Embed code
You're sick eyes is very proud toward
the social impact award to we should
feel and which is a professor of
computer science and the chair of
information science at the university
of Colorado wood older. Um and I wanted
to say just a little bit about the
social impact award when I looked at
the definition of this award it said
it's given to individuals to promote
the application of human computer
interaction research to pressing social
needs and I thought that combination of
applying original research to pressing
social needs was I really apt
description of relations contributions
where she's done great research in the
area of crisis informatics a field that
deals roughly I'm legislation define it
in great detail it deals with how
people use information communications
technology to deal with crises in what
could be more of a pressing need in
that so please join me in welcoming
mention Hi everyone thinks line things
that I really for this award as you
will see in this talk the inroads that
we have made in this area area are the
result of many many hands. So I'd like
to begin by offering my simple working
definition of what races informatics is
Chris informatics is the study of
information communication technology in
relation to actual or potential mass
emergencies with a particular focus on
the role of social computing in such
situations now it's a social media is
one large classes social computing
technology and its advance has brought
major attention to the nature of large
social movements which include disaster
response that we don't often think of
disasters in quite that way either
working on developing this field for
the past decade starting just after the
December two thousand and four indian
ocean tsunami before hurricane Catriona
hit in the US that falling August
mobile phone diffusion was finally
highly in the US after lagging behind
the rest of the world. But the US still
did not a recent data services so in
the US we were not texting believe it
or not still in two thousand and five
not not greatly the very first camera
phones were making their appearance in
the summer of two thousand five but
again this was happening outside the US
and the world over there was very
little in the way of social media in
the way we think of it today and we did
not have that term available to us at
the stage was set for significant
change after having worked in the area
of mobile telephony and social
computing in a range of what we call
everyday environment is you know I
began to wonder what could be possible
when there was a collective turn to a
set of time and safety critical needs
in what situation for the very idea of
collaborative technology you put to the
test more so than during the massive
social disruption of disaster events
Catriona came along and started showing
a west ass what that could mean for
that I spent that spring semester
camped out in the university of
Colorado natural hazards library so
looking for the conceptual theoretical
and methodological connections between
our home field of human computer
interaction my other home field of
computer supported corporate of work
and social computing with what is a
rich social science literature on
collective action social convergence in
relation to disaster. So what I'd like
to do is talk is accomplished a couple
of things a little too formally is
outline the field of christ's
informatics as I see it today really
kind of you know from my point of view
but trying to imagine the concerns of
many but I want to do that by way of
telling the story of how it came about
through collaboration. I include this
is part of the account not just because
acknowledging the contributions of
others is the right thing to do of
course but because I think I'd like to
have you consider how the
diversification of skill sets had
direct bearing on this field scores
sent to whatever extent it has achieved
it to its success. So let's see you.
This workgroup from one investigator
basically too soon after my very first
group of graduate students and you will
know these folks and see reviewing is
here I don't know if she ended up there
she and that and these these folks are
now at universities and research
institutes and government agencies of
their own doing some terrific work
expanding this work even for their in
nine in two thousand nine a large NSF
grant expanded and intellectually
diversified faculty base it see you. Um
as well as with the our partner
university of California remind with
queer mark how many of you will know
but other than Gloria these we're not
HCI faculty these were people in our
people in computational linguistics a
point sharing telecom policy and so on.
And then more students including join
white movies think tennessee's work all
also be talking about today who are
getting ready to graduate shortly and
the menu students were affiliated with
these other faculty themselves to are
intellectually diverse as all people
are but I would like to make the point
that I found I have found that these
are students have come from a range of
disciplines in their bachelor's or
master's degrees sticks come work in
this area and I think that's really has
led to the richness and I think the
broader impact of the work new club
faculty collaborations of the last two
years have further diversified our
portfolio if you will with colleagues
in civil engineering communication
environmental design and what the
national centre for atmospheric
research with meteorologist you're all
just to themselves are trained as also
trying to social sciences and then I
have to recognise two of our colleagues
that you will know from the kite
community I mean there's Ascii and
Alexander source that could impose
talking with a grip over these ten
years. And that he was waving people
working in this area our our traffic
set of PHD students including two who
are here at high this week Melissa
because and rubber so and then rubber
said will be speaking later on Thursday
I'll tell you more a little bit about
that so you can have the opportunity
attend attend that talk as well. And so
you can see this is the real
correlation of effort across
disciplines a lot of cross disciplinary
advising interaction and cooperation.
But I think has been an important part
of this account. So together we had
publisher and study research on events
utterly events listed here in this
rough time line and collected data on
many others social media data basically
but other data kind as well. And if and
we've of course a published across
events as well we theorise about issues
that are driven by events alone. And in
total are group has produced over
seventy papers dissertations and MSI
thesis documents in this deck in this
time frame which I'm really proud of
and we've had a commitment to making
this work available to a very very wide
audience which is mostly good but
actually introduces some concerns that
we have to think about how we think
about what what happens when our
research can have impact on talk about
that and just a second including as
well a very large practitioner audience
in this again as I think a central
objective in our in our research the
development of this new area would not
have happened without the generous
support of the US national science
foundation and it's here again that I'd
like to make the point that the
diversification beyond HCI allowed for
the diversification of grants forces
that then was able to find that work
and then became cut this virtuous
cycle. So we've been able to seek not
just funding from the computer science
directorate which in the American
context is where much HCI research is
funded but also in the engineering
directorate and in the geo spacing
geotechnical director it's and so
because this problem as you can imagine
is is pretty cross kind cut cutting and
so again I I just want to state how
valuable this is into building and
sustaining momentum to allow us to
really drive in a dedicated fashion
towards a topic that I think I think
deserves deserves that kind of
attention and and having that kind of
reach and I I reflect what's happened I
see how the HCI sensibility plated
persisted role in all of this even
though there has been a great deal of
specialised leadership across are group
and across time and give me pause
because I often think that the role and
I'm speaking a little bit I'm speaking
for myself and I'm really pretty happy
projecting a little bit on to you but I
think it's fair to say that many of us
often feel that HCI when we're trying
to collaborate on large far reaching
projects it's often tacked on as kind
of something that in service to others
I think this happens a both research
and industry and I think perhaps in
resistance to that sometimes we find it
easier to collaborate with people who
are much more like ourselves because we
believe or at least I believe there is
basic research to be found in HCI
itself. So what happens if we went a
step further and consider what could
happen when the HCI sensibility will be
this multidisciplinary projects what
happens when it leaves different
disciplines with an engineering
computer science in different
disciplines within social science I
