EMForum.org Presentation — October 8, 2008

Crisis Informatics
The Evolving Role of Information and Communication Technologies (ICT)

Colin Whitmore, EMT-B
Emergency Management Analyst
ICF International

Leysia Palen, Ph.D.
Assistant Professor, Computer Science
University of Colorado, Boulder

Amy Sebring
EMForum.org Moderator

The following version of the transcript has been edited for easier reading and comprehension. A text transcript is available from our archives. See our home page at http://www.emforum.org

[Welcome / Introduction]

Amy Sebring: Good morning/afternoon everyone. Welcome to EMForum.org! We are pleased you could join us today, including our first-timers. We want you to feel comfortable about participating, and we will be giving instructions as we go along.

Our topic today is "Crisis Informatics: The Evolving Role of Information and Communication Technologies." Please see the Background Page for links to related information and Websites. Our focus is the impact of the rapidly evolving social networking technologies in emergencies and disasters.

There is a related poll/survey question on our home page: "Where would the function of monitoring social networks go?: a) ESF #5, EM/Situational Awareness b) ESF #15, Public Information c) Other/Don't Know." Please take a moment after our session to respond if you have not voted already, and review the results to date.

Now it is my pleasure to introduce our guests. Colin Whitmore is an Emergency Management Analyst with ICF International. Previously he was a student at Virginia Tech where he was a member of the Rescue Squad February 2004 until May 2007. During that time he held the ranks of Vice President, President, and Lieutenant. As a Lieutenant he coordinated all aspects of training for the squad’s 40+ members. On April 16, 2007 Colin served as the EMS Commander for the tragic shootings on the Virginia Tech campus, coordinating the emergency medical response to Norris Hall.

Dr. Leysia Palen is an Assistant Professor of Computer Science at the University of Colorado, Boulder and a faculty fellow with the Institute for the Alliance of Technology, Learning and Society (ATLAS) and the Institute of Cognitive Science (ICS). Prior to her appointment at Colorado, she completed her Ph.D. at the University of California, Irvine in Information and Computer Science and her undergraduate education in Cognitive Science at the University of California, San Diego. Professor Palen, who partners with CU's Natural Hazards Center, was awarded a 2006 National Science Federation Early CAREER Grant to study information dissemination in disaster events.

Welcome to you both and thank you for being with us today. I now turn the floor over to Colin to start us off please.


Colin Whitmore: Thank you, Amy. I’m happy to have chance to talk about this topic with the people I think need to learn about it the most. I’ve given this presentation several times in person, but never in an online forum, so hopefully this is an opportunity to spread the message a little further.

On April 16, 2007, a student at Virginia Tech opened fire on students in West Ambler Johnston and Norris Halls. He took the lives of 32 students before taking his own. Just like any disaster, man-made or natural, people everywhere want information, many of them need information. If you were half-way across the country, you most likely heard what was happening in Blacksburg, VA from the anchor on CNN or in between sets of music on your car radio. As we know, however, these news sources take time to get their facts right. It would be hours before CNN was broadcasting the information that students on campus were exchanging in the same way they would communicate about any event.

This is our first lesson: people use the communications channels they are most comfortable and familiar with when disaster strikes -- for a growing demographic of people, those channels are Web-centric applications.

What makes the Virginia Tech incident a unique and powerful case study is that a disaster occurred in a community of people who are often the most active users of social networks and related applications. More interesting still, was that the lead EMS agency, Virginia Tech Rescue Squad, was comprised almost entirely of volunteer student EMTs and medics. Our command post was just as likely to watch Facebook for information as we were to turn on CNN. We were far more likely to exchange a text message with each other than to attempt a phone call. Some of our responders heard of the incident via text message -- a sort of SMS dispatch. Additionally, the police department at Virginia Tech is used to patrolling social networks for intelligence gathering and investigation and so they turned there as well. This is not a theory we’re dealing with -- it is the environment we live in and must learn to adapt to.

