Presentation to National Academy of Science workshop on Public Response to Alerts and Warnings Using Social Media. I argued that the citizen science model, in which volunteers contribute to substantive scientific research, is a great model for how to involve the general public in making accurate, actionable social media posts (Twitter, Twitvid, Facebook) that first responders can use to direct their efforts in a disaster.
Applying citizen science model to disaster management
1. Applying the citizen science
model to disaster management
W. David Stephenson
Stephenson Strategies
NAS Public Response to Alerts and Warnings using Social Media Workshop
February 29, 2012
What
if
social
media
were
to
make
the
public
full
partners
in
emergency
communications?
I
will
outline
a
scenario
in
my
talk
by
which
I
believe
training
the
public
to
provide
accurate
information
would
be
feasible.
That’s
not
the
situation
at
present.
2. While
of>icials
are
now
belatedly
coming
around
to
using
social
media
as
an
alternative
broadcasting
medium
to
disseminate
of>icial
information
in
disasters,
most
remain
leery
of
actively
soliciting
information
from
the
public
via
social
media
and
mobile
devices,
usually
citing
questions
about
its
reliability.
As
FEMA
Administrator
Craig
Fugate
summarized
the
prevailing
wisdom
in
a
speech
earlier
this
month,
most
of
his
colleagues
see
the
public
as
“…not
worthy,
not
trustworthy,
you
have
not
been
cleared.
You
do
not
have
a
background
check…
you
panic
in
an
disaster,
and
cannot
be
relied
upon
to
think
rationally...”
Fugate,
by
contrast,
is
open
to
public
input
via
the
social
media.
As
he
said
in
Congressional
testimony:
“I
often
say
that
individuals,
families
and
communities
are
our
nation's
‘>irst'
>irst
responders.
The
sooner
we
are
able
to
ascertain
the
on-‐the-‐ground
reality
of
a
situation,
the
better
we
will
be
able
to
coordinate
our
response
effort
in
support
of
our
citizens
and
>irst
responders.
Through
the
use
of
social
media,
we
can
disseminate
important
information
to
individuals
and
communities,
while
also
receiving
essential
real-‐time
updates
from
those
with
>irst-‐hand
awareness…”
But
even
Fugate
sets
a
low
bar
for
accuracy
of
information
gathered
via
the
social
media,
emphasizing
that
what
he’s
interested
in
is
primarily
a
large
number
of
data
points
that
indicate
the
need
for
prompt
action,
rather
than
expecting
that
information
necessarily
being
scrupulously
accurate.
As
he
said,
“
..
we
looked
at
social
media
as
the
public
telling
us
enough
information
to
suggest
this
is
worse
than
we
thought
to
make
decisions
to
spend
…
your
money
without
….
waiting
for
a
formal
request,
without
assessments.
….
All
I
need
is
enough
information
to
hit
my
tipping
point.
I
don’t
need
a
lot
of
information….”
3. Emergent behavior
While
I
understand
his
point,
I
think
that
Administrator
Fugate
sets
too
low
expectations
for
the
public.
Why
can’t
we
provide
information
that
is
not
only
valuable
because
of
its
volume,
but
also
its
accuracy?
If
individuals
actually
did
provide
accurate
information
that
would
really
be
actionable,
and
provide
substantive
situational
awareness,
that
would
be
consistent
with
what
we
know
of
behavior
in
general
during
disasters.
Unlike
the
stereotype
of
un-‐credentialed,
panic-‐prone
individuals,
years
of
research
at
the
two
primary
disaster
research
centers,
the
Universities
of
Delaware
and
Colorado,
have
demonstrated
that
what
happens
in
emergencies
is
the
kind
of
“emergent
behavior”
that
this
termite
colony,
and
the
other
social
insects
demonstrate.
A
large
number
of
individuals,
acting
largely
on
their
own
and
self-‐directed,
cobble
together
highly
sophisticated
collaborative
actions.
