them (Sellen
  et al., 92) to aid retrieval?
• When is such a system useful?