WEBVTT Kind: captions Language: en-US 00:00:01.080 --> 00:00:12.200 [inaudible background conversations] 00:00:12.200 --> 00:00:16.970 Hi, everyone. Hi. We’re going to go ahead and get started. 00:00:16.970 --> 00:00:21.660 Thank you all for coming here. Just a few quick announcements. 00:00:21.660 --> 00:00:24.240 We will be having another seminar tomorrow. 00:00:24.240 --> 00:00:26.720 Our speaker is going to be Sam Johnson. 00:00:26.720 --> 00:00:29.240 He’s going to be talking about the northern San Andreas Fault. 00:00:29.250 --> 00:00:33.290 And this is a co-hosted seminar with GeoMAC, and it’s going to be at 11:00, 00:00:33.290 --> 00:00:37.840 not at 10:30, for anyone who wants to join us again tomorrow. 00:00:37.840 --> 00:00:41.820 All right. I would like to introduce our speaker for today. 00:00:41.820 --> 00:00:48.239 Gitanjali Bhattacharjee has a bachelor’s of science in architectural engineering 00:00:48.239 --> 00:00:52.910 and a bachelor’s degree in liberal arts from the University of … 00:00:52.910 --> 00:00:55.740 - Texas. - Texas. Yes. Thank you. 00:00:55.740 --> 00:00:59.060 And she’s got her master’s degree in structural engineering, and she has just 00:00:59.070 --> 00:01:03.699 finished her first year at Stanford at the Urban Resilience Group with 00:01:03.699 --> 00:01:08.409 Jack Baker, who spoke with us – who spoke here a few weeks ago. 00:01:08.409 --> 00:01:12.420 And, without, further ado, Gitanjali, please. Thank you. 00:01:17.020 --> 00:01:19.159 - Good morning, everyone. Thank you for having me. 00:01:19.159 --> 00:01:23.940 Thank you for the invitation to come talk about some work that I’ve done. 00:01:23.940 --> 00:01:27.690 So my main work actually focuses more on infrastructure management. 00:01:27.690 --> 00:01:30.740 This is just some exciting stuff that I’ve been lucky enough 00:01:30.740 --> 00:01:33.420 to get involved with at my time at Stanford. 00:01:33.420 --> 00:01:36.469 And specifically, I’ll be talking about developing a better understanding 00:01:36.469 --> 00:01:41.000 of post-earthquake building damage information needs and use. 00:01:42.160 --> 00:01:45.860 So, to give you some context for how this work came about, 00:01:45.860 --> 00:01:49.500 I got involved with some research in crowdsourcing building damage 00:01:49.500 --> 00:01:53.439 assessments after earthquakes. And, as part of that larger research 00:01:53.439 --> 00:01:56.299 effort, we talked a lot about previous efforts to crowdsource 00:01:56.299 --> 00:01:59.530 building damage assessments, and so we talked a lot about 00:01:59.530 --> 00:02:02.540 the response to the 2010 earthquake in Haiti. 00:02:02.540 --> 00:02:05.040 And what was notable – well, there were many notable things about 00:02:05.049 --> 00:02:09.979 that earthquake, but GEO-CAN was initialized specifically after that 00:02:09.979 --> 00:02:13.400 earthquake to crowdsource building damage assessments. 00:02:13.400 --> 00:02:16.310 And the crowd in that case was a group of expert volunteers. They were 00:02:16.310 --> 00:02:21.950 structural engineers, people with lots of relevant experience to the task at hand. 00:02:21.950 --> 00:02:24.830 And they were looking at satellite images and trying to assess the 00:02:24.830 --> 00:02:28.810 damage to individual buildings. And the whole goal of this massive 00:02:28.810 --> 00:02:32.640 GEO-CAN effort was to inform a very specific document called 00:02:32.640 --> 00:02:37.560 the Post-Disaster Needs Assessment, or PDNA for short. 00:02:37.560 --> 00:02:41.430 And the Post-Disaster Needs Assessment is a document that the government of 00:02:41.430 --> 00:02:47.819 the affected country will put together after a disaster in order to, first of all, 00:02:47.819 --> 00:02:54.130 take stock of the losses from a disaster to plan for recovery and reconstruction, 00:02:54.130 --> 00:02:57.260 and very crucially, it’s the basis upon which they make requests for 00:02:57.260 --> 00:03:01.620 financial assistance to other governments around the world. 00:03:02.300 --> 00:03:05.300 So GEO-CAN was really trying to inform 00:03:05.300 --> 00:03:09.140 the loss estimates that were going to be in the PDNA. 00:03:10.400 --> 00:03:14.080 And if you look at what actually gets tabulated in the PDNA, 00:03:14.080 --> 00:03:17.640 it looks something like this. So this table is actually from the 00:03:17.640 --> 00:03:22.060 2010 Haiti earthquake PDNA. And you can see it’s pretty simple. 00:03:22.070 --> 00:03:26.670 You have a number of damaged houses, the corresponding dollar value, 00:03:26.670 --> 00:03:30.860 a number of destroyed houses, the corresponding dollar value. 00:03:30.860 --> 00:03:36.260 And, at the end, you come up with a number – about $2.3 billion of damage. 00:03:37.300 --> 00:03:41.140 When we were talking to people involved in putting this document together, 00:03:41.150 --> 00:03:44.550 they really emphasized that they just needed broad regional estimates 00:03:44.550 --> 00:03:49.040 of building damage. They didn’t need individual building damage assessments, 00:03:49.040 --> 00:03:52.340 which is what GEO-CAN had been working so hard to provide. 00:03:53.180 --> 00:03:57.960 And so we saw that there was a pretty clear mismatch between the enormous 00:03:57.960 --> 00:04:01.870 very generous effort that these volunteers in GEO-CAN were 00:04:01.870 --> 00:04:06.800 putting forth and the main end user’s information needs. 