think what happens is that because we
had the training that we have and the
sensibilities that we have we're able
to carefully articulate a set of
research questions that can set a
course for ambitious cross cutting
research. So one and offering today in
this talk is a structure for how to
think about the field of Christ
informatics as it exists today and it
came about from the doing of it's as a
very interactive way of looking at the
field in looking at the all the
products of the various groups now that
are working in the space and I hope you
see that you see I sensibility coming
through in addition to the mix of all
these other disciplines. So before I
get into this so I need to do a couple
of access staging one of discussing I
will win a little bit of detail
actually but our system a logical
commitments are to this area and it's
this is also an idea I've been kind of
trotting out playing with and I hope to
do more of it this summer in at at some
other venues. But I think it service
both the purpose of seeing what you
think about it. But also that kind of
against it waiting ourselves there. So
that's when I posted in the second act
will be to then move into the disaster
area the domain and then do some
articulation as some differences I'd
like you to be aware of their before we
don't talk about the five branches. So
I see currently that there are three
major epistemology isn't each the I am
willing to be wrong about this and had
this for the refined but just for the
purposes of being clear about where we
are and where I am this is how I see
the world. Um and I see first as impure
cold research as a major at this module
stance of HCI and this is in general we
must associate with the sciences and so
many others I think still in HCI feel
the need to work in reaction to a
different forms of empirical work to
justify stick critical age the eye for
example so this is that you Alexei I
this is a very dominant kind of stands
critical studies and theories are what
we most a sissy with humanities but
we're seeing more and more of that in
in in HCI and then this terms I like so
much research to design which is
anywhere phrase and I believe Jerry for
these Ian Johnson women are the ones to
credit with that term I think it's a
new term better reflects this I think
and at a an ongoing commitment that
we've long had that the ability to
understand things first comes to the
creation of things especially if there
don't have analogies for whatever it is
it's being created in in the that
already exist or group and I imagine
that I I live in a computer science
department so I have to explain this
often I work and speak to different
emergency manager groups I have to
explain this quite a bit to time
perhaps doing a little over explaining
here but I'm just gonna go with it for
now but we're strongly impure call I
think as our group is growing and in
diversifying working T worse trying to
bring in more critical. So analyses
into the work that we do and I think
perhaps I underestimate how much
critical analysis we do because as
you'll see when I talk about branch one
I feel that what we've done is sort of
abolish with the typical troops that we
assume are is the story of disaster and
you can't do that if you want to get a
critical view but only the very
strongly into this kind of impure cool
view empirical research that then
follows. So let's so let me just for
their dive into this empirical form of
great this is going to develop as a
matrix as you'll soon see so empirical
science we have impact positivist an
interpreter this forms of enquiry
positivist very simply forms of enquiry
is what we most strongly associated the
sciences many of us are taught the
scientific method in school so we think
a poetic of experimentation the goal of
this parts of us research is to prove
knowledge and so what we have there is
hypothesis testing but of course what
happens when you don't what we don't
yet know what knowledge needs proving
well then you find yourself perhaps
wanting to align more with interpretive
S forms of enquiry and it's also impure
call it's also progress as many of you
know it builds knowledge it formulates
thesis statements its data driven it's
highly interactive and I think the
basic building work that we've been
doing in christ's informatics falls
mostly into these interpreters and and
interprets find it in great and and so
and then there's what this means to H
the alliance this with the matrix comes
and I think there are two ways to
describe the goals of HCI and that's
summit informative and somebody
research is to study something that is
preexisting or is already occurred
without deliberate intervention whereas
and there can be a and and some of
research could be your positivist or
interpret that so where as part of this
research asks does X for example
influence why and how much interpretive
is some and somebody research asks how
does how does X influence why it makes
the you know particularly you have very
strongly I think formative goals where
the research is meant to be conducted
in such a way as to inform the design
artifacts but also policies procedures
and so on. And it's this idea of
formative that drives that the you know
the very familiar radii tenants that we
espouse of iterative design engaging
with users that's the formative
research that we are committed to more
many of us are committed to and this
too can be there positivist we're
interpret this. So positivist formative
work might ask does this design choice
have some effect and that's where AB
studies will sits right large scale
maybe studies for example among others.
Um interpreted this formative research
might then ask well how does this
design have an effect. And so I would
say that in the work over these ten
years there are examples of where we
worked in most of these cells at one
point or another but we work I would
say in this interpret this line of
enquiry primarily and we work in this
formative interpretive list point of
intersection. But certainly
interpretive is but it is the most
dominant my work personally and and and
the reason I'm making the point of
saying this is one to kind of sit
against it with the work suggests that
this field is a growing field that
we're trying to discover what the
problems are and how to articulate them
fast to then do sit subsequent work
design work other kinds of
interventions and so on. Um but I would
also like to say that constitutes basic
research this goes back to a little bit
of Lawrence introduction and the reason
why I feel I it's important to say this
is that I think the social import
hacked of our work is really important.
But I think I think even I thought not
long ago that to have social impact one
had to be various it very closely to
applied research which is not a bad
thing but as it turns out I think it's
very possible and in fact an obligation
of scholarship in the world today to
conduct basic research and imagine what
the cat impact can be bike carefully
sort of monitoring and imagining what
those translation efforts will be so
one can do basic research and have
social impact that's that's the that's
my but my argument okay I'd like to set
the stage once more by distinguishing
between I'd like to I would like to
know it'd into discussion of the the
the different kinds of crises that we
might considering what do we mean by
crisis and before I talk in further
detail I want to make the distinction
between hazards with exhaustion is
verses and dodges agents my group has
worked across a range of hazards
actually both kinds of hazards but our
biggest contribution has certainly been
in examining social media behaviour
social computing behaviour in relation
to events that have exalted this agents
those that cannot be apprehended or
easily apprehended like a hurricane
like a tornado like a big terrorist act
that that apparently comes from the
outside and cannot be immediately dealt
with in the event of the experience of
the disaster as it's being lived in
that moment. And this is that just a
decision to endogenous agents agents
that are from within and this often
means that there's a perception that it
can somehow be stopped and so often
this behaviour is perceived as criminal
okay so the reason why this is
important to distinguish is that I
think it has significant impact on what
we understand the online online proud
to do it also has impact on what the
offline crowd as already in place crowd
but significantly I think it has even
greater chain attacked on the on line
crowd because he there's such a large
observant audience coming sometimes
from the whole world looking at some of
these major events. So one examples of
Virginia tech shootings which is
something we did examine with
therapeutic and others that you saw on
the screen earlier and in that and then
it was the names of those who were
killed or injured he was cut was the
subject of what the crowd was trying to
search for and I know the shooter we
set at the scene there so there was
nothing to apprehend at that point in
the Boston bombings it was the pursuit
of the perpetrators that was the
salient problem that the and and and as
many of you might know the crowd got
that wrong the rhetoric community got