You’ve all probably heard the term "Web 2.0" thrown around lately. But what is Web 2.0? It is important to note that the term does not denote any upgrade of the hardware that makes the Internet work. What Web 2.0 refers to is the culture shift in the way we experience the Internet. It’s a bottom-up model where the value of web-based tools (Like Facebook, Twitter, Wikipedia, etc.) comes from the contributions of its users just like you and I. At Virginia Tech, students created groups on Facebook, wrote Wikipedia articles and edited the facts as they became clear, and exchanged information via blogs, MySpace, and text message.

The important concept here is that the information comes from an endless supply of sensors because everyone can contribute. Much the way a group of editors checks facts of a writer, users check facts supplied by other users. We begin to see patterns of emergency behavior where groups form to accomplish tasks and work cohesively to share up to date information.

Recent changes in the way we view emergency management force us to be in the intelligence and information business as emergency managers. Our responsibility to the people and property we protect is to provide them with the most accurate information available. If we intend to do that in an effective manner, we cannot ignore the segments of our population that rely on these non-traditional forms of communication. This is increasingly important as our populations age and more members of our community are less likely to turn on their televisions and more likely to check their iPhones for disaster information.

There are certainly challenges in moving toward a participatory role for emergency management in this environment. What are those challenges? What do you worry about as an emergency manager with regard to these information sources?

I have hinted at a participatory role for emergency management. I usually see a lot of people shifting in their chairs when I say that to a group. In fact, when I last looked at the poll on emform.org, 67% of respondents thought that social network monitoring was an ESF #15 Public Information function. I challenge that point.

Certainly, we need to learn to distribute information through these sources and we need to address rumors that may start in the same way we have always done with traditional media. However, if you were to watch users of Twitter during the recent wildfires in California, you would see that people were posting information about what they saw, heard, felt or needed. During Hurricane Ike users on the same site were discussing conditions in their areas. We have, at our disposal, an endless network on sensors that could provide us with a wealth of situational awareness. We have to learn to harness the power of that resource.

There is no denying that a primary concern for all of us is making sure that the information we gather from these sources is accurate. We are comfortable with the information we receive from traditional media because we accept that there is a certain degree of responsibility for accuracy that a major news network accepts as part of their jobs. We think of the Internet as this vast sea of information -- the wild west of information even. To the contrary, what we see is that, because these groups want information that is more accurate and timely than they can find on CNN, they self-police. We see a trend of individuals checking the facts of their peers and consensus-forming.

I’d like to caution each and every one of you from trying to address these sources in the same way we might normally adapt to change. Our tendency is to try and corral new sources of information. We control information because we respect the power that it has. We build levees to steer information where we want it to go -- and we do it effectively. However, these communities are not going to be corralled. We cannot, for instance, start our own emergency management social network and decide that we’ve addressed this concern.

Hundreds of new Web 2.0 applications are developed each year and only a very few of them last. Even fewer become as successful as Facebook and MySpace. The key here is that users will select the applications that work for them. What makes them work is an unpredictable combination of timing, popularity, features, and ease of use. We have to learn how to be current and interact with the community as it exists -- not as we wish it existed. This is undoubtedly our biggest challenge.

With that, I'd like to turn the floor over to Dr. Palen who is far more versed in these trends than myself, as she has been researching them for some time.

Leysia Palen: Greetings all. Thanks, Colin, for your terrific remarks, and to the EM Forum for the invitation to participate here.

My students and colleagues at the University of Colorado at Boulder have been conducting research in the area of "Crisis Informatics." The work is a union of information science, computer science and social science, areas of expertise that we bring together to understand the changing role of technology in society during times of emergency, disaster, and perhaps other forms of crisis.

The goal of Crisis Informatics is to understand the information- and technology- needs of all players in emergency situations. This includes both formal responders AND members of the public. This is our starting point. If we understand these needs, then we can better shape the development of new information technology, its deployment, and the policies that support and shape its use.

