Slidecast version of the short (50 min) version my "What's Wrong with Online Reading" presentation. Audio recorded during a Nov 2011 guest lecture. Unfortunately, some of the vocal commentary for the last dozen slides was lost.
This talk discusses how a wide range of research in web usability, psychology, education, and communication theory provides corroborating evidence that on-line reading is transforming cognition, learning, and the very nature of knowledge in some disturbing ways.
3. Discuss how recent research in web
usability, psychology, physiology, cognitive
science,
science political science media studies
science, studies,
and education provides a great deal of
corroborating evidence that
online reading is not
nearly as g
y good as it seems
4. In fact, the evidence convinces me
that is d
th t it i downright
i ht
dangerous
for our cognitive powers
and
for the future of democratic society
society.
8. “young people scan online pages very
rapidly ( y especially) and click
p y (boys p y)
extensively on hyperlinks – rather than
reading sequentially … they tend to
move rapidly from page to page,
spending little time reading or
digesting information.”
I. Rowlands and D. Nicholas, Information Behavior of the Researcher of the Future (2008)
9. “our empirical study seems to indicate
… that hypertext degrades the q
yp g quality
y
of reader’s engagement during
reading.
reading ”
David S. Miall and Teresa Dobson, “Reading hypertext and the experience of literature,” Journal of Digital Information 2 (2001)
10. “hypertext presentation resulted in a
lower comprehension p
p performance.”
Rouet et al, “Effects of online reading on popular science comprehension,” Science Communication 25 (2) 2003.
11. Readers with low domain knowledge
comprehend significantly better with
p g y
highly coherent texts (books).
Readers with high domain knowledge
comprehend significantly better with
low coherent texts (i.e., hypertext).
L. Salmeron et al, “Reading Strategies and Hypertext Comprehension,” Discourse Processess 40 (2005)
12. “the net total effect of the web is
actually to reduce learning compared
y g p
to print presentation.”
Eveland and Dunwoody, “An investigation of elaboration and selective scanning as mediators of learning from the web versus print,” Journal of
Broadcasting & Electronic Media 46 (1) 2002.
13. In a longitudinal study comparing digital literacy
in 2002 and 2009 across generations:
Improvements in technical li
I i h i l literacy amongst the older cohorts
h ld h
Big decreases in tasks requiring creative and critical thinking
amongst younger cohorts
“For the
“F th more critical and creative skills …
iti l d ti kill
experience and exposure to [online]
information seem t h
i f ti to have a negative effect
ti ff t
on the user’s performance.”
Eshet‐Alkalai, “Changes over time in Digital Literacy,” CyberPsychology & Behavior 12 (6) 2009
14. Both
user control theory
and
structural isomorphism theory
(
(communication/learning theories)
g )
predicted that
reading comprehension
g p
would be improved
online in comparison to p
p print.
Eveland and Dunwoody, “User Control and Structural Isomorphism or Disorientation and Cognitive Load,” Communication Research 28 (1) 2001.
17. Usability experts have observed
that over past 5-6 years
the nature of web usage
has dramatically changed.
18. Most web usage has
switched from Surfing to
Information Foraging
18
19. Information f
I f i foragers
(informavores)
are seeking very specific prey
20. Information foragers
rely on search engines
to get to the “information patch
information patch”
Because search engines make it easy to f d patches,
h k find h
foragers will spend little time looking for prey.
According to Information Foraging Theory:
Kvalue Expected value of knowledge
Ctime Cost (in time) to gain knowledge
21. As informavores,
we are “hard-wired”
hard-wired
to prefer quick finding
and processing of information
information.
24. long do you spend viewing
your average web page?
25. 25% of all web pages
are displayed for less than
p y
four seconds!
Weinreich et al, “Off the Beaten Tracks: Exploring Three Aspects of Web Navigation”, IW3C2 2006
26. 52% of all visits
are shorter than
ten seconds!
Only about 11% are visited for
more than 2 minutes.
Weinreich et al, “Off the Beaten Tracks: Exploring Three Aspects of Web Navigation”, IW3C2 2006
28. “users most often spent
p
approximately 10 seconds
viewing those documents
that they eventually
h h ll
identified as relevant and
also those that they
eventually did not mark as
relevant.”
Diane Kelly and Nicholas J. Belkin, “Reading Time, Scrolling and Interaction: Exploring Implicit Sources for User Preferences for Relevance
Feedback”, Proceedings of the 24th annual international ACM SIGIR conference on Research and development in information retrieval (2001)
29. Server-record analysis
hints that these studies actually
over state the average stay time
g y
(i.e., actual average stay is even briefer).
WHY?
30. Because adult
sites appear to be
the largest single
category of web
g y
site (with email a
close second)...