Emergent
behavior
is
a
higher
level
of
collective
behavior
-‐-‐
and
combined
intelligence
-‐-‐
that
couldn’t
be
predicted
from
the
behavior
of
individuals.
The
group
becomes
a
highly
capable
“superorganism.”
4. As
the
Delaware
researchers
report,
“Studies
of
evacuation
at
times
of
crises
have
now
been
undertaken
for
the
last
50
years.
They
have
consistently
shown
that
at
times
of
great
crises,
much
of
the
organized
behavior
is
emergent
rather
than
traditional
[such
as
this
evacuation
from
lower
Manhattan
on
9/11].
In
addition,
it
is
of
a
very
decentralized
nature,
with
the
dominance
of
pluralistic
decision
making,
and
the
appearance
of
imaginative
and
innovative
new
attempts
to
cope
with
the
contingencies
that
typically
appear
in
major
disasters.”
What
could
be
more
ideal
than
the
combination
of
existing
patterns
of
emergent
behavior
in
disasters
and
the
advent
of
social
media
that
directly
encourage
and
facilitate
emergent
behavior?
5. It
seems
to
me
that
there
are
two
relevant
analogies:
World
War
II
plane
spotters
and
the
growing
phenomenon
of
web-‐enabled
“citizen
science.”
During
World
War
II,
those
on
the
homefront,
from
housewives
to
elderly
men,
were
trained
to
spot
both
Allied
and
Axis
planes
through
tools
such
as
these
decks
of
playing
cards.
On
a
number
of
occasions
the
plane
spotters
were
able
to
identify
US
planes
that
crashed,
and
even
though
I
haven’t
been
able
to
document
any
cases
where
they
actually
spotted
Axis
planes
over
the
continental
U.S.,
I
have
no
doubts
that
the
attention
to
detail
in
the
playing
cards
and
other
training
and
the
volunteers’
zeal
would
have
made
that
feasible
if
needed.
Updating
that
approach,
I
was
one
of
many
volunteers
who
“searched”
for
Steve
Fawcett’s
downed
plane
by
using
Amazon’s
Mechanical
Turk
system.
We
were
given
a
visualization
of
what
the
plane
might
look
like
if
crashed
into
a
mountainous
area
and
then
asked
to
examine
photos
of
a
number
of
quadrants
in
the
area
where
the
plane
was
believed
to
have
crashed
to
see
if
there
was
any
similar
visual
evidence.
6. Citizen science
More
relevant
to
using
the
social
media
and
mobile
devices
in
disasters
is
the
current
phenomenon
of
citizen
science,
which
harnesses
the
interest
of
everyday
people
in
using
21st-‐
century
technology
to
participate
in
scholarly
research.
According
to
the
Citizen
Science
Alliance,
citizen
science:
•
“..
[is
able]
to
cope
with
extremely
large
data
sets
•
..
provides
quantitative
estimates
of
error.
This
is
an
essential
part
of
the
'wisdom
of
crowds',
allowing
us
to
understand
the
accuracy
of
the
data
we
provide.
•
…
naturally
provide[s]
large
and
powerful
training
sets
for
machine
learning
approaches
to
classi>ication
problems.
This
is
an
essential
part
of
our
future;
as
data
sets
continue
to
grow
we
will
need
to
hand
off
more
and
more
of
the
routine
tasks
to
machines;
by
doing
citizen
science
today
we
can
help
train
them.
•
…
[can
lead
to
serendipitous
discovery
]Serendipitous
discovery
is
a
natural
consequence
of
exposing
data
to
large
numbers
of
users,
and
is
something
that
is
very
dif>icult
to
program
into
automatic
routines.
Humans
are
naturally
programmed
to
keep
an
eye
out
for
the
weird
and
the
odd,
even
while
sorting
most
objects
into
more
mundane
categories.
•
While
the
primary
goal
of
our
projects
is
to
produce
academic
research,
by
their
very
nature
they
are
also
outreach
projects.