This is an area of research that is gaining ground and that we have worked extensively to build at CU with my colleagues at the Natural Hazards Center, Dr. Jeannette Sutton and Prof. Kathleen Tierney, and my hardworking PhD students Amanda Hughes, Sophia Liu, and Sarah Vieweg, Prof. Ken Anderson at CU, and my collaborator out of University of California, Irvine, Dr. Irina Shklovski.

There is other work to be aware of as well: Prof. Christine Hagar, of Dominican University is a library information scientist who did some even earlier work on "crisis informatics" during the early 2000s UK Farming Crisis in the early 2000s, and coined the term. Other researchers from other disciplines are doing some informative work in the broad area of disaster and information technology. For references to these other sources, I kindly refer you to the papers we have written on this topic.

At the heart of the research we are doing at CU, we connect knowledge of how people communicate using technology under normal conditions to explain how they do and will use technology under non-routine, emergency conditions. Because we are a society that increasingly uses information and communications technology (ICT) to manage our lives by connecting to information and people we need, it is quite natural and not surprising to know that people will turn to those same resources in disaster situations, and will do so at an increasing pace. Furthermore, we know that in times of disaster, pre-dating today's ICT, people turn to each other---family and friends--for help. People are natural information seekers, and ICT provides another means for them to find and provide information to each other. The difference is the scale and speed of these interactions.

In the Virginia Tech Shooting, students turned to each other to learn about what was going on---to make sense of the event on limited information (because information is always limited in times of crisis and emergency). After doing safety and welfare checks with family members, students, who were told to stay in their dorm rooms, turned to their computers to connect to their vast social networks. But they couldn't connect directly to everyone---people on university campuses casually know lots of people after all. Among other things, they turned to applications that would let them detect the on-line presence of their friends and acquaintances. This included instant messaging applications and social networking sites, like Facebook. There, students didn't have to directly talk to each other necessarily; at a minimum, they could figure out who was okay by who was on-line at the same time, or had been on-line recently.

This congregation of people on-line---and their ability to pool information, as well as to make inferences about their welfare---set the stage for a massive, international, highly socially distributed problem solving activity. Once the public knew how many people had been killed and injured, this was the impetus for members of the public to pool knowledge to figure out who those fatalities were. It sounds morbid and voyeuristic, but it wasn't. The listing of the victims was done with great respect, seriousness, and concern for accuracy. Lots of people wanted to help during and after the Virginia Tech event. Pooling information and doing 'research' of this sort was one way for people to help, even from a distance.

The results were remarkable. In just a few of the places that we were able to study, there were no inaccuracies on the lists. Furthermore, the lists were compiled in different orders, and no one list was complete. Yet across even our sample, all the names of fatality victims had been correctly deduced, mostly in advance of the official announcement from the university. This is proof of a kind of "collective intelligence" in the way that Roxanne Hiltz and Murray Turoff long ago predicted, where the public produces accurate information together, and without executive orchestration.

The results of the work do NOT have any kind of implications that would suggest that VT officials should have worked faster. Let me re-iterate that this is not the lesson. Rather, the conditions of this emergency were such that they created a "closed problem solving environment" that let us definitively see features of collective intelligence in action.

When we later turned our attention to the Southern California Fires one year ago, the highly diffuse nature of the event made it harder to point to tell-tale indications of collective intelligence in quite the same way. But critical information sharing by members of the public was present. This is something that is very difficult to quantify and empirically measure, so we make no claims to how much of this was happening during the event. But when we look at the some of the types of communications and information exchange that went on, we can see what the future will look like, and I think make some informed predications of what we should be thinking about for the future of emergency response.

In that event, the major take away lessons in this research I did with Irina Shklovski and Jeannette Sutton was that people will use the "backchannel" (the informal channel) to supplement "front channel" or formal communications. Those members of the public who are close to the action can come to serve as local experts-eyes on the ground--and can provide information that is often more accurate than what broadcast media can provide simply because of their previous local knowledge of the area, their personal investment in what is going to happen, and their ability to communicate that local knowledge with each other. They do this because it helps them cope. They want to help and they will help in this way.