that wrong too great detriment actually
and so within tardiness agents the
system drives itself towards the
apprehension identification of in
individuals is a general statement I'd
like to make but I think it's an
empirical one that can be that can be
something worth examining even further
but in general I observe that the
character of the interaction between
people online and with the perception
of the agent is quite different and it
dries towards issues of blame justice
and forensics in the case of natural
hazards or other exhaustion as agents
there are lots of problems to solve the
but it's a much more diffuse kind of
information in question answer kind of
environment there are the the problems
are still significant they're not
necessarily minor but there are just so
many that it defies the crowd's
attention writes the crowd is doing
lots of different kinds of things. And
so therefore this online crowd
instructions itself differently social
structures are different in these
different kinds of environments. So the
field of christ's informatics it should
it continue when I hope it does should
address a whole range of hazard see all
the bad things that can happen right I
mean that's fine but but like you more
bombings and NX but now that field is
maturing and I think it's important
that we make it distinct distinction
between the expectations of a field in
expectations of a lab or the
expectation that all there are the
expectations of the paper and so now
that we have that a wide audience
people we one paper and try to make
some very broad implications from that
without appreciating what these
concerns are so in other words just
because there is all this behaviour
occurring on mine when bad things
happen doesn't mean that it's all the
same thing it doesn't mean that there
should be a singular interpretation of
what's going on of course in that
actually a very technologically
deterministic you things to to think
such things. Um and I think because
crisis informatics research is now
being read by a very wide academic and
practitioner audience it's being
misread because readers don't
understand this even just this
distinction between different kinds of
eight agents and even authors do not
understand this distinction. So though
there's this wonderful story perhaps
the impact of this field there's now
off now this thing that we have to be
very careful how we tried and how we
communicate to our audience what it is
that we mean what the impacts are and
we turn in fact some basic ideas in HCI
like technological determinism and and
other things I'll be talking about just
a little bit about how about what and
how this means of how we do our
research. So the court phenomena of
interests just just to reiterate is not
social media the core phenomena are the
different kinds of hazards we should
see different kinds of collective
action information seeking into in
relation to different kinds of events
and we do okay. So now they turn to the
five branch that christ's informatics
what I'm going to be doing is I'm going
to animate them with examples largely
from our own work which again is mostly
about exhaustion is agents. So the
branches I think still stand up even
when you talk about endogenous agents
but the branches. I've written them for
the start in such a way that they don't
dictate what the results are they
describe the different topics with in
this field that could be examined. So
some of the results all used animate
the ideas might be in contrast to other
advantage you read about or perhaps you
yourself study. So this would just be
another conversation that we would have
about what about what that what what
what what the city of these different
branches might mean okay I think that's
one the consequence of having done this
work for one being able to have the
privilege to be able to be dedicated to
it for a while so that we can make
those distinctions that we and that we
should make those distinctions okay so
the five print as an overview artists
first I'll talk about social computing
as it's perceived in relation to
professional emergency management and
why it's hard for emergency management
to adopt segments I will show how a
second line of research capitalises on
the spontaneous social media activity
that actually occurs in response to
disasters and what people are trying to
do their mother problems and
opportunities are there there all talk
about the challenges of collecting
crisis data and that how and
specifically I mean social media crisis
data and how the collection is already
a kind of sampling decision that limits
the kinds of things that can be done
that down the road how that's another
point of caution about how we're going
to read and conduct this research going
forward. And then for all spend the
most time summarising the internal
social structures that arise online in
response to disaster events but I will
come back to in a sense branch one with
how what happens nominally speaking
online is has connections to emergency
we're actually in the physical world.
This is a point I bring up here because
we've been this that criticism that's
been level that is pretty severely
about you know you know all that stuff
about what's happening online but how
does that help people on the ground and
so I'd like to talk about that book is
that some it informative kind of
research enterprise and so we'll close
there okay so to help keep track I have
colour coded each sections is the blue
section and this will hope you keep
track of where we are and talk maybe
even help me keep track of we're gonna
talk as and some of the topics as
you'll see some of the ideas are
overlapping and often deliberately so
with this will help you keep track of
where we are so what about to talk
about now I'm eventually going to the
ideas of how to fix professional
emergency management but to the extent
that perhaps not units audience how are
not familiar with disaster I am also
imagining that in a sense we the
general public belongs in this category
of concern to and it's here where I'd
like to dismantle that ropes of what
disaster response is and how people
react to it any bearing that has on
social media and how we study that and
yeah so let me start there okay so
Christ informatics as or research field
disrupts more comfortable frames of
reference about who does what in
disaster. So when we consider
technology solutions are impacts we
must expand our understanding to go
beyond what formal formal responders
need these emergency responders you see
here perhaps in terms of situational
awareness which is of course something
that everyone's trying to achieve an
emergency managers are trying to
achieve have an overview situation to
know what's going on people to make
decisions about where resources are
gonna be allocated because a disaster
that's a necessary condition disaster
is a disaster if the problems XXC
outstrip the resources that can be
allocated to solving a problem right.
So hard problems have to be made you
and you make them by having an overview
of what's going on. So situational
awareness is as holy grail but we have
to go beyond honest and recognise that
people on the ground to air victims of
disaster are trying to do the very same
thing they're still trying to figure
out what to do even when bad things
happen to them. So look what they did
here here in the in the aftermath of
the two thousand five hurricane katrina
and then read it which closely followed
it one of the few walls in a huge use
master damage the big sports arena to
make it a bulletin board for missing
persons and they could no pictures
member they had no phones no camera
phones they didn't know they were going
to be the pictures that they have them
there would be no way to print them out
and put them out and distribute them
they're all just worn not pieces of
paper and cardboard these bulletin
board is really natural phenomena that
he certainly before the advent of ICT
and he's built imports happened
everywhere across the region in the
more than one hundred eleven official
right across shelters. But also in
other places hotels and other public
areas across the US gold coast region.
And it's the best they can do under the
circumstances now it probably didn't
work out very well as a way to find
people unless unless these photographs
themselves by female helped perhaps
because the only people going to be
shelters where the people who were
bought the rights to be the shelters
they were all tagged with wrist tax
right. So it's not entirely clear who
they were advertising too but as zero
you one set in one of our lab meetings
in two thousand and seven people do
this thing do this kind of activity
because they're desperate the desperate
move right. So if they know that this
isn't going to be the way they can
solve their problems but there's trying
to find their absolutely it makes the
point that everyone is striving for
situational awareness everyone even
under their very limited impoverished
circumstances information harbours
circumstances and otherwise are
striving for situational awareness. So
so you just two important things to
this landscape first expands the
audience to get outside that shelter.