00:04:08.140 --> 00:04:11.760 And so, as a group, we started thinking that there’s been a lot of work – 00:04:11.760 --> 00:04:16.200 very valuable, exciting, and worthwhile work done to advance building damage 00:04:16.200 --> 00:04:20.759 assessment techniques, but we have a comparatively limited understanding 00:04:20.760 --> 00:04:26.620 of how responders actually use and need building damage information. 00:04:26.620 --> 00:04:29.920 And why does that matter? Well, first of all, it limits the ability 00:04:29.930 --> 00:04:34.439 of organizations to respond effectively to disaster. 00:04:34.439 --> 00:04:37.930 Not having a common understanding of information needs can also limit 00:04:37.930 --> 00:04:42.220 the ability of organizations to effectively share information. 00:04:42.220 --> 00:04:45.039 And, in a broader sense, if you think about the bigger 00:04:45.039 --> 00:04:48.039 disaster community, it hampers the ability of 00:04:48.040 --> 00:04:52.480 the research community to address practitioners’ needs. 00:04:53.700 --> 00:04:57.439 One of the people that we talked to gave us a really concrete example 00:04:57.439 --> 00:05:01.220 of why having an understanding of information needs is important. 00:05:01.220 --> 00:05:04.860 She said, when damage assessment was ongoing on the ground after 00:05:04.860 --> 00:05:09.961 this 2014 event, there were apparently 14 different surveys. I mean, they were 00:05:09.961 --> 00:05:13.560 the same surveys, asking the same question, doing the same damage 00:05:13.560 --> 00:05:19.020 assessment on the same house, but 14 times by different organizations. 00:05:19.020 --> 00:05:24.440 That’s an incredible amount of effort to get at one thing. 00:05:26.720 --> 00:05:30.610 So what makes understanding information needs difficult in the first place? 00:05:30.610 --> 00:05:35.279 I think perhaps the first two bullet points are very self-explanatory, 00:05:35.279 --> 00:05:39.710 but maybe they’re worth repeating. Post-disaster contexts are very chaotic. 00:05:39.710 --> 00:05:41.949 And no two disaster responses are the same. 00:05:41.949 --> 00:05:44.620 So it’s very hard to come up with hard-and-fast rules about 00:05:44.620 --> 00:05:49.419 who needs what, when, and how. But previous work in trying to 00:05:49.419 --> 00:05:53.689 understand information needs has also found that, even if you were able to get 00:05:53.689 --> 00:05:59.490 expert practitioners in a room together talking about their work, they find it 00:05:59.490 --> 00:06:04.200 very difficult to actually enumerate the information that they use and need. 00:06:05.930 --> 00:06:12.040 So more recently, some researchers have called to shift the focus from 00:06:12.060 --> 00:06:17.940 information needs to how information informs specific activities or tasks. 00:06:17.940 --> 00:06:21.860 So, in one paper, Tapia and Moore wrote, while improved information 00:06:21.870 --> 00:06:27.840 quality and sharing are noble goals, the real aim is to improve relief services. 00:06:27.840 --> 00:06:33.199 In another paper, Lallemant et al. wrote, damage assessments are only relevant 00:06:33.199 --> 00:06:38.800 in their relation to the response and recovery activities that they enable. 00:06:38.800 --> 00:06:42.780 And so we, as a group, were thinking, how can we reframe our question 00:06:42.780 --> 00:06:45.090 so that it’s more tractable to answer? 00:06:45.090 --> 00:06:49.150 And one way we thought we could do that is to ask, how does building 00:06:49.150 --> 00:06:53.960 damage information enable disaster responders to carry out their work? 00:06:55.680 --> 00:07:00.500 So the objectives of the work that I’m going to talk about today are as follows. 00:07:00.500 --> 00:07:03.979 First of all, we simply wanted to identify post-disaster tasks 00:07:03.980 --> 00:07:07.000 that rely on building damage information. 00:07:07.000 --> 00:07:11.260 We’re certainly not trying to identify all of them, but at least make a start. 00:07:12.430 --> 00:07:16.000 Then our second objective was to describe each task and to do so 00:07:16.000 --> 00:07:20.069 according to the people involved in carrying it out according to the 00:07:20.069 --> 00:07:23.809 particular building damage information that they require. 00:07:23.809 --> 00:07:26.469 And also according to the characteristics of the building 00:07:26.469 --> 00:07:30.720 damage information that make it actionable for those users. 00:07:31.520 --> 00:07:35.400 So the first half of this talk is really going to focus on the first two bullet 00:07:35.400 --> 00:07:40.050 points, which are information needs. And then, in the second half of the talk, 00:07:40.050 --> 00:07:45.919 I’m going to focus on information uses and articulate how organizations 00:07:45.920 --> 00:07:50.240 actually use the building damage information that they say they need. 00:07:51.640 --> 00:07:55.940 So, to give you an idea of our research design, it was pretty straightforward. 00:07:55.949 --> 00:07:59.639 The first part involved doing a survey of expert practitioners. 00:07:59.639 --> 00:08:02.970 And I’ll give you a better idea of who those are in a second. 00:08:02.970 --> 00:08:06.279 To do that survey, we first did 11 in-depth interviews 00:08:06.279 --> 00:08:08.639 with those expert practitioners. 00:08:08.639 --> 00:08:13.199 We also disseminated a short online survey that got 12 responses. 