Furthermore, they will often correct what they hear on big media. AND, importantly, even though information on the backchannel might also be incorrect, they will correct that too. What we have the potential for here in disaster response is a massive self-checking and correcting system, if we begin to think in these broader terms.

These are my brief remarks on a topic that I know many of you have a great deal of practical experience and wisdom about, and I look forward to discussing these with you today. I will turn the floor back over to our Moderator to start the discussion.

Amy Sebring: Thank you very much Colin and Leysia. Incidentally, you can find the research papers Dr. Palen referred to from a link on our Background Page for today's session. [See http://www.cs.colorado.edu/users/palen/new_papers.html]

Now, to proceed to your questions or comments. We especially want to hear your comments about any specific experience you may have had in this area.

[Audience Questions & Answers]

Hal Grieb: Along with keeping up to date with current Web 2.0 trends, we need to filter what we read and what we put out as far as information. ESF 5 is responsible for reviewing all information internal and external and making decisions based on that information. By using ESF 15 as not only a filter, but also a fast response update tool to the public, seems wise. My concern is what may become "cyber terrorism" through organized false reporting that we may view as credible during big unfolding incidents. What steps do you think could help filter for an accurate and timely picture?

Colin Whitmore: This is a fairly common concern - the idea of a sort of counter-intelligence terrorism attack, especially in the law enforcement community. I'd love to tell you that there is a magic answer to somehow "filter" for this sort of thing, but currently there isn't. In theory, however, I think there are two keys to addressing the issue:

First, how likely do we really think this is? The feasibility of a group of people organizing to disseminate information incorrectly as an act of terrorism without a greater number of ‘community’ members pointing out their wrong-doing seems highly improbable to me.

Second, I think this is a matter of deciding how we interact with the social networks on a case-by-case basis for each disaster. Obviously there is little impetus to disrupt hurricane response, and a far smaller chance of being successful in doing so. This would be in opposition to the chances that after a terrorist attack the perpetrators or sympathizers might attempt to spread the effects through counter-intelligence. I hope that addresses your questions somewhat.

Leysia Palen: I'll chime in just a bit, a couple of things: I am not convinced that the destination for publicly-generated information goes to the PIO function or the intelligence functions yet. I think we don't know the answer. I think that the massive changes of information dissemination in the public sphere might have even more significant changes in the formal organization of EM. Also, I think we tend first to think about cyber-terrorism as the primary concern, when I think the greater concern is that we miss out on all the valuable information that is NOT cyber terrorism.

And finally, this is a huge open question that we need to do research in. This is something we are trying to do at CU, where we are bringing together domain expertise like yours, social research, AND computer science techniques. The solutions are both social and technical -- social in the sense of both "human interaction as it works peer to peer" but also in our institutions and organizations that manage emergencies.

Jake Westfall: Talking with others in the EM community there is a lot of skepticism about Web 2.0 communication because of their funding models. They are either funded solely by venture capital that could eventually run out, or by advertising that may not be acceptable for your agency. Directly funding a communication system builds more confidence that you can depend on it. How can this concern be dealt with?

Colin Whitmore: I'm not precisely sure what you mean by skepticism, of what are you skeptical? I'd say that if you're skeptical about the reliability of the network, then you're looking at this from the wrong perspective. One primary key to adapting to these new modes of communication is that we need to be as flexible as they are. Colin Whitmore: If you want to use Web 2.0, then you have think like Web 2.0. Web 2.0 is bottom up, i.e., not funded by your friends at the local courthouse.

Leysia Palen: I concur with Colin, and I'll summarize by saying that what people are going to be using tomorrow to communicate will be different then in the next 6 months, and different again in 1 year, and so on. We need to prepare the organization of EM for this, not try to outsmart the technical solutions that are being created every day.