As
it
involves
our
volunteers
directly
in
the
process
of
research,
citizen
science
is
a
powerful
tool
for
both
formal
and
informal
education.
Unlike
traditional
education
programs,
from
the
moment
users
>irst
interact
with
one
of
our
project,
they
are
not
only
learning
but
also
contributing
to
science.”
Doesn’t
that
emphasis
on
classifying
a
large
number
of
data
points
and
serendipitous
discovery
sound
similar
to
>irst
responders’
need
for
rapid
reporting
of
wide
assortments
of
disaster
information?
These
approaches
have
led
to
signi>icant
scienti>ic
progress.
One
amateur
astronomy
researcher
participating
in
the
Galaxy
Zoo
project
found
“Hanny’s
Voorwerp,
a
dust
cloud
generated
by
a
nearby
quasar,
and
participants
In
the
National
Geographic’s
Field
Expedition:
Mongolia
analyze
GeoEye
satellite
images
to
identify
potential
dig
sites
for
archaeologists
to
explore
in
Mongolia.
The
information
provided
by
amateurs
is
detailed
and
scienti>ically
valid.
This
scienti>ic
rigor
doesn’t
just
happen:
the
supervising
institutions
put
a
lot
of
effect
into
creating
simple,
easy-‐to-‐follow
educational
programs
to
train
the
volunteers.
Why
can’t
that
be
the
case
for
emergency
response
as
well?
7. #wxreport
One
government
agency
that
I
think
does
follow
the
citizen
science
model
about
soliciting
speci>ic
information
from
the
public
during
a
disaster
is
the
National
Weather
Service.
It
urges
people
observing
exceptional
weather
patterns
to
submit
Tweets
using
the
hashtag
#wxreport.
The
hashtag
makes
the
Tweet
machine-‐readable,
so
that
the
information
becomes
additional
data
points
for
forecasters
to
consider.
Given
the
importance
of
hyperlocal
conditions
such
as
microbursts,
this
can
be
invaluable
information
for
the
NWS.
8. Applying
the
citizen
science
model
to
disaster
response
is
not
just
an
academic
intererest
for
me:
I
have
been
directly
involved
in
several
projects
over
the
past
decade
that
I
believe
show
it
is
possible
to
use
the
same
technologies
that
individuals
can
use
to
report
disaster
to
train
them
and/or
give
them
the
tools
in
advance
needed
to
provide
accurate,
actionalble
information.
Nine
years
ago
I
created
what
I
believe
is
still
the
only
comprehensive
program
for
smartphones
to
give
the
average
person
–
in
his
or
her
hand
–
all
of
the
relevant
information
needed
to
prepare
for,
report
and
respond
to
a
disaster.
9. Later
I
prepared
a
series
of
YouTube
videos
with
speci>ic
information
on
social
media
and
other
tools
for
disaster
response
–
a
project
that
I’m
belatedly
subjecting
to
some
long-‐
overdue
updating.
10. VITA
Wireless when you need it most
Finally,
I
designed
strategies
for
the
Wireless
Foundation
and
National
Public
Radio
to
train
the
public
about
how
to
use
social
media
in
combination
with
their
wireless
devices
to
report
to
authorities
during
a
disaster.
If
I
could
do
this
by
myself
and
with
limited
resources,
imagine
what
could
be
done
if
government
and
industry
groups
such
as
the
Wireless
Foundation
were
to
partner
to
design
high-‐quality,
compelling
outreach
programs?
11. Tweak the Tweet
I’d
like
to
conclude
by
focusing
on
two
areas
consistent
with
the
citizen
science
approach
that
I
think
might
be
particularly
effective
ways
to
increase
the
accuracy
of
citizen
reporting
in
disasters.
One
is
to
actively
promote
the
Tweak
the
Tweet
program
that
Leysia
Palen
and
her
Project
EPIC
colleagues
created
to
make
tweets
more
focused
and
machine
readable
in
disasters,
similar
to
the
NWS
hashtags.