… and on average the stay time for
adult and email requests
is significantly longer
than
th non-adult and non-email requests.
d lt d il t
31. For most pages with an average amount of text
p g g
(600 words), users will only take the time to
read at best about a quarter, or, more likely, a
fifth of the text.
34. In a very interesting study comparing the
y g y p g
time spent reading a paper-based academic
article and the on-line equivalent, the
researchers f
h found that
d h
“a
“ very l large proportion
i
of [online] full-text views
full text
were extremely brief and
possibly cursory.”
David Nicholas et al, “Viewing and reading behavior in a virtual environment”, ASLIB Proceedings: New Information Perspectives 60 (2008)
35. Average reading times for 10+ page p
g g p g printed
academic paper varied between 22 to 45
minutes based on the discipline.
Average reading times for on-line version
averaged about 74 seconds.
Yet academics reported that they spent
between 5-15 minutes reading the online
version (even though they didn’t).
David Nicholas et al, “Viewing and reading behavior in a virtual environment”, ASLIB Proceedings: New Information Perspectives 60 (2008)
36. Average for academics: 74 seconds
g
Average for students: 100 seconds
Average for life science academics: 112 s
Average for business academics: 60 s
Average for computer science academics: 55 s
Research-university faculty spent longer than
teaching university
teaching-university faculty.
David Nicholas et al, “Viewing and reading behavior in a virtual environment”, ASLIB Proceedings: New Information Perspectives 60 (2008)
37. Study examining stay times for ScienceDirect
articles: 38 seconds on average
Study comparing reading of academic papers on
computer monitor versus paper:
Paper readers: 80% progressed past first page
and spent 50% of time on p g 2-n
p pages
Online readers: Almost no one progressed past
first page and those that did spent only 17% of
time on pages 2-n
Nicholas, D. et al, “Viewing and reading behaviour in a virtual environment.” Perspectives 60.3 (2008).
Huntington, P. et al, "Website Usage Metrics: A Re‐Assessment of Session Data." Information Processing & Management 44.1 (2008).
45. We
W are only able to see
l bl
things clearly and in focus
in the fovea
46. Word Skipping: Implications
Eye movements in reading are characterized by
short periods of steadiness (fixations) followed by
fast movements (saccades). Saccades are needed
to bring new information into the centre of the
visual field where acuity is best; fixations are
required to recognized words. … Some words are
q g
fixated more than once, some are initially not
fixated but immediately afterwards regressed to,
and some are not fixated at all.
Marc Brysbaert and Francoise Vitu, “Word Skipping: Implications for Theories of Eye Movement Control in Reading,” Eye Guidance in Reading and
Scene Perception (Elsevier Science, 1998)
47. Results of an eye-tracking experiment
eye tracking
in which subjects were being tested
for which text layout was easier to read;
notice that even when subjects
were being asked to read, very little reading
(i.e., fixations – shown as circles)
was actually done
51. Notice
Red areas show
only first two
words in headlines
are
scanned
Nielsen Group, “Email Newsletters: Surviving Inbox Congestion,” http://www.useit.com/alertbox/newsletters.html (June 12, 2006)
52. More recent research shows
Use s ead only the st
Users read o ly t e first
eleven characters
of an online h dli
f li headline
(forget about the body text).
More recent research shows …
12345678901
Nielsen Group, “First 2 Words: A Signal for the Scanning Eye,” http://www.useit.com/alertbox/nanocontent.html (April 6, 2009)
54. … but it has a long evolutionary past in
adapting cognitive t it f swift
d ti iti traits for ift
processing and responses to audiovisual
cues.”
”
Grabe et al, “Informing Citizens: How people with Different Levels of Education process Television, Newspaper, and Web News,” Journal of
Broadcasting & Electronic Media 53 (1) 2009.
55. Reading is unnatural
unnatural,
but scanning is not.
Humans are hard-wired
to excel at fast scanning
56.
57. We have fooled ourselves
into thinking we are reading
when consuming web pages,
g p g ,
in reality,
we most often are not reading,
g,
and indeed,
we are often blithely
unaware that we are not reading.
65. “Collectively, the models presented
illustrate that
ill t t th t as j journal archives came
l hi
online … citations became more
concentrated within fewer articles ”
articles.
“by enabling scientists to quickly
by
reach and converge with
prevailing opinion, electronic
ili i i l t i
journals hasten scientific
consensus”
James A Evans, “Electronic Publication and the Narrowing of Science and Scholarship,” Science 321 (July 18, 2008)
66. Power Law Distribution
rules the web (and more).
http://www.congo‐education.net/wealth‐of‐networks/figure‐7‐4.gif
p g g g
68. Whether you look at the web
y
as a whole or any subsection within it
(blogs, political sites, sports sites, etc) you
(bl liti l it t it t )
see power law distributions.
70. Google search and res lt pages
result
account for almost
a quarter of all pages
f ll
Weinreich et al, “Not Quite the Average: An Empirical Study of Web Usage”, ACM Transactions on the Web (February 2008)
71. It facilitates the quick
scanning and foraging
behavior of contemporary
web usage.