Amy Sebring: These next two questions are related, so let's combine them.

Isabel McCurdy: Colin, how does one ensure privacy of victim's rights in these mediums?

Matt Lane: What about the liability issues with socially driven information sharing, i.e., HIPAA issues, identity protection and so forth? Wouldn't there be a concern by the emergency management community of sharing personal information via these venues when typical lines of communication are hampered or no longer exist?

Colin Whitmore: We can only address that as EMs from the standpoint of how WE act in these communities. I don’t think there is any motivation for us to be discussing private aspects of a victim's situation. I've received the question many times and when I reply with "What would you like to post on Twitter that violates your citizen's privacy rights?" No one can give me an answer.

I think the concern comes from our involvement, and perhaps associated facilitation or approval of, in the way others share information online, and we don't want to be associated with the sharing of private information. That being said, the communications are already taking place and I don't read about privacy concerns today in the paper from victims of Hurricane Ike. I have no reason to think our participating would change that. Leysia, anything to add?

Leysia Palen: I'll just say that my student Sarah Vieweg and I looked at Facebook behavior at the Northern Illinois Shootings in February, to see how their behavior was different, if at all, from Virginia Tech, and what we found is that the precedent of VT guided behavior around what kind of information people thought was ok to share. So, many VT students told NIU students NOT to post victim's names. So again, we see some self-regulation here. But otherwise, yes, I concur with Colin that just because we participate in on-line environments for some kinds of information, it doesn't mean that we release publicly information about private citizen matters.

John Tommaney: Isn't there the danger of "information overload" using social networking in emergencies? So many different systems and users that you can't possibly wade through all the information (correct and incorrect) and stay on top of trends?

Leysia Palen: I think we are living in a world of information overload right now! I think that is a cusp that society is on right now, which is trying to figure out how to manage ALL this information that we could look at, even under normal conditions, but in times of emergency, this is a question I do hear about: How do emergency managers deal with information overload?

My answer is not a specific solution-based answer, but instead a perspective-based one, which is, that we know that members of the public share information with each other in times of crisis because THEY experience information dearth. : In other words, information overload and information dearth are two sides of the same coin, so what I think we are ultimately talking about here are information bottlenecks that are only made more ‘bottlenecky’ when we try to incorrectly control flow.

Colin Whitmore: I agree with Dr. Palen that it’s more an issue of perspective-shifting than buying a solution to manage the information. That's what I want to chime in on from the industry side. I caution against buying the latest greatest piece of software to manage all your social network accounts or intelligence monitoring programs. There are some great solutions out there, but you're really just trying to hold water in a burlap sack. By the time you learn your fancy new software, another new technology will exist that it cannot monitor. Just my two cents.

Rick Tobin: Has anyone evaluated what happens to the generation using the ad hoc systems when those sources of information aren't available for decision making? In other words, will they trust the other established information paths and respond appropriately without peer feedback?

Leysia Palen: Good question. I don't know of research on this point, but I think, to use Colin's water metaphor, what will happen is that the water flow will seek out the next best alternate path. I don't think it is the case that people don't trust established sources, it is that they ALSO trust sources that they might have established as trustworthy using other criteria. Even predating ICT, we know from the risk communication and warning empirical literature that people always checked with their family and friends to see if they were really going to evacuate, so I think people will endlessly pursue information, and get it wherever they can.

Amy Sebring: Two more related questions we can combine:

Gerald Smith: A few years ago our Province had a disaster. The Government website had 900,000 hits per day from people looking for information. Why not develop a section of this Government website to accept information from people? I like the idea of having many "ground sensors". Do you know of any negative implications of using a Government website? If we receive a vast quantity of information, how can this be processed in a timely manner in order to be beneficial?

Hal Grieb: Both of you seem to be against PIO run network sites. What do you feel the proper route should be, and why?