The
system
is
made
up
of
a
series
of
simple,
easy-‐to-‐remember
and
short
–
so
they
don’t
use
too
many
of
the
precious
140
characters
in
a
tweet
–hashtags
that
serve
to
identify
the
data
that
follow
them.
Like
the
NWS
hashtag,
the
TtT
hashtags
make
the
data
that
follows
each
of
them
machine
readable.
The
system
was
rushed
into
service
during
the
height
of
the
Haiti
earthquake
recovery,
and
proved
a
valuable
way
of
structuring
the
information
that
residents
and
aid
workers
were
reporting
from
the
scene
of
the
diaster.
For
my
own
use
I’ve
prepared
a
laminated
wallet-‐sized
card
with
my
emergency
contact
information
on
one
side
and
the
TtT
hashtags
on
the
other,
to
make
certain
that
I’d
have
the
information
on
me
if
needed.
Why
couldn’t
similar
cards
be
mass-‐produced?
12. Twitter videos
Finally,
I’d
like
to
focus
on
one
aspect
of
social
media
that
I
don’t
believe
gets
enough
attention
from
any
of>icials
in
disasters:
streaming
media
sources
such
as
Twitvid,
posterous,
or
vimio.
Many
of
us
now
carry
smartphones
capable
of
recording
and
broadcasting,
in
real
time,
many
minutes
of
High
De>inition
video.
Of
course
there
are
many
disasters
where
bandwidth
is
severely
limited,
making
Twitter
the
tool
of
choice
because
of
its
low
bandwidth
requirements,
but
when
infrastructure
is
intact,
consider
the
wealth
of
information
that
could
be
conveyed
by
a
willing
volunteer
panning
his
or
her
smartphone
while
also
narrating
what
is
being
said:
ν! an
assessment
of
priority
damage
(downed
utilities,
buildings
hit
by
explosions,
residences
destroyed,
etc.)
ν! placing
the
video
in
context
(including
content
such
as
landmarks)
ν! identifying
possible
terrorists
who
run
from
the
scene
Multiple
videos
shot
from
multiple
perspectives
can
help
authorities
assemble
a
literal
and/or
a
>igurative
comprehensive
view
-‐-‐
talk
about
situational
awareness!
One
need
only
remember
the
video
shot
by
a
Virginia
Tech
student
immediately
after
the
mass
shootings
there
that
was
rebroadcast
ad
nauseum
by
the
cable
news
channels,
or
the
countless
videos
of
violence
during
the
Arab
Spring
demonstrates
to
appreciate
the
wealth
of
information
that
could
be
conveyed
by
such
a
video.
Yet,
I
have
been
unable
to
>ind
any
public
agency
that
gives
guidance
on
what
to
include
in
such
videos
and
how
to
submit
them.
13. For more information:
W. David Stephenson
Stephenson Strategies
D.Stephenson@stephensonstrategies.com
508 740-8918
... And don’t forget to read Data Dynamite: how
liberating information will transform our world.
I
agree
with
Administrator
Fugate
that
sometimes
the
barest
of
real-‐time
information
can
be
invaluable
to
>irst
responders
in
determining
the
scope
of
the
affected
area
and
the
resources
to
mobilize.
But
does
that
mean
that
we
should
automatically
settle
for
the
lowest
common-‐denominator
information
from
citizen
responders?
That’s
not
good
enough
for
citizen
science
experiments,
which
have
proven
that
amateurs
are
able
to
provide
accurate
information
if
they
are
educated
in
advance
about
what
is
needed.
I
believe
that,
if
given
the
information
they
need
in
easily
accessed
forms,
citizen
responders
will
provide
information
that
is
not
only
timely
but
also
increasingly
accurate.
Let’s
launch
education
programs
that
will
do
just
that.
Thank
you.