00:08:13.199 --> 00:08:16.089 And then, in the second step, we wanted to develop a framework 00:08:16.089 --> 00:08:19.229 that would help us understand and contextualize responders’ 00:08:19.229 --> 00:08:21.720 building damage information needs. 00:08:23.100 --> 00:08:26.520 So, to give you a sense of the people that we were talking to, they worked 00:08:26.520 --> 00:08:31.069 in quite a few different capacities. Those included disaster risk 00:08:31.069 --> 00:08:34.780 management reduction and financing, urban search and rescue, 00:08:34.780 --> 00:08:38.220 emergency mapping, disaster loss estimation, 00:08:38.229 --> 00:08:41.550 and also information management and GIS support. 00:08:41.550 --> 00:08:45.269 They worked at organizations including the United Nations Office for the 00:08:45.269 --> 00:08:48.610 Coordination of Humanitarian Affairs, the American Red Cross, 00:08:48.610 --> 00:08:52.199 the World Bank, Pacific Community, ITHACA, the Humanitarian 00:08:52.199 --> 00:08:55.639 OpenStreetMap Team, and then we had a couple representatives 00:08:55.639 --> 00:08:59.699 from urban search and rescue organizations from the Netherlands 00:08:59.699 --> 00:09:03.820 and also from the International Search and Rescue Advisory Group, 00:09:03.820 --> 00:09:06.360 which is part of the U.N. 00:09:09.060 --> 00:09:12.040 They also had varied backgrounds – seismology, 00:09:12.040 --> 00:09:14.880 earthquake structural engineering, economics, geomatics, 00:09:14.880 --> 00:09:18.180 marine science, and hazard and risk assessment. 00:09:21.170 --> 00:09:26.500 In terms of their practical experience, seven of our 11 interviewees had 00:09:26.519 --> 00:09:30.699 been involved in the response to the 2015 earthquake in Nepal. 00:09:30.699 --> 00:09:36.610 And five of the 11 had been involved in the 2010 Haiti earthquake response. 00:09:36.610 --> 00:09:42.519 And then, smaller numbers had been involved in earthquakes and other 00:09:42.519 --> 00:09:46.270 disasters going back to the 2004 Al Hoceima earthquake in 00:09:46.270 --> 00:09:52.140 Morocco and most recently in the Amatrice earthquakes in Italy, 2016. 00:09:54.560 --> 00:09:58.260 So the interviews that we conducted were semi-structured. 00:09:58.260 --> 00:10:02.800 First we wanted to get a sense of who the people are that we were talking to. 00:10:02.800 --> 00:10:06.720 So, for example, we might ask, what were your major responsibilities? 00:10:06.720 --> 00:10:09.260 Can you give me examples of the types of work that you 00:10:09.260 --> 00:10:12.200 were doing, and who were you working with? 00:10:12.880 --> 00:10:16.060 But we also really wanted to drill down into the nitty-gritty of the particular 00:10:16.069 --> 00:10:21.190 building damage information they were using as the basis for decisions. 00:10:21.190 --> 00:10:25.529 We also wanted to know, if they didn’t get to use building damage data, 00:10:25.529 --> 00:10:28.110 would they have liked to, and how would that information 00:10:28.110 --> 00:10:30.540 have been useful for their work? 00:10:30.540 --> 00:10:36.300 So these semi-structured interviews, we used these questions as a guiding basis. 00:10:36.300 --> 00:10:40.060 But we really let the conversation go where they directed it. 00:10:41.630 --> 00:10:45.940 So our interviews indicated six particular tasks that rely heavily 00:10:45.940 --> 00:10:50.260 on building damage information. They are urban search and rescue 00:10:50.260 --> 00:10:55.760 operations, which focus on identifying live victims of 00:10:55.760 --> 00:11:00.750 building collapse and extracting them from those buildings. 00:11:00.750 --> 00:11:05.649 Then the U.N. Office for the Coordination of Humanitarian Affairs 00:11:05.649 --> 00:11:10.230 will put together a Multi-cluster Initial Rapid Assessment, called the MIRA. 00:11:10.230 --> 00:11:13.910 And that happens soon after a disaster. 00:11:13.910 --> 00:11:17.230 The World Bank will also put together a Rapid Impact Assessment, 00:11:17.230 --> 00:11:20.040 which is primarily for their internal use. 00:11:20.940 --> 00:11:23.480 There’s, of course, the Post-Disaster Needs Assessment, 00:11:23.480 --> 00:11:27.720 which I described earlier, that the government of the affected country 00:11:27.720 --> 00:11:30.240 will put together in partnership with the 00:11:30.240 --> 00:11:33.130 World Bank and other development partners. 00:11:34.060 --> 00:11:37.040 Then, in the longer term, the government of the affected country may 00:11:37.050 --> 00:11:42.069 offer some housing recovery support to people affected by the disaster. 00:11:42.069 --> 00:11:45.699 And then, even longer term, different groups may be interested in developing 00:11:45.700 --> 00:11:51.300 and calibrating fragility or loss functions using data from a particular event. 00:11:53.580 --> 00:11:58.720 Now, we found these tasks, and listing them is fine, but we really wanted to 00:11:58.720 --> 00:12:03.860 contextualize them and give them some relationships to each other. 00:12:03.860 --> 00:12:08.300 So we came up with this very simple framework that I will describe here. 00:12:08.300 --> 00:12:11.940 It’s basically a plot, and on the X axis, we’re looking at the time elapsed 00:12:11.949 --> 00:12:16.370 since the disaster. On the Y axis, we have the operational 00:12:16.370 --> 00:12:21.080 spatial resolution of the underlying building damage information. 