Colin Whitmore: Excellent questions. I think you should start your website that allows input from the public tomorrow if you can. But let me caution you, there are many agencies around the world that have done this and the most common complaint they receive is that no one responds to the messages! People sending you messages on your website might expect you to be as responsive as their peers would be in a social forum. You can't possibly do that during an emergency, but if you can at least have people who are assigned to addressing these concerns or letting people know that a real person reads them, you'll have better luck.

As for how to process the information, I don't know of a specific answer for that. I'd say its part of the disaster life-cycle -- not enough information becomes more than you can handle very quickly when you start seeking it.

As for PIO-run websites, I am not AGAINST a PIO-run website as you might gather, but I don't think it has anything to do with social network participation. Social networks are not going to be successfully created by the folks in Washington, DC. They are selected by the people who use them, and prosper based on their ability to meet needs and usability. The proper route to participation, in my view, is to understand that you cannot control this information but you can benefit from it. Then learn to participate just like every other community member, but use the information you can glean to help fulfill you role.

Leysia Palen: I think Colin answered this well--and let me say that I am also NOT against PIO-run sites. Not at all -- they have a place. It is just that people are looking for frequent updates, and will need robust sites (sites that stay up), so there need to be multiple places for people to go. People are always looking for authoritative information, but how they judge authoritative information might be calling their friend who stayed up on the evacuated mountain because he's got a cell phone and can drive around to see the neighborhoods.

Amy Sebring: We are running out of time and will not get to all pending issues, but I wanted to highlight that this is not just text messaging, but includes posting damage photos, collaborative mapping etc. Leysia, you recently looked at Flickr I believe; can you tell us just a little about that?

Leysia Palen: Yes, led by my student Sophia Liu. It is available on-line. It really gets at how these on-line environments are themselves adapting to be useful in some fashion to disaster response. For Flickr and other photo-sharing sites, this is all ‘in process' I'd say, but what we do see there is people trying to organize and find ways to help and participate in the event, even from afar.

I think the lesson of that study is to once again say that people want to participate in disaster events--it is a human need--so they'll do it in creative ways. Some of this doesn't have to be managed by emergency response; it is just there and speaks to this mass-collaboration phenomenon.

Amy Sebring: Colin, other types of collaboration or ways people get their information?

Colin Whitmore: I recently came across a fascinating diagram done by Fred Cavazza called "Social Media Landscape" which he has, ironically, posted to Flickr. http://www.flickr.com/photos/fredcavazza/2564571564/. I would recommend looking at that to get an idea of how vast this environment is...

We're not going to be able to ever sit and list all the possible ways for us to interact using these applications, and by the time we get a partial list aggregated, it’s outdated. I think the entire point here today is to draw attention to this. Much in the way that people's need for information on a large scale creates the motivation to create these tools, our motivation as a community of EMs and academics to understand this will eventually lead to innovative ways for us to participate.


Amy Sebring: Time to wrap for today. Thank you very much Colin and Leysia for an excellent job. And thanks to all our participants for great questions and comments. I am sorry we do not have time to get to all of them. Please stand by just a moment while we make a couple of quick announcements.

Again, the formatted transcript will be available later today. If you are not on our mailing list and would like to get notices of future sessions and availability of transcripts, just go to our home page to Subscribe.

Don't forget to vote in our poll, and PLEASE take a moment to do the rating/review! These folks deserve "excellent"!

We are planning an Open House event to try out Live Meeting, and more info will be coming out on our mailing list about that it in the near future. We hope you will participate, as we will need your help. Today is the International Day for Disaster Reduction, and this year's theme is "Hospitals Safe From Disasters." In honor of the campaign, our next regular program will be October 22nd, when the topic will be the PAHO/WHO Hospital Safety Index with Pat Bittner. We hope to be using Live Meeting for that session.

Thanks to everyone for participating today. We stand adjourned but before you go, please help me show our appreciation to both Colin and Leysia for an excellent job.

Colin Whitmore: Thank you very much for having us. If I can be of assistance to any of you, please feel free to contact me at [email protected]. Thanks!