So here are to be reports that are you
located and and I envisioned here on a
to earth representation from the Google
person finder database and their
representatives tweaks and these were
the missing person finder people
reports following the tsunami in Japan
okay it's the same behaviour but now
there's potentially a global audience
to assist second social media exposes
the informal work conducted by the
public. And because we're trained in
HCI and perhaps ECW we know that the
very idea of informality isn't is
important to understanding real life
and of course. I could we have to think
of Lucy such men's contributions and
her colleagues yeah and her work in
plan since action that sticks this so
clearly the idea is that what we say we
will to does not reflect how things get
actually actually get done. This is
them or idea to keep in mind we build
technology we build policy or
especially when we built our policy
about technology because all of that is
an artifice the construct build on
rationalise logic about how things how
people do things and how people think
should be done because it's so much
easier to design for things how should
be done that how they are done of
course what happens is we miss all this
very important informal work that has
already happened in disaster has always
happen to disaster but digital traces
expose it to us in new ways. So we
think about disasters which we know are
not part of anyone's plan. Um we have
to really think about even we were
trained in this kind of thinking what
what we're doing when we bring our own
kinds of pisces to this problem so I I
love this but I find it extremely
moving and I think it illustrates this
point really well so here we see the
public mixed with formal response you
could tell by the the clothing that
they're wearing in the rescue
activities from the two thousand and
two earthquake in Turkey which was a
devastating earthquake they have many
devastating earthquakes that one was
particularly bad and we see this
blender formal and informal roles at
work here and we see orderly
this to this photo. So if we had lots
and lots and lots of tweaks like this
we may be able to sell the situational
probably earns problem next week to
driving data but there are other things
that are making this hard in addition
to the absence of a lot of this kind of
data is that you know social media data
are mounting in volume everyday there's
balding styles of communication there's
mixes of languages in any one one
hundred forty characters we we see it
in blogs and they spoke as well. Um but
there are also multiple housings that
occur on the planet at any one time so
when we were trying to distinguish
between between someone earthquake in
there it with another it became very
hard to do. So just in one a random
sample of a day was it was a big
earthquake doesn't well "'cause" rick
earthquake I give an assignment to my
students I didn't have time to work for
the homework assignment night at want
them to answer some questions about
retreating behaviour well turned out
they couldn't do it because they at
least at least ten other earthquakes on
the planet that day that we're being
wheat we didn't read about much more
small much smaller than the Costa Rica
earthquake but you couldn't tell easily
with how manual Reading what we treat
or associated with what so this is
become really challenging so there's
different automated process ease that
we and others thankfully around the
world are trying to do inc and also
human computations a part of this. But
the problem continues to be hard
because there was that perfectly I told
you I told you about but here's a more
typical tweaked okay so usually people
laugh more money. So that way you can
laugh so this is not really this is a
more difficult we but we should not be
dismissive of the street this week is
actually quite valuable tells a lot of
information you don't know that yet but
you will so hold on to whatever
thoughts you have and what we're to
that to you in a minute. So branch
three brings as to how we collect
social media data in response to
disaster so and it's how you it's how
you collect the data that you know that
that was a valuable to eat beyond what
it said it was how you collect it and
that's why the collection of this is so
much a part of the science of and
interpretation of what we do with our
data. So the next slide is really the
most profound results of all the risky
research we've done over a decade. So
here it is okay so actually I know I A
But we know this right. So then why do
we treat our data and what I if you
look at social computing research in
general not even just for disaster
research but especially where I live I
see it in disaster research all the
time. We see that people are collecting
on high signal stuff right "'cause" and
and this happens for any kind of big at
that right there only collecting on the
it's there's so much volume of the only
they have to choose what they're
collecting ons they're clicking on the
high signal stuff and ignores the
context of the before after this
happens in machine learning this
happens in human analysis you know
we'll have an adheres on the annotate
we try to eat human computation tasks
will also look at this to be but we in
absence of the before after we have
done this we're trying to correct this
problem by introducing new ways of
looking at this. But still have it be
automated supplementing are qualitative
analysis but this is what you see in
their public record today and it's a
real problem and it's going to prohibit
our ability to move forward in ways
that really start getting at what will
become the more complex issues around
disastrous all show you just a minute.
So you would never do this in real life
you would never take a real centre one
sentence out of context when you're
passing by the hon here always people
talk about different things. Um but we
do it here and we do it especially with
weather. So now we go back to this to
be now imagine you could be whatever
you'd like to be could be a machine we
can be human whatever you'd like to
imagine in your head if your machine
you get ten thousand it's to liberty
like this if you're human you get a
thousand to annotate it. And this is
presented to you and it's all you have
just this content here. And you have to
make judgements about whether it's on
topic or off topic if it's relevant or
if it's truthful. And I'm betting that
most of you would say you just can't
know the answer to that right
especially because of the two
euphemisms we have a properly spot over
but it's the wrong for so as I said in
context history turns out to be pretty
informative. So I'm gonna show it you
know in context. And I'd like to tell
you why we're looking at this kind of
thing so so you can imagine the problem
that we're after so one the things that
we're sitting in our group right now
with and car is how the public makes
decisions about risk a risk around
coastal hazard in particular and in
particular hurricanes. So there's a
whole line of enquiry around what it's
called protective decision making so
how people perceive risk and how they
decide to evacuate we know that you
know they get about an evacuation order
they won't necessarily evacuate this
because they're not people are rational
but not rational on the way we think
they are so beep so so we're just
responders others get really upset like
about what they just listen to the
evacuation order well it turns out you
they can't exactly maybe they are L
maybe have you know a single mother
with three children or you know they
don't have any money to go somewhere or
it turns out really going that way
isn't a very good idea based on other
information that they have and so what
there's just larger idea of how people
take protective action even if they
can't evacuate and then there's the
question of when how what kind of
protected action today take before they
actually decide to evacuate in in in
face of this direct. So that's the
general problem that we're trying to
examine and use a social media record
as a real time lots of we hope these
kind of decisions some research or try
to figure out if we can use it that way
to get these this kind of precision
about where one people make decisions.
So okay so here's our same guy and he
starts out with that we that says
people are really overreacting about
this damn hurricane. So this is a very
now I'm actually gonna you're machinery
human and you're looking at just this
isolation I think you would just to be
on topic and relevant okay and we would
catch it probably because of the word
hurricane although a lot of us we're
not filtering on her fingers was too
noisy for her king sandy so we probably
wouldn't of cotton street but it was
put to you I think you would think it
was a valuable and we would think of
this person is dismissive right but
then look at what happens. So he says
I'm about to put my I found on the
charger that's all he tweaks if it if
you all you saw was this in isolation
you would not know this had anything to
do the hurricane you could not make a
judgement about that. And then you know
thing in order then we get our dear
flying in this guide to eat and then we
get our fourth week that for me
personally resolves what happens
before. So then he he says I'm about to
make something to eat before the power
goes out and I'm willing as a human
judge it would make an adjustment to
believe this has something to do with a
hurricane the power going out does that
for me it helps me resolve the I found
on the charger is probably about power
and then it tells me that that you're
flying in the sky to be is the weather
eat right but the weather report right
and then of course also we see the turn
in his the changes his mind from people
are overreacting about this damn
hurricane to stuff is really out of
control I better charging my phone to
get something right but that classic
way of collecting data in this space
for these extreme events lots of
activity is to collect on keywords
maybe there have tags maybe they're not
but they're like as unique as possible
place names and all sorts of things. So
this is the only time we capture this
guy is with this thing to eat where
he's writing the governor Christie use
the governor of new jersey want the
state so that that is affected. And he
there's no problem down here so it's
actually kind of funny knowing we is
it's kind of finding deteriorates the
governor to ask when is it safe again
to head back on that clearly is make
you the hurricane but we're not sure if
it has anything to do with his
protective decision making right. It's
almost like he's act asking on behalf
of others because the wiki phrased in
the way he's writing to governor
christie. And then he finally says and
this the only evidence we get he says
it feels so great to be home sigh. So
never did he say you've got a tyrant we
finally gave you an extra never easy
say I'm evacuating I'm getting out of
here I'm going home I'm going away like
phrases that you might think would be
good phrases to detect evacuation our
project a decision making right. So we
have to look at the whole of the stream
to get some sense of what's going on
why do we do this why do we continue
what is that a war in social computing
research to look at things that we
might we were statement by statement.