00:12:21.080 --> 00:12:24.450 By operational spatial resolution, I just mean the minimum 00:12:24.450 --> 00:12:28.990 required spatial resolution for that information to be useful 00:12:28.990 --> 00:12:31.400 to the people carrying out the task. 00:12:32.400 --> 00:12:34.580 So we’re looking at time going from response 00:12:34.580 --> 00:12:38.380 all the way through recovery and reconstruction. 00:12:38.380 --> 00:12:43.600 And then, in terms of spatial resolutions, sort of the least precise would be 00:12:43.600 --> 00:12:48.860 having aggregated information of that building damage at the national level 00:12:48.860 --> 00:12:54.220 all the way down to really detailed building-level damage information. 00:12:54.220 --> 00:12:59.550 And so we can start to situate the tasks that we found in this framework. 00:12:59.550 --> 00:13:04.029 So urban search and rescue operations happen – they can start happening 00:13:04.029 --> 00:13:08.460 within six hours of an event anywhere in the world. 00:13:08.460 --> 00:13:13.040 And those operations continue, at maximum, to about 10 days 00:13:13.040 --> 00:13:18.319 after the onset of the event. And they really rely on very detailed 00:13:18.319 --> 00:13:22.220 building-level damage information. They need to know construction 00:13:22.220 --> 00:13:27.210 details that might be specific to the particular country that they’re in. 00:13:27.210 --> 00:13:31.860 And that may affect the likelihood of people surviving if buildings collapse. 00:13:32.960 --> 00:13:35.720 At the same time, the World Bank will start putting together its 00:13:35.720 --> 00:13:40.959 Rapid Impact Assessment, which is sort of a coarse first-order 00:13:40.960 --> 00:13:45.240 look at the losses that come out of a particular event. 00:13:46.040 --> 00:13:48.240 And that’s primarily for internal use but is also 00:13:48.260 --> 00:13:51.440 used to check government statistics. 00:13:52.980 --> 00:13:59.240 Then, as I mentioned, the MIRA, or the Multi-cluster Initial Rapid Assessment, 00:13:59.250 --> 00:14:03.220 is put together at the same time as these other assessments. 00:14:03.220 --> 00:14:05.329 And it really focuses on humanitarian needs. 00:14:05.329 --> 00:14:08.339 So what are the food, water, and shelter needs of people 00:14:08.339 --> 00:14:11.430 who have been affected by an event? 00:14:11.430 --> 00:14:16.079 And they really rely on city-level building damage information. 00:14:16.079 --> 00:14:18.149 In contrast to the Rapid Impact Assessment, 00:14:18.149 --> 00:14:21.420 which only really needs regional information. 00:14:22.040 --> 00:14:25.800 As we move through the recovery timeline, the PDNA starts about 00:14:25.800 --> 00:14:32.060 two months after an earthquake or other disaster. 00:14:32.060 --> 00:14:34.779 And that’s put together by the national government in partnership with 00:14:34.779 --> 00:14:38.829 many development partners. And, again, they rely on regional 00:14:38.829 --> 00:14:42.900 building-level – sorry – regional building damage information. 00:14:44.240 --> 00:14:49.040 And then, as we go on, housing recovery support that might be offered 00:14:49.040 --> 00:14:51.910 by the national government and other development partners 00:14:51.910 --> 00:14:54.709 will require really detailed information about the damage 00:14:54.709 --> 00:14:58.480 to specific buildings as well as property ownership. 00:15:00.040 --> 00:15:03.860 And, to calibrate fragility and loss functions, you might even 00:15:03.860 --> 00:15:06.750 need component-level damage information 00:15:06.750 --> 00:15:10.080 that’s more detailed than building-level damage information. 00:15:11.780 --> 00:15:15.960 So now that we’ve plotted all of these tasks that rely on building damage 00:15:15.960 --> 00:15:20.240 information, what does that help us see that we couldn’t see before? 00:15:20.240 --> 00:15:25.200 Well, first of all, you’ve noticed that the tails on these different numbered 00:15:25.200 --> 00:15:31.660 bubbles have different lengths, and that indicates the task duration. 00:15:31.660 --> 00:15:37.399 So all of these tasks take some amount of time, and different information 00:15:37.399 --> 00:15:40.560 will be useful to different people at different times. 00:15:40.560 --> 00:15:45.959 So this helps us see, for example, if you come into – if you happen to 00:15:45.959 --> 00:15:49.779 have some regional damage information, 00:15:49.780 --> 00:15:53.470 it might be useful to different people at different times. 00:15:54.380 --> 00:15:57.279 We can also see that there are many opportunities for information 00:15:57.280 --> 00:16:01.460 sharing that might not have been taken advantage of. 00:16:01.460 --> 00:16:06.500 So different groups may need the same building-level damage information – 00:16:06.509 --> 00:16:10.709 so information gathered earlier by urban search and rescue teams 00:16:10.709 --> 00:16:15.329 might be useful to housing recovery support and calibrating fragility 00:16:15.329 --> 00:16:19.220 parameters in later stages of the recovery timeline. 00:16:20.360 --> 00:16:25.139 And we can also see that there are really two spatial precisions at which 00:16:25.139 --> 00:16:28.690 building damage information is needed, primarily. 00:16:28.690 --> 00:16:32.639 There’s highly detailed building-level damage information that’s required for 00:16:32.639 --> 00:16:38.189 different operations, and then there’s regional estimates of building damage. 00:16:38.189 --> 00:16:44.060 Anything in between is less useful to different groups. 