blue / papers.hotmobile2015-numerator · Commits

Commit f672f896d6a3bf6dc4e153e1355b57fcbc15a10e

Authored by Geoffrey Challen
1 parent c46c0622

Trimming.

Inline Side-by-side
Showing 5 changed files with 25 additions and 32 deletions
conclusion.tex
  View file @f672f89
... ... @@ -21,4 +21,4 @@ awards
21 21 \href{http://www.nsf.gov/awardsearch/showAward.do?AwardNumber=1205656}{1205656}
22 22 and
23 23 \href{http://www.nsf.gov/awardsearch/showAward.do?AwardNumber=1423215}{1423215}.
24   -
  24 +The authors thank the anonymous reviewers for their feedback.
... ...
metric.tex
  View file @f672f89
... ... @@ -17,10 +17,10 @@ user is paying attention to. Number of starts is also a potentially-useful
17 17 input, as may be the distribution of interaction times across all times that
18 18 the app was brought to the foreground.
19 19  
20   -While these measures of contact time are intuitive, there are obvious cases
  20 +\sloppypar{While these measures of contact time are intuitive, there are obvious cases
21 21 in which they fail, particularly for apps that spend a great deal of time
22 22 running in the background in order to deliver a small amount of useful
23   -foreground information---such as a pedometer app.
  23 +foreground information---such as a pedometer app.}
24 24  
25 25 \subsection{User Interface Statistics}
26 26  
... ... @@ -28,23 +28,23 @@ Patterns of interaction may also be useful to observe, and inputs such as
28 28 keystrokes and touchscreen events are simple to track. However, there is more
29 29 obvious differentiation between app interaction patterns between
30 30 categories---users deliver far more keystrokes to a chat client than to a
31   -video player---so it is clear that interaction statistics will have to be
32   -used in conjunction with complementary value measure components that offset
33   -the differences between high-interaction and low-interaction apps. This
34   -approach also fails in the case where apps deploy confusing or unnecessary
35   -interfaces that require a great deal of unnecessary interaction to accomplish
36   -simple tasks. Clearly such apps should not be rewarded.
  31 +video player---so interaction statistics will have to be used in conjunction
  32 +with complementary value measure components that offset the differences
  33 +between high-interaction and low-interaction apps. This approach also fails
  34 +in the case where apps deploy confusing or unnecessary interfaces that
  35 +require a great deal of unnecessary interaction to accomplish simple tasks.
  36 +Clearly such apps should not be rewarded.
37 37  
38 38 \subsection{Notification Click-Through Rates}
39 39  
40 40 Another interesting statistic that could provide insight on app value is how
41 41 often users view or click through app notifications. When notifications are
42 42 delivered but not viewed, then it is unclear whether the app needed to
43   -deliver them at all. When clickable notifications---such as those for new
  43 +deliver them. When clickable notifications---such as those for new
44 44 email---provide a way for users to immediately launch the app, the percentage
45   -of notifications that are actually clicked as opposed to ignored could be
46   -used to at least evaluate how effective the notifications are, and may also
47   -reflect on overall app value.
  45 +of notifications that are clicked versus ignored could be used to at least
  46 +evaluate how effective the notifications are, and may also reflect on overall
  47 +app value.
48 48  
49 49 Notification view and click-through rates also help put into context the
50 50 energy used by apps when they are running in the background. Legitimate
... ... @@ -75,8 +75,8 @@ previously considered can fit into this framework:
75 75 \item \textbf{Chat client:} the content is the messages exchanged by users,
76 76 and efficiency is determined by the amount of screen time and interaction
77 77 required to retrieve and render incoming messages and generate outgoing
78   -messages as replies. Value is measured by the content of the messages and
79   -efficient chat clients send and receive a large number of messages per joule.
  78 +messages as replies. Value is measured by the content of the messages.
  79 +Efficient chat clients exchange many messages per joule.
80 80  
81 81 \item \textbf{Video player:} the content is the video delivered to the user
82 82 and efficiency is determined by the amount of network bandwidth and processing
... ...
paper.tex
  View file @f672f89
... ... @@ -77,7 +77,6 @@ Apps}
77 77 \input{usage.tex}
78 78 \input{metric.tex}
79 79 \input{results.tex}
80   -%\input{related.tex}
81 80  
82 81 \input{conclusion.tex}
83 82  
... ...
results.tex
  View file @f672f89
... ... @@ -39,6 +39,8 @@ it through a survey completed by 47~experiment participants. Unfortunately,
39 39 our results are inconclusive and open to several possible interpretations
40 40 which we discuss.
41 41  
  42 +\newpage
  43 +
42 44 \subsection{Total Energy}
43 45  
44 46 \input{./figures/tables/tableALL.tex}
... ... @@ -115,7 +117,7 @@ redraws.
115 117 \centering
116 118 \includegraphics[width=\textwidth]{./figures/survey.pdf}
117 119  
118   -\caption{\textbf{Survey Results.} The height of each bar demonstrates how
  120 +\caption{\small \textbf{Survey Results.} The height of each bar demonstrates how
119 121 many of the suggested apps the user is willing to remove for better battery
120 122 life, with suggestions based on overall usage or our new content-delivery
121 123 efficiency measure. Our new measure does not convincingly out-perform the
... ...
usage.tex
  View file @f672f89
... ... @@ -82,17 +82,13 @@ different app features or app configurations used by Alice and Bob.
82 82 By computing value and, thus, energy efficiency, we can overcome these
83 83 weaknesses. A value measure should allow us to compare the efficiency of two
84 84 apps in different categories based on how efficiently they use energy to
85   -deliver user value.
86   -%, making it possible to compare games to email clients to video players.
87   -Comparisons within the same app category should allow users to
88   -select the most efficient email client or web browser. Aggregating results
89   -over all users, differences in app energy efficiency should reflect how well
90   -the app is written and how well it predicts and adapts to users, not just
91   -differences in the core features it provides. When comparing two users using
92   -the same app, differences in efficiency should reflect differences in
  85 +deliver user value. Comparisons within the same app category should allow
  86 +users to select the most efficient email client or web browser. Aggregating
  87 +results over all users, differences in app energy efficiency should reflect
  88 +how well the app is written and how well it predicts and adapts to users, not
  89 +just differences in the core features it provides. When comparing two users
  90 +using the same app, differences in efficiency should reflect differences in
93 91 app configurations or app features.
94   -%different app configurations or differences in how efficiently the app provides certain
95   -%features.
96 92  
97 93 \subsection{Evaluating App Changes}
98 94  
... ... @@ -101,11 +97,7 @@ and deliver more value per joule. Today's energy profiling tools may be able
101 97 to show the energy impact of adding a new feature or changing the way that a
102 98 particular feature is implemented, but energy consumption alone is not
103 99 sufficient to apply Amdahl's Law properly to the problem of improving app
104   -energy efficiency.
105   -%For example, if a particular feature consumes a great deal
106   -%of energy but adds little value, it is possible that it should be eliminated,
107   -%not improved. Overall
108   -Developers should strive to make the parts of their app
  100 +energy efficiency. Developers should strive to make the parts of their app
109 101 that generate a large amount of value as energy-efficient as possible, remove
110 102 parts that generate little value while consuming a great deal of energy, and
111 103 defer work on everything else.
... ...