And I think it's this to any of the
data structure that's doing it to us
and I'm really worried about this I
think the tear any of the data
structure in this case the tear any of
the two iterator structure compromises
research design we forget that it's
actually dictating what it is that
right kind of choices that we're making
it must be really really careful about
this for all social computing research
in general. So what do you do to solve
this problem well it's not
automatically easy but if you're
dedicated to working in this you've got
a your do it or work with others you
worked Rory had the competition
capability to first first clicked on
keyword collection and then go through
you get your keywords you could your
treats based on the keywords you find
you find your users by having collected
on those terms discover you have you
know user three for example that said
two things. And you you're curious what
I wanna know what here she said before
after so then you go back what we do is
we go back and collect we call the
contextual streams. So we get all the
streams of what they said before and
after these found rates. So the finding
of the first weeks is just the first
step then there's at least a second
step in here. And then I'd like to this
helps is bridging to branch for is we
do this for everybody or whatever
sample it is that we decided that we
need to do this for based on whatever
the criteria are and this is just of
course impressionistic but what I'm
trying to show you that this is where
then all the interaction between other
people lies we need to have this data
in order to see what it is people are
doing with each other and all we do our
collect on these words that we thought
would be important then we really don't
understand for example what it means to
evacuate. So the problem of course is
that takes data just explode when you
do this ago becomes and the manageable
and so one must have automated
assistance in order to even get at
eventually content or qualitative or at
that you know the virtual ethnographic
analysis or whatever it might be that
you might do with this there's no
question that you need some automated
assistant so we can't analyse it just
linguistically "'cause" the linguistics
and telling us enough so we need new
techniques for example and that
includes filtering on what people do
through non linguistic behaviour and
social media and then looking at what
they say then build models from that so
let me give you some examples of that
in keeping in mind that like first anti
for example they were five million
people who participated in that
conversation and certified figure out
who the evaluators were a five million
people you must you must actually have
assistance anyway so some of things
that we're looking at that we have a
pet and are looking at now are are are
these things so one kid starting
Christmas machine were at Colorado they
were looking at follower deltas so it
was more telling to look at the
changing of power count then look at
the initial power council changing
power count was a good predictor of
people who had very good information
and therefore we're actually closest to
the ground so was a good predictor of
people who were on the ground
experiencing an event by looking at the
following alters deltas pacing is
another thing that we might that we're
starting to look at so if somebody's
treating for example a great deal and
then set it suddenly stopped weaning
well that tells you a lot about that
person that suggest you might want to
investigate what's happening there or
they don't we very much at all meant
suddenly they start Reading a lot.
That's another kind of signal geo
typing switching is another thing that
we're looking at so the people choose
to turn geo tagging on off. I think can
tell you something about what the what
the audiences that they're trying to
reach and that is another dimension
another behavioural dimension that
might be valuable and then another
thing that was doing a lot of time on
right now is this idea of movement
derivation which connects to the
evacuation research I was telling you
about and the idea is here again to
look at among the various small number
of people who do you have to tagging on
can we figure out by comparing to prior
behaviour if they have evacuated or if
they have sheltered place turns out
this is incredibly messy it sounds like
it should be just like a oh but it's
not that's because people once you know
they'll go to the shore don't take
pictures and all go for coffee as
they're really trying to figure out
what should we do here. They do all
this kind of movie that sort of
disguises itself as regular activity
turns out it's it's connected to
evacuation so there's a there's a fair
amount of work at inferences has begun
here but the idea is what we're going
to is is that we can build we configure
the evacuated are are project to they
might be through their movement
behaviour and then we look at their
linguistic behaviour build our our
theories and then eventually are
machine classifier models based on that
linguistic behaviour the people that we
know we have found and then you can
apply apply those classifiers to the
much much larger world of attention of
those who participated with out geo
coding. This is work that Jennings the
Anderson is leading cap and still mean
a coke and and are in car colleagues
another lesson we've learned in all of
this in this kind of attraction to
driving information and people having
ideas about how disaster should be
right the movie version of disaster is
to be where the lure that information
this is like the second stomping ground
for people who want to work in the
space for situational awareness and one
that that's be pretty hard is able you
know all figure out where all the bad
information is and and especially in
hazard with sergeants agents I mean I
think that the indulgent agents we have
different story going on as explained
earlier but this is they presume is
also to be to be equally true with its
exactness agents and I think it's a bit
of a red herring I think it's a hard
problem to solve but it's not that hard
it's it it's the easiest of the hard
problems I worry and set about the
hyper locality in hypertext centrality
of data. So data that's true for me
what about where it might evacuate
might not be true for you in the back
of the room and were as close as you
know we can reasonably expect to be and
so I might have to evacuate this way
and you might have to back with that by
the by the way I checked those doors
Nazarene block so you can't exactly
that way. So you might also have to
evacuate this way the point is is that
even within very small amounts of
spaces and the information that is true
right this like this pursue that is it
true. So much research around is social
media data true which is which is a
strange question to me is we discovered
very quickly that it's a relative
concept same with it the temporally of
data. So information that scroll true
at time one is not really time to and
being able to discern this data we
don't I don't have a way to figure out
if this you know where is this bad
actor coming from three IBIP addresses
or whatever else that is really the
critical problem. We're starting to do
some work in that area but only just
barely and I really hope there are more
people who continue to work in the
space okay. So branch for so branch for
has to do with how the social
structures a rice how social structures
arise from the primordial soup of
social media. So we've come to
understand like how this happens by
looking at different levels different
units of analysis interpersonal group
organisation and institutional levels
an hour of analysis money just kind of
show this to hear hear very starkly
representing the informal and the
formal emergency management and the
public here I hope you realise this is
kind of a this is just for visual
purposes and it's the best I can do
with my graphics skills but anyway the
point is is that these social
structures are rise out of of people
I'm help people begin to share
information with each other. And they
witness other sharing information
that's this interpersonal unit of
analysis we want to know much like I
described earlier what is that that
they're saying and how are the
expressing themselves. Um the second
level of analysis is what I I just what
happens next which is what people start
working together and they operate on
and through the communications that
they produce because remember to tell
those it'll traces become the material
upon which they can work it becomes the
sight of production right. So not only
their own interactions with the
interactions that they're witnessing
and seen others do. And they so they
start newly mobilising around digital
digitally distributed information not
everyone does this but some will start
seeing things happen and say oh I bet I
could ask so once I get bad asses
person about where I can get the
supplier Clyde barrow that generator
they know something about this piece of
information do they know where somebody
is or I saw somebody needed help
antibiotics I've got antibiotics let me
start doing yes mister helping here and
so what happens is that these groups
star forming and developing and some
really develop into serious group
sometimes they last for days sometimes
they last for weeks sometimes alas just
for hours that they try to accomplish
something. And utilities grassroots
groups and this completely parallels
what we see happening in the physical
space in crap and bars where speak
about this very nice looking at the
hundred year history of disaster events
and this gets us to this organisational
level of analysis and then where I
think some really interesting questions
are merging now that we and I hope
others will be investigating is that
we're starting to see shapes of course
in these information relationships
within the whole of the institution of
emergency management so we can't ignore
what is happening from to eat. But the
level of to eat all we have to
institution emergency management and we
we fall it this that really follow and
how we get there okay so as I said
those people once found if they find
each other through the expression of
their questions and answers and
concerns express on source yeah well
sometimes start working together just
they do in the physical world. And this
quote but that was from work that Kate
starboard and I publish back and
doesn't eleven about the he dearest