00:16:45.300 --> 00:16:48.880 We can also get a sense of who’s involved in all of these different tasks 00:16:48.889 --> 00:16:52.439 that rely on building damage information. 00:16:52.439 --> 00:16:56.810 And what becomes clear very quickly is that the national government happens 00:16:56.810 --> 00:17:01.569 to be involved in nearly all of these activities that have to do with buildings. 00:17:01.569 --> 00:17:05.430 And so it’s important, as other researchers have pointed out, to support 00:17:05.430 --> 00:17:10.300 them getting back up to capacity if they’re affected by the disaster as well. 00:17:12.000 --> 00:17:16.459 So what we can see from this first part on responders’ information needs is that 00:17:16.459 --> 00:17:21.569 building damage information needs among responders vary by operational 00:17:21.569 --> 00:17:26.380 spatial precision and also by the time at which information is required. 00:17:26.380 --> 00:17:30.360 But by doing this work to link building damage information to the responders 00:17:30.360 --> 00:17:34.539 and the tasks that it supports, we can highlight strategic opportunities 00:17:34.540 --> 00:17:38.860 for research for information sharing and other partnerships. 00:17:40.120 --> 00:17:44.260 We also a Blume Report that includes more information on this if you’re 00:17:44.260 --> 00:17:49.550 interested. And, of course, I’m happy to take questions about that as well. 00:17:49.550 --> 00:17:54.840 We also wanted to look beyond frameworks to information use. 00:17:54.840 --> 00:17:57.620 So what happened after we came up with this framework and analyzed 00:17:57.620 --> 00:18:01.470 what we could from it, is that we saw it really captures the availability 00:18:01.470 --> 00:18:05.720 and the suitability of information for users’ needs. 00:18:06.580 --> 00:18:09.740 But when we went back and looked at all the conversations we had with 00:18:09.750 --> 00:18:14.330 practitioners, they had so many stories that seemed to suggest there are more 00:18:14.330 --> 00:18:19.980 factors that influence whether or not they use building damage information. 00:18:21.809 --> 00:18:26.300 And so we wanted to take account of those stories and anecdotes that they 00:18:26.310 --> 00:18:31.020 had provided, and so we conducted a qualitative analysis of the 00:18:31.020 --> 00:18:34.780 interviews and the survey data to better understand responders’ 00:18:34.780 --> 00:18:39.060 information use as distinct from their information needs. 00:18:40.440 --> 00:18:46.340 And in doing that qualitative analysis, we identified some other factors that 00:18:46.340 --> 00:18:51.730 tend to influence responders’ use of building damage information. 00:18:51.730 --> 00:18:55.070 And these are impediments to information sharing, 00:18:55.070 --> 00:19:00.120 implications of information adoption, varying understandings of disasters, 00:19:00.120 --> 00:19:03.400 and their attitudes toward emerging technologies. 00:19:05.140 --> 00:19:10.460 So I want to address each of these and give you a sense of what they mean just 00:19:10.470 --> 00:19:15.990 by sharing some of the anecdotes that we gathered from expert practitioners. 00:19:15.990 --> 00:19:19.071 So when it comes to impediments to information sharing, we found that there 00:19:19.071 --> 00:19:25.910 are really three issues that were present in our responders’ experiences. 00:19:25.910 --> 00:19:29.770 And those were issues of semantic interoperability, 00:19:29.770 --> 00:19:33.220 systemic disincentives to sharing information, 00:19:33.220 --> 00:19:38.280 and a lack of feedback from information users to information producers. 00:19:39.909 --> 00:19:45.440 So semantic interoperability we’ll define as the ability to share information 00:19:45.450 --> 00:19:48.890 without creating ambiguities in what it means. 00:19:48.890 --> 00:19:52.240 And something that happens after an earthquake, for example, 00:19:52.240 --> 00:19:57.080 is that different responders will tag buildings according to different 00:19:57.080 --> 00:20:01.390 schemes to indicate their status. So urban search and rescue teams 00:20:01.390 --> 00:20:06.960 will tag buildings to indicate whether they’re safe for urban search and 00:20:06.960 --> 00:20:11.630 rescuers to enter to rescue people. And building damage assessment teams 00:20:11.630 --> 00:20:16.460 will tag buildings to indicate whether they’re safe to reoccupy. 00:20:16.460 --> 00:20:22.659 Something that one of our interviewees mentioned in Haiti was that there was 00:20:22.659 --> 00:20:26.510 a misinterpretation, and that was that the color coding for safety 00:20:26.510 --> 00:20:31.710 for the purpose of rescue was interpreted as the color coding for viability. 00:20:31.710 --> 00:20:37.520 Those things are, in fact, very different and should not be confused. 00:20:39.460 --> 00:20:42.780 Another very interesting anecdote we heard from the urban search 00:20:42.789 --> 00:20:50.240 and rescue team leaders that we talked to is about systemic disincentives 00:20:50.240 --> 00:20:56.130 to sharing information. So, after an earthquake or other disaster happens, 00:20:56.130 --> 00:21:00.240 multiple urban search and rescue teams may deploy to the affected region. 00:21:00.240 --> 00:21:04.080 And each of them is designed to be a self-sufficient unit. 00:21:04.090 --> 00:21:08.620 And so they have to have a set of detailed coordination guidelines. 