72. Google is so good that …
g g
75% of users stick to first page of SERP
50% of users click on 1st choice
20% of users click on 2nd choice
Majority behavior if not clicking on first two choices?
Reformulate search
Nielsen + Loranger, Prioritizing Web usability, 2006
73. “Students in this study seemed to
y
have a great deal of confidence
in their abilities to distinguish
g
the good sites from the bad.”
Yet
“Students are also not
consistently able t diff
i t tl bl to differentiate
ti t
between advertising and fact.”
Graham and Metaxis, “Of Course it’s true; I saw it on the Internet,” Communications of the ACM (2003)
74. “Overall only about 1 in 6 searchers …
Overall
can consistently distinguish between
paid and unpaid results ”
results.
Pew Internet and American Life Project, “Search Engine Users,” (2005)
76. Usability analyst Jakob Nielsen calls it:
y y
Google
g
Gullibility
Nielsen Group, “User Skills Improving, But Only Slightly,” http://www.useit.com/alertbox/user‐skills.html (Feb 4, 2008)
77. Michel’s 1911 iron law of oligarchy is a political theory
that t t that ll forms of organization will eventually
th t states th t all f f i ti ill t ll
and inevitably develop into oligarchies.
My iron l
law of googlearchy states that all f
f l h h ll forms of search-
f h
optimized web-based information will eventually and
inevitably develop into oligarchies in which a small
y p g
number of sites absolutely dominate the discourse on any
given subject.
79. Some studies say that datacenters account for
between 1.2 to 2.0 percent of the electricity
consumed in the United States.
By some estimates, if you were to view datacenters
as an industry unto themselves, U.S. datacenters
would be approaching the top five industries in
terms of energy use.
http://technet.microsoft.com/en‐us/magazine/2007.10.green.aspx
80.
81.
82. US data centers thus produce higher g
p g gas
emissions than the countries of Argentina
and the Netherlands.
Even worse, these numbers did not
include Google’s power usage.
Google s usage
83. Q: How much does it take to power a Google data center?
A: It's none of your business.
Google considers power usage to be a trade secret
http://blogs.techrepublic.com.com/datacenter/?p=118
84. One
O estimate:
i
Every time y search Google y could
y you g you
power an 11-watt light bulb for an hour.
http://www.gimmiethescoop.com/data‐center‐power‐consumption‐global‐warming‐will‐the‐web‐crash
85. These numbers did not include data center power usage.
Moberg et al, "Screening environmental life cycle assessment of printed, web based and tablet e‐paper newspaper," Reports
from the KTH Centre for Sustainable Communications, 2007
87. This is the key one …
but is still under-studied
88. Is li
I online reading
di
actually changing our
cognitive abilities,
bl
perhaps for the worse?
89. There have been some claims
that in fact the new media environment
is making us smarter.
g
90. These claims are mainly founded on
y
the Flynn Effect
(Q
(IQ test scores have been rising 3-5 points
g p
per decade since 1930s)
This growth has however been in scores below the
median, not above it.
Sundet et al, “The end of the Flynn effect?” Intelligence 32 (2004)
91. Recent research indicates Flynn Effect has reversed in
the
th past d
t decade.
d
Sundet et al, “The end of the Flynn effect?” Intelligence 32 (2004)
Teasdale and Owen, “Secular declines in cognitive test scores: A reversal of the Flynn Effect” Intelligence 36 (2008)
( )
Teasdale and Owen, “A long‐term rise and recent decline in intelligence test performance: The Flynn Effect in reverse”
Intelligence 39 (2005)
93. That is, the brain can change
is
radically due to novel stimuli
94. Neuroscientists have found that the
brain appears to rewire itself significantly
after prolonged internet usage
p g g
and
that brain ti it for
th t b i activity f paper reading
di
changes after internet usage.
95. Like McLuhan argued
in his “Narcissus as Narcosis” paper,
accepting new t h l i means we
ti technologies
undergo displacement in our perceptions
96. At this point,
the cognitive science
is still
not clear enough
for us to know
whether the adoption
of online scanning
results
in irreversible changes
in our brain’s ability
brain s
to read
in a traditional way
way.
97. Recap:
p
Five Things
Wrong
with online
reading
107. “If we are building a transportation system to serve the automobile, the
Spadina Expressway would be a good place to start But if we are
start.
building a transportation system to serve people, the Spadina
Expressway is a good place to stop.” Ontario Premier Davis
116. Leisure paper-based reading
paper based
still remains one of the
strongest correlates of post-
secondary success.
Gallik, “Do they read for pleasure? Recreational reading habits of college students,” Journal of Adolescent & Adult Literacy 42 (1999)
Kaiser Family Foundation, Generation M (2005)