quick I think explains this really well
in one of our subjects wrote to us and
said in the beginning I worked alone I
sorta recognising people this is on to
what are you seem to have good
information and we would support each
other we read to each other we help
find information for each other. And
then whenever colleagues in all of this
a person became a colleague because
they were one of the people who found
each other says eight days after the
events says I am stand we've gotten
supplies then we say people from the
rubble you brought them doctors we had
best team we are following putters and
he's going to call themselves onto
eaters and crisis treaters. And the
subset of this group then went on to
further formalise into a nonprofit
organisation and it was one story of
not a large number but a significant
number of groups that did this kind of
formalisation they developed it sounds
the cottage industry that rose out of
these new opportunities for these
volunteer technical communities as they
call themselves to emerge and operate
disaster anyways open street map is a
large organisation is that you might
know about it on the pocket PDF of maps
and they to this was a different a
slightly different story but I followed
a similar trajectory so they already
existed right they already had a
purpose of generating geospatial data
that was open and available for number
purposes. So the over some meaty
themselves internally mobilised in the
wake of he if you remember many
government buildings were destroyed
many government officials were killed
there was a real need for geospatial
data this is what open street map look
like the day before the earthquake
there's very little content and I don't
know if you can see you maybe can see
this is this there's a great deal of
content now here and if you could see
on the slide this project version of
the slide twenty one days after the
events of the whole country was really
matter very very high level of detail
in work that rubber so then Jennings
Anderson and others did we found there
were four hundred forty six mapper is
eighty three of them were brand new but
they were already a lesser members and
so and then we saw the same behaviour
happened four years later in the
Philippines in preparation for in in
the wake of typhoon you'll on the high
and and rubber this is the paper that I
mentioned earlier the Robert so it will
be speaking about later in the week
about how and why the patterns of
mapping actually look quite different
these colours represent different users
here between these two events even of
the phenomena were very similar and it
was a technical consequence avenue
mapping innovation that supported
mapping but he you will talk about how
it comes about through this idea of
always them scene itself as a as a
different kind of player and this
larger ecosystem of data production and
data reuse. So it's really a story of
institutional change that gets realise
of course in these technological and
social innovations and the way we see
it is that was some of the data
producing organisation was responding
to lots of external pressures in the
sort of in the given that they're
they've experienced a great deal of
success in achieving what they wanted
to be common source of good geospatial
data they then have this obligation I
think is how they see themselves to
then think about how they deliver that
data to others and so this is what you
see I think in this in this change it's
happening OSM so come Thursday last
session last talk. It'll be worth it
remote you for we're doing at a little
bit more work in this area to around
the Alex of course then we're trying to
open up the database which is again
another kind of tear any of the
database your near the data structure
tyranny of the database where am I cook
a PD as rubber will explain always them
can't see itself very well. It's not
like with the PDA where you have all
the documentation with talk pages in
history pages we can study with those
collaborations look like very very hard
to do all this time because the
databases impenetrable. So getting the
Anderson rubber. So they can understand
Rena coke in time you a Mccallum there
are mere and are working on this and we
have some open source code now
available to be able to look at the
user centred view of who's making club
edits to the map and how much and when
and all of that and this work is being
presented this week actually in Chicago
by Jennings at the American association
for Java first okay my last scratch
okay this is good almost there okay so
I one of the so what I think they
mentioned is one the criticisms that's
an levelled at S and I think it's been
it's great great gets a chance to
respond so now I bulk or will for are
you are you here oh I can say things
about "'em" is not here I know he's
here please not here so he has been I
think especially critical but it's been
very productive for us about saying you
know this is all you know all the stuff
that's happening online but what what
relevance does it have to what's
happening on the ground and is there
any relevance and I think the answer is
yes and I'm not not this is not
speaking of advocacy but I think the
challenge and setting it well means
that we can if we can study it well
then we have no right to write a paper
about it doesn't tell you what that
looks like because it's a very hard
thing to study right you have to I
think the connections have been very
kind of the femoral and tiny and minor
right and actually minor they're
important but there we know of course
people look at things online to get
help from people online I can't quite
remember if they got it from their
friend on face book core when sell
their neighbour on the street you know
later that day they can't make those
kinds of connections it's hard to say
that we that makes it anything
meaningful to say to you but what I
think we are seeing is that as these
events are unfortunately happening and
as these social structures are starting
to have some persistent sometimes and
then hours trying to replicate
themselves before starting see patterns
we're discovering in a new they're
being exposed other ways in which
people are using social media in other
kinds of environments makes ten that
really quickly what's happening in this
disaster that they're experiencing.
They are there is more transfer more of
a relationship between what's happening
normally speaking offline and online
that then therefore becomes study about
we can witness it now "'cause" it's
extended has some kind of duration. So
now we can study it it also means
sometimes have to be in the right place
at the right time to study it and this
is once again where the floods we
hadn't call rather be possible for us
just to be there to know what what to
study the other answer I have to that
criticism that's is that we must design
and create those opportunities those
connections as well so that goes back
to both the summit informative research
that I argue in the beginning that is
in play here right so that design
interventions when we did have just a
few examples before I close. So we have
two three forty what we have told video
what oh darn okay so okay so we so in
what what do I do and why we looked at
again response where floods the
evacuation of thirty eight courses that
were marooned or skimming them around
on a mountaintop you don't be an I You
to be on a mountaintop in your room and
what all the roads going out are no
longer passable and so in her
dissertation work she looks at how this
ensemble the people who could only meet
virtually because they were not in
colour out of many of them they we're
not on this on top was came together
and what they what the cast kind of
wide net for this expertise around what
it means to be a horse person to be
able to figure out this unusual problem
exactly thirty horses down mountain a
three hour journey down to a new ranch
and so the story goes is that the
expertise the interest of it it's
objective knowledge that expertise
shares enabled a degree of trust around
strangers that enabled the planetary
support of remote planning with many
unknowns but was still solitaire and
hyperbole and that's often where the
social media activity gets dismissed
mystics and what's not thirty forces
you told us it was fifty and so what
silliness is happening online here in
social social medial and well so they
have this area gets propagated lots of
things happen around it then what
happens is once they come on to site
after big workers orchestration once
again this problem gets mitigated once
again through their expertise through
online witnessing of the are on site
listing of the obituary their own
materiality working demonstration of
their own expertise which we establish
credibility and then kind of put we
measured what the response should be
we're developing outsourcing systems
around Boston found issues we are
working a great deal in the American
west this goes to the designing the
interventions and what with still a
Colorado on their own experiment
apparent how to include volunteers in
responses to natural has which we're
not getting into law enforcement stuff
but around natural hazards and then
most recently we're trying to the topic
of resilient infrastructures and
resilient communities and this once
again opens up the opportunities for a
diverse disciplinary commitment to this
topic so we are working with civil
engineers environmental designer's
decisions scientists urban planners
with at the university actor in Norway
where I have an evaluation as well as
you inverse of Colorado and so this
again speaks is kind of diversification
of what HCI research can be so the five
branches of Christ informatics research
as I see it today about a close now. So
just bear with me okay thank you. So
okay so as you can see disasters of the
sites of human innovation victims are
creative helpers are creative
responders are created and researchers
must be creative and vigilant and
respectful and integrative of so many
different forms of investigation and
all of that is I would argue is what a
CI researchers researchers are so good
at doing we are hard workers I think we
want to tackle very hard problems and I
think it's possible first H see eye to
accomplish a great deal on behalf of
many it is possible to manage both be
applied without losing any commitments
to the basic scholarship that also
drives research in in the most
essential ways and I think it's in fact
what scholarship demands of us today.