00:21:08.620 --> 00:21:12.260 And they do. A key part of those coordination guidelines 00:21:12.260 --> 00:21:15.570 is tagging buildings. So here at the bottom of the slide, 00:21:15.570 --> 00:21:19.520 I’ve included a couple of examples of what those tags might look like. 00:21:19.520 --> 00:21:24.820 So, on the left, you can see they’re in fluorescent ink. 00:21:24.820 --> 00:21:28.250 They indicate some dates when work is happening. 00:21:28.250 --> 00:21:31.270 At the top, they might indicate some hazards. 00:21:31.270 --> 00:21:35.100 On the left, we have asbestos. On the right, we’ve got chemicals, 00:21:35.100 --> 00:21:41.260 gases, and rats. So that’s important for them – for their teams to know. 00:21:41.260 --> 00:21:46.380 And if you look at the tag on the right, at the bottom, you’ll see a count. 00:21:46.380 --> 00:21:50.190 It says, 12, question mark, two dead in elevator. 00:21:50.190 --> 00:21:55.370 And so teams are supposed to note the potential for rescuing 00:21:55.370 --> 00:21:58.280 live victims from buildings. 00:21:59.860 --> 00:22:03.900 And urban search and rescue teams also have multiple other information-sharing 00:22:03.900 --> 00:22:06.980 mechanisms. They have a virtual coordination platform. 00:22:06.980 --> 00:22:10.070 They have emails. They share paper forms. 00:22:10.070 --> 00:22:12.570 They do phone calls. And they have coordination meetings. 00:22:12.570 --> 00:22:17.070 So there’s no shortage of mechanisms for them to share information. 00:22:17.070 --> 00:22:21.720 But when we were talking with a USAR team leader, he listed all of those 00:22:21.720 --> 00:22:25.390 coordination and information-sharing mechanisms, and then he sort of took 00:22:25.390 --> 00:22:28.640 a step back and said, well, there’s another one, and that is, 00:22:28.640 --> 00:22:31.559 you don’t share information. And, of course, we were curious. 00:22:31.559 --> 00:22:35.770 Why would you not share information if there are lives potentially at risk? 00:22:35.770 --> 00:22:40.210 And he sort of went on to say, because information is valuable. 00:22:40.210 --> 00:22:43.140 The moment you go to another country and you try to save people, 00:22:43.140 --> 00:22:45.820 it’s very important that you save somebody because it’s an 00:22:45.820 --> 00:22:50.780 operation that will cost a lot of money. And we all have sponsors. 00:22:50.780 --> 00:22:54.720 So he wanted to explain that there are teams who will occasionally 00:22:54.730 --> 00:23:00.799 not tag buildings or share information about locations of potential victims 00:23:00.799 --> 00:23:05.100 or locations where there have been confirmed fatalities because they 00:23:05.100 --> 00:23:10.360 don’t want other teams to get a head start in finding live people. 00:23:10.360 --> 00:23:13.549 And that’s because there are external pressures to do with 00:23:13.549 --> 00:23:19.700 how these teams are funded that disincentivize information sharing. 00:23:20.440 --> 00:23:23.920 Even though they have so many coordination mechanisms. 00:23:26.200 --> 00:23:30.399 A less dramatic example, perhaps, is just having a lack of feedback 00:23:30.399 --> 00:23:33.920 from the people using your information. 00:23:33.920 --> 00:23:37.090 So we talked to people who do emergency mapping in the wake 00:23:37.090 --> 00:23:41.919 of disasters. And one of them noted, well, there’s a mechanism of a feedback 00:23:41.920 --> 00:23:45.820 form in the Copernicus Emergency Mapping Service. 00:23:45.820 --> 00:23:50.380 But it’s quite difficult to have details in how they actually use that analysis. 00:23:50.380 --> 00:23:53.670 There’s no direct relationship between the people who are doing the 00:23:53.670 --> 00:23:57.550 emergency mapping and the people using the emergency maps. 00:23:57.550 --> 00:23:59.929 And this lack of feedback was something he cited as 00:23:59.929 --> 00:24:03.080 frustrating and that could inhibit improvements to the 00:24:03.080 --> 00:24:06.220 production of building damage information. 00:24:08.980 --> 00:24:13.480 Another theme that we noticed is that information adoption has certain 00:24:13.490 --> 00:24:18.590 implications. Using information is not neutral, in other words. 00:24:18.590 --> 00:24:22.490 So if you choose to use a piece of information, in some situations, 00:24:22.490 --> 00:24:26.880 you’re actually sending a signal of trust to the person who’s providing 00:24:26.880 --> 00:24:31.429 you that information – a signal that you trust their version of events 00:24:31.429 --> 00:24:33.740 and you trust what they have to say. 00:24:33.740 --> 00:24:39.200 And, in some cases, you might want, as a responder, to send that signal. 00:24:39.200 --> 00:24:43.100 You might want to establish or maintain or even build a good 00:24:43.100 --> 00:24:47.420 working relationship with the people who are giving you information. 00:24:47.420 --> 00:24:54.420 And we sort of noticed this because of two very different behaviors 00:24:54.429 --> 00:24:59.820 that came up in our interviews. One was by members of a World Bank 00:24:59.820 --> 00:25:02.580 team who are working with the government of the affected country 00:25:02.580 --> 00:25:09.019 long-term. So they work together on the PDNA and then on efforts to recover 00:25:09.020 --> 00:25:13.820 and reconstruct in the affected country. And then we also noticed a very 00:25:13.820 --> 00:25:16.980 different behavior from urban search and rescue teams. 00:25:16.980 --> 00:25:20.600 Now, they have very time-limited deployments to the affected countries, 00:25:20.600 --> 00:25:23.159 which means they have different relationships with the governments 00:25:23.159 --> 00:25:26.140 of those countries. And so, when we were talking 00:25:26.140 --> 00:25:31.059 to the World Bank employee about how they used building damage information, 00:25:31.059 --> 00:25:35.091 they said, well, unfortunately, in the context of PDNAs or 00:25:35.091 --> 00:25:38.400 rapid assessments that we do, we have to use the government 00:25:38.400 --> 00:25:40.880 report that they send. They have their own way of 00:25:40.880 --> 00:25:44.760 collecting data on the ground, although it’s not very good. 00:25:47.550 --> 00:25:52.980 This stands in really dramatic contrast to how the urban search and rescue team 00:25:52.980 --> 00:25:58.710 leader talked about his team’s deployment after the 2015 Nepal 00:25:58.710 --> 00:26:04.860 earthquake. He said, initially the government had sent them to a 00:26:04.860 --> 00:26:07.480 particular area, or said that they should go there. 00:26:07.480 --> 00:26:11.210 And he said, in the end, we sent a reconnaissance party to that area, 00:26:11.210 --> 00:26:15.440 just to discover that the government really was telling us the truth, but we 00:26:15.440 --> 00:26:18.840 had so much information coming from other people who said you have to go to 00:26:18.840 --> 00:26:23.740 these other places, we said, well, let’s just confirm and see if it’s really true. 00:26:24.750 --> 00:26:30.820 So these examples show, first of all, that the relationships between the 00:26:30.820 --> 00:26:34.400 urban search and rescue teams and World Bank teams and the 00:26:34.400 --> 00:26:37.600 government are very different. And they’re interested in sending 00:26:37.610 --> 00:26:41.780 different signals to the government. So the World Bank has that 00:26:41.780 --> 00:26:45.630 long-lasting relationship. It’s important to build that relationship. 00:26:45.630 --> 00:26:50.420 And so signaling trust by using information, even if you consider it 00:26:50.420 --> 00:26:54.580 substandard, may be in your interest for other reasons. 00:26:55.280 --> 00:26:58.980 If you’re a urban search and rescue team leader, and you’re going to be out of 00:26:58.980 --> 00:27:02.590 the country within a week, you really don’t care, perhaps, 00:27:02.590 --> 00:27:05.370 about your relationship with the government of the affected country. 00:27:05.370 --> 00:27:07.940 You don’t need to signal that you trust them. 00:27:07.940 --> 00:27:10.680 You can operate in a very different manner. 00:27:12.420 --> 00:27:17.740 Another theme that emerged from our analysis is that different responders 00:27:17.750 --> 00:27:22.350 have really varying understandings of the same disaster. 00:27:22.350 --> 00:27:27.610 And those varying understandings stem from things like language differences, 00:27:27.610 --> 00:27:31.070 from the different organizational mandates that they carry out, 00:27:31.070 --> 00:27:35.060 and from the differences in their roles or in their backgrounds. 00:27:37.049 --> 00:27:41.059 So something that came up when we were talking to a loss estimation 00:27:41.059 --> 00:27:45.309 specialist was that the languages responders speak may actually shape 00:27:45.309 --> 00:27:49.290 their understanding of how a disaster happened or unfolded. 00:27:49.290 --> 00:27:54.400 So he noted, after the Nepal earthquake – the one in 2015 – 00:27:54.400 --> 00:27:57.500 he said, all the information was actually coming out. 00:27:57.500 --> 00:28:01.240 It was just all in Nepalese. So I’m afraid millions of different 00:28:01.240 --> 00:28:04.889 papers said it took the government four days or five days to get around to 00:28:04.889 --> 00:28:08.760 bringing out loss estimates, but that’s not correct. 00:28:09.560 --> 00:28:14.260 And so here he’s showing that many people’s understanding of what was 00:28:14.260 --> 00:28:18.970 happening after the Nepal earthquake was actually not accurate 00:28:18.970 --> 00:28:22.549 because they didn’t take into account that 00:28:22.549 --> 00:28:25.500 the information might be in a different language. 00:28:27.030 --> 00:28:32.340 He also made a note about language barriers that can influence the building 00:28:32.340 --> 00:28:36.340 damage data to which responders actually have access. 00:28:37.250 --> 00:28:42.580 So he said, in most countries, 90% of the best data is in their local language. 00:28:42.590 --> 00:28:45.000 So, in China, you don’t search in English. 00:28:45.000 --> 00:28:47.679 If you want a Chinese vulnerability function, if you want Chinese 00:28:47.679 --> 00:28:52.889 damage data, you search in Chinese. He made the point that not many people 00:28:52.889 --> 00:28:57.320 think to do that when they’re putting together loss estimates. 00:28:59.760 --> 00:29:04.659 In addition, how responders actually understand building damage information 00:29:04.659 --> 00:29:08.789 and act upon it can differ pretty significantly according to the particular 00:29:08.789 --> 00:29:12.889 organization that they work for and the mission of that organization. 00:29:12.889 --> 00:29:16.700 So the person that we were talking to here worked on doing building 00:29:16.700 --> 00:29:20.360 damage assessments after the 2010 earthquake in Haiti. 00:29:20.360 --> 00:29:23.630 Those were structural damage estimates. 