So I propose that together we consider
in our research going forward how can
or it's the artwork have greater social
impact. And I propose to you and to me
is it a matter of expanding the base of
collaborations for more intellectual
diversity no we sure that the research
we're doing is it being foiled by the
very technologies the new technologies
that were encountering. Um are we being
pulled in by its imperatives making us
unwittingly complicit in technological
determinism. And can we return to basic
H the I tenants in constructs that help
guide or understanding that behaviours
for becoming differently complex in the
face of all this new innovation. And
finally imagine the powerful things
that can happen when it's the eye
leads. Thank you very much for your
attention I didn't leave much time for
questions I'm so sorry we start a
little bit late and I'm here for other
question two minutes points as we have
two minutes oh two questions. So thank
you for the talk my name is jasmine
Jones I'm there JP university my
question about the okay this article
you can these things a little louder
fairly my question is about so the
epistemology not logical it using
making with the separation between
exactly in that it is it was wondering
if you have any thoughts on situations
that might be a mixture of both I'm
thinking about like people look right
where it was kind of this big exactness
issue but there is also this issue
quarantine which kind of yeah great
that's a great example I think I
thought about that one in just in
graduate Colorado for a little while
when you were an undergraduate yeah
yeah great nexus you can so the think
about ebola is that in other kinds of
had an X is that of course at other
button inhabited by people. So it
becomes this endogenous agent right so
it becomes like you know tracking who
is inhabiting the agents we after more
fries that so I would say that would
behave a lot like an endogenous intense
in in my opinion. But of course there's
still then I mean and there might be
you know there is just one single
answer to it to we might depending on
the way we're whatever research
question we have about this we might
need to play with the frame right about
whether it's important to think about
it for this research question as an
exhaustion agent for example versus
others as an endogenous agent I would
say I didn't do much social your
research on that but just kinda very
casually looking at anecdotally what
was going on. I would say that the kind
of behaviours we were seen in social
media we're we're very much more
aligned with and dodge in the US sorry
in the US we're very much aligned with
this kind of and dodging this way of
thinking around about the age of that
we have to act we handed watching you
know study and figure out who's
responsible for you know letting the
nurse walk out of the hospital or
something. So that would be my answer
that thank you thanks hi yeah enjoyed
your talk a lot john Thomas from
problem solving international. So I am
wondering what you think about global
climate change I realise that it's
happening over a long period of time
but maybe in terms of our ability to do
the things that are necessary in order
to impact the maybe it is a crisis and
maybe some of the things in crisis
informatics really need to be brought
to bear yeah you know so sure I think
yeah white of course and I I do you
know this is where the there's okay so
couple things there's there's always
struggle with the limits of english.
And crisis and disasters and
emergencies and what we call things and
how things and fall in or out of that
because we have to label them I think
four ish I think the attention that the
field is moving in is attending to the
outcomes of class climate change so
it's attending currently to the the
string of hurricanes and the damage
that it can do it's and we struggle
with how to deal with even things like
trout which are also protracted and
prolonged an outcome of climate change
that unclear how to what it I clear
that the social on computing frame is
the thing we should put two that we may
be looking at that differently we might
be looking at for example changing
behaviour right incentive icing
behaviour to do to behave differently I
would say at least in the war in in the
contributions that we've been able to
make the limit our work is does have
its limits for sure we have not looked
we have not tried to change behaviour
as a result of these kinds of threats
but I think that ability that does that
need to change behaviours we see in
other forms of each yeah I like people
becoming more green than exercising
more and all this that could absolutely
find a place here which is why so often
try to bring many people to the fold
here and try to describe a try to
explain the disaster is not an easy
problem it cuts across all aspects of
human society all aspects and and it's
a great place to be thinking about
these problems people don't have the
horizon is just so far out so I think
it's possible but I don't think is a
huge amount work being done in that
cooperative work even though it's
clearly improvised work right. It's not
planned work nobody was planning for
that's the lesson here is that we
simply will not be willing to hear once
again and for this context that
informal roles an informal work exists
and that they accomplish important
things such an understanding counteract
the mess of victim hood and other false
depictions of public conduct in
disaster especially those with
exhaustion as agents contrary to what
you see in the movies people are not
panicked they are not in a daze they
are not looting these are things that
get magnified hugely if we continue to
think about disasters as things that
need to be policed because people are
perceived as helpless and therefore
suddenly somehow a dangerous then we
are first of all doing a humane job
responding to disaster. And second all
this energy that we think about what we
think the problems are is being
misplaced. So if you ever gonna
disaster event yourself you will know
this terrible things could be happening
around you trouble things might be
happening to you but if you're not an
injured victim and maybe even if you
are you are still assessing the
situation you're still making decisions
you might be distraught you but you
still have your brain you are still
acting based on your ongoing analysis
of what's going on on around you you
never stop being smart right. So my
point is this if we believe the movie
version of disaster we bring those
ideas to social media front as well.
And it's another which than their place
of social convergence we magnify the
problems that we imagine must be there
and the more we do that the more we are
distracted by things like bad
information without really
understanding what it is we mean when
we say that information that it
deserves its own analysis of its own.
And we are characterising that very
well and therefore we can solve those
problems very well. If they're they're
even as we imagine them. Um and then of
course we miss this fascinating stuff
the version of this that happens
online. So our group is on a great deal
of work we have emergency managers in a
participatory fashion in various ways
over the years. And it's pretty clear
that the problem of adoption social
media for them is is bigger than most
imagine a lot of criticism is levelled
emergency managers for being slow and
not quick to to adopt new technology in
being afraid of new technology the
course problems are actually far more
complex than that there's really issues
of liability that they have to navigate
around. So how do they suddenly again
have to make decisions when they have
fewer resources the problems they have
and they turn social media can we can
they be sure word how when the state of
the art disliking so far behind them
them being able to get a situational
awareness kind of depiction of what's
going on. They have to be very cautious
about making decisions in relation to
it that's a very kind of simple
simplified way of looking at the array
of things that are facing but that's I
think a fair statement to say about
what the what the critical issue is but
I believe that this will come about
through change and it's going to come
about through change the practise so I
think of all the things that we have
looked that I think that the points the
the most fundamental point can be
boiled down to one takeaway lesson here
and that's that changes and changes in
social media policy and they do exist
even other partial we'll come about
through changes in local practise and I
have a fort we excerpt from hurricane
see any into doesn't it well as it was
rolling in and making landfall in the
in the US eastern seaboard that I think
brings this point home really
parsimonious lee. So this was a part of
a set of research that was presented by
Amanda Hughes and we think tennis at
high last year. So we have portraits
this is the first we just by the fire
department of new York these are we
produce exactly except we have
anonymous a little bit information as
agency. Um and FDNY says to announces
makes this declaration please note do
not tweak emergency calls please call
nine one one which is the emergency
line in the US if it's not an emergency
please call three one one and then some
text. And by the way has tags were born
out of disaster itself we credit
Christmas scene the who invented the
convention of the hash tag in the at in
the weight in the experience of the two
thousand seven southern California
wildfires that's another message and
all this is the disaster site of
innovation. So not even an hour later
about forty five minutes later somebody
writes directly to FTNY and says my
sister's family is at this address
water's rising twelve feet need help
phone number first floor drowned kids
it's care FCN vibrates back a couple
minutes later and says please keep
trying to call nine one one we see the
first hedge. I'd I will try to reach
just batteries now. So she's should
start their policies already starting
to you rode we're not even an hour in
CD about yeah and then many treats go
by there's lots of interaction I'm
showing you an excerpt here FT and my
breaks back not three hours after the
initial declaration and writes what
number of people and says don't want
new York street rely on this as an
alternative to nine one one button
notifying dispatchers of all
emergencies tweed and so this is the
point change happens in these so it's
your technical systems and all systems
really through bottom up processes
bottom up practise it's not top down
one so this is situated action situated
cognition in action and this is how
these things work. And this is how
we'll see change okay branch to the
biggest point of entry I'm moving along
because I have too much for this talk
so I actually actually probably see
that even a little bit more so the
biggest point of entry into the space
is this idea of how to drive leverage
leverage is a common word leverage and
drive data from the naturalistic leek
rings or some media streams that make a
disaster and it brings a lot of people
into this problem space even certainly
those two you may or may not care about
say social science commitments to the
problem space and there I am glad for
it actually because this is going to be
a multidisciplinary solution it's gonna
be a whole field solution to figure out
how to deal with two major constraints
for good data derivation and that's
true that is the optimisation for speed
and the optimisation for
comprehensiveness we currently candy
both that's the holy grail and so we
have to think in the meantime as we're
as we're trying to contribute to this
as others are trying to solve this
problem along with us and in the
application of other domains we ask
ourselves in our group what can we do
to drive data it's are still hopeful to
responders but that it's over a period
when things are happening but doesn't
get us into ethical dilemmas one we
can't be comprehensive about reporting
how many people are injured for example
or dealing with the absence of
information that we don't know yet know
how to quantify or qualify what that
means the absence of information could
mean somebody's impact really hurt not
that they can be heard right they could
in fact be really hurt okay so so one
area that we've done some work in this
area it was we try to apply this
compromise is in the area of
geotechnical reconnaissance audio
technical reconnaissance in so this is
what we do with civil engineers and
with the the department transportation
in the US and environmental design. So
do you think or common sense is spongy
gear teens geotechnical extreme event
reconnaissance teams to pull a two
disasters around the world or
international teams it appointed
disastrous internationally you know
five to six people to study the impacts
of hazards on the build infrastructure
naturalistic Lisa like us they want to
use the world is their natural
laboratory to get accurate accurate
rails of the hazard load on the
physical build environment. But the
mobilisation of international teams is
challenging it six pence who I mean
more than usual because there are no
flights into that area like truck
cancelled flights are but there are no
hotel rooms there are no rental cars
"'cause" they're all taken up that's
what happens disasters as massive
convergence of tons and tons of people.
So it's really hard to get into a
disaster site even if yeah I even if
it's not the transportation that's been
compromised it can sometimes be
dangerous so there was a team that was
going to planning to deploy to Japan
after these not me there but the risk
of radioactivity curtailed what they
were trying to do and it could be just
so easy to get wrong how do you deploy
team a five to six people to study the
isolated spots of so we'll look or
factions that occurs after seismic
activity when what you want to know you
wanna be able to see these places a
lick of actually to be able to see what
the effects are on the building
structures right now I will then have
future attacks on the building codes a
new forms of engineering of of our
infrastructure. So how do you how do
they find those points of the
confection along this eccentric longer
fault line so the idea is and also
adjustable my thing is the cleanup
begins right away even in most
devastating earthquakes and disasters
so so you can't take these measures if
you're not there right away. So the
idea is to use social media to help the
teams navigate within a highly diffuse
events in the geographical space
account having some intelligent
navigational tasks that then that also
than say by the way don't go that way
as you normally would you have to go
this way because the road is damage. Um
but also to use as a proxy for when you
can't be everywhere at once which which
can't so we tackle this using are only
for the okay so in Colorado in false
doesn't thirteen we had a massive
massive flash flooding we had an annual
a year's worth of rainfall in four
days. So we had flash flooding
everywhere and this is the kind of
topic as you might imagine that if you
really want to do meaningful work you
often have to be there and so alas it
came to us and we were able to conduct
a series of studies then that that then
that were much like this. So these
photos that and this but these two
photos initially you I'm illustrate the
kind of data thereafter so this is
water rushing down but turns out to be
a multi use path so you can see
obviously the before. And the after and
it I don't know if the resolution on
this projectors showing you but there's
just sheets of rain coming down so this
is the kind of data thereafter and so
what we did was we quickly extracted
videos and photos an overly those data
with floodplain data a satellite
imagery other kinds of data to use it
as a cases really have formative design
kind of study to imagine what we could
do to support this geotechnical
reconnaissance. Um the open circles
that you see here are known places a
bridge flooding and the closed dots are
that we that we were able to locate
nearby and with this we can retrieve
obviously the cup we content and verify
relative to other things that we see
like the satellite imagery but we also
learned that with your is whether who
that whatever's our right and that's
valuable because if that person is
located nearby we can contact them if
that person located nearby actually
this person turned out to live right at
that intersection of those two areas.
So we know this person is already
there. So we're not deploying new
people to the space so one temptations
is to say well why don't you just make
the crowd problem and ask people to
send you in reports of what they see
because you don't know tapping at this
intersection what we can do that
because if we don't know what's going
on we send people there we don't know
what they're encountering right and so
this is why we have to be very
concerned about how these even tiny
with seem like small interventions that
happening other crowd sourcing
situations have different applications
here. Um so anyway but we can find this
person he's already there I know to he
he's already there he might have other
photos another videos that you didn't
think to post right so that's more
data. So it's not just that use of the
whatever date that was a social media
data it's using it to find who you
could use for providing data and not
introduce more problems I also would
like to offer this as an example for
how you can think about testing
solutions that don't introduce
additional problems so when we're
thinking about how you extracted a
rapidly we can afford to get it wrong
for doing geotechnical reconnaissance
but we still are taking the people
cycles also have to be meaningful you
can just be a plate waiting right. But
we can still get it wrong if we don't
have a representative sample it's okay
we haven't presumably haven't heard
anybody. So a lot of people ask people
I'm doing technology for disaster how
do we test it all on the things you
might do with something like this is
testing in a problem that will let you
make Schwartz working properly whatever
that solution might be but not risking
human life and limb and there are other
things to say about that but I'm gonna
skip over that for now okay so but
building solutions to really help the
team situational awareness this is like
what I said as I said earlier the holy
grail and this is because of course
tweaks for example as you might know
are by and large on their own not
individually content for they're
they're not saying they're not there
but they they don't occur in wrap it in
large numbers and they're hard to find
and so this is the world's most perfect
we I think in in my opinion we founded
in two thousand and nine and it's still
my favourite sweet tutor was different
than a bit. So remember the metadata
was different. So this year is doing
some really clever things here. She we
get the time stamp and date stamp for
free. But she includes a link to it
will pick which is of this feature
which she tells us she confirms that
the red river in Winnipeg which was
under threat of flight. And it's north
of the university of Manitoba behind
the log building in university college
so it's really precise location
information which you could improvise
geo coding information I think provide
us I mean provide whoever's whomever
she doctor audience was okay and so we
get the who what where and this this a
a person you to confirm the way. So she
was doing recipient design. So she
wasn't present meaning that just
because she posted it on April a that
was taken on April eighth. So she was
really thinking about this is data
likewise it is going to be treated. She
didn't she presumed for audience should
not presume that it was taken. I
wouldn't know when it was taken so she
attaches that information to this to

Share this talk: 

Conference program

SIGCHI Social Impact Award Talk
Leysia Palen, University of Colorado Boulder, Boulder, United States
20 avr. 2015 · 2:34 après-midi

Recommended talks

Small Devices for Big Impact
Nuria Oliver, PhD, Scientific Director Telefonica Research
19 sep. 2013 · 3:05 après-midi
Design for Conversation
Henrik Vendelbo, The Digital Gap / Zurich, Switzerland
27 nov. 2016 · 2:30 après-midi