00:29:23.630 --> 00:29:27.590 She was working on doing those assessments to inform the PDNA 00:29:27.590 --> 00:29:31.519 and not for humanitarian purposes. And she was reflecting on her 00:29:31.519 --> 00:29:36.019 experience, saying, you also get an idea of what’s left, and that’s really 00:29:36.019 --> 00:29:39.760 important because that’s your elasticity. That’s where you can absorb people 00:29:39.760 --> 00:29:43.929 into rental shared housing, temporary housing, into what’s left. 00:29:43.929 --> 00:29:46.300 And we never accounted for that in Haiti. 00:29:46.300 --> 00:29:49.420 We were just, like, oh my god, all of the stuff’s been damaged. 00:29:49.420 --> 00:29:52.360 We didn’t say what’s left, and where is it, and how can we 00:29:52.360 --> 00:29:55.700 open that up so people can use that housing? 00:29:55.700 --> 00:29:58.520 That, of course, would have been really valuable from a humanitarian 00:29:58.520 --> 00:30:02.860 needs perspective, but she didn’t think of it at the time because 00:30:02.860 --> 00:30:06.539 her organizational mandate was completely different. 00:30:08.600 --> 00:30:13.020 And then, people’s different roles and backgrounds can also really 00:30:13.030 --> 00:30:18.600 influence how they receive information that’s packaged differently. 00:30:18.600 --> 00:30:21.500 One person we were talking to – an information manager – sort of summed 00:30:21.500 --> 00:30:26.899 it up, saying, maps, figures, different people take data in different ways. 00:30:27.520 --> 00:30:33.200 A disaster reduction specialist reflected a little bit more, saying, showing a 00:30:33.210 --> 00:30:38.230 disaster manager a map, it’s not going to enthuse them if they just want numbers. 00:30:38.230 --> 00:30:41.429 And sometimes these numbers are something as simple as a table that 00:30:41.429 --> 00:30:46.180 shows this is what’s been impacted and the needs are then this. 00:30:47.429 --> 00:30:50.539 So those different roles and backgrounds can really impact 00:30:50.540 --> 00:30:54.680 how people react to information that’s packaged in different ways. 00:30:56.250 --> 00:30:59.300 And then the final theme that emerged really strongly from the qualitative 00:30:59.309 --> 00:31:04.230 analysis that we did is that the responders all have different attitudes 00:31:04.230 --> 00:31:10.500 toward emerging technologies. Previous work looking at responders’ 00:31:10.500 --> 00:31:14.860 attitudes toward emerging technologies has stated that they tend to be reluctant 00:31:14.860 --> 00:31:21.610 to adopt those technologies if they’re not problem-free at the outset or if, to do so, 00:31:21.610 --> 00:31:25.919 they would be replacing tried-and-true solutions – methods that they’ve used 00:31:25.919 --> 00:31:31.130 in the past that have worked before. But we found out something different. 00:31:31.130 --> 00:31:34.529 We found that responders actually have a really wide range of attitudes 00:31:34.529 --> 00:31:39.019 toward emerging technologies. And those attitudes range from, 00:31:39.019 --> 00:31:44.950 like, real pessimism toward extant technologies to real optimism about 00:31:44.950 --> 00:31:49.480 technologies with which they have no operational experience or that 00:31:49.480 --> 00:31:54.600 might not even exist yet. We also found that they tend to 00:31:54.600 --> 00:31:58.590 use emerging technologies in tandem with tried-and-true methods. 00:31:58.590 --> 00:32:01.649 They don’t replace those tried-and-true methods. 00:32:01.649 --> 00:32:04.409 And we also have this sort of conditional optimism about 00:32:04.409 --> 00:32:08.110 emerging technologies that I’ll describe in a minute. 00:32:08.110 --> 00:32:12.220 So one way to illustrate this – we asked a lot of people about their thoughts on 00:32:12.220 --> 00:32:18.139 using drone imagery of buildings. And we got many different responses. 00:32:18.139 --> 00:32:21.290 So the last estimation specialist we talked to was really enthused 00:32:21.290 --> 00:32:24.740 about drone imagery. He said, you can get that 3D image. 00:32:24.740 --> 00:32:28.660 You can get a lot better idea of the damage sites in the location. 00:32:29.420 --> 00:32:31.980 When we talked to someone who works in humanitarian needs 00:32:31.980 --> 00:32:37.100 assessments, he was not at all thrilled. He said, there are too many other things 00:32:37.110 --> 00:32:40.990 flying around in the air, then trying to do an aerial imagery assessment, 00:32:40.990 --> 00:32:45.360 that could be done faster by getting in a Jeep and driving 30 miles an hour. 00:32:46.100 --> 00:32:50.600 We also talked to the urban search and rescue team leaders about how they 00:32:50.610 --> 00:32:54.920 might use drone imagery to inform their operations, and they didn’t 00:32:54.920 --> 00:32:59.029 really see it as useful at all. They said it can’t tell me specific 00:32:59.029 --> 00:33:04.440 information about which building is probably a possible rescue situation. 00:33:07.520 --> 00:33:12.240 Some interviews also showed really great optimism towards untested 00:33:12.259 --> 00:33:15.679 technologies. So when we were talking to the urban search and rescue team 00:33:15.679 --> 00:33:20.380 leader, we asked whether he thought about using social media scraping. 00:33:20.380 --> 00:33:24.130 And he got really excited about the potential of that. 00:33:24.130 --> 00:33:28.090 So he said, so if you can take – if you can actually make a platform 00:33:28.090 --> 00:33:30.990 where you can analyze these tweets and pick a place. 00:33:30.990 --> 00:33:35.380 We see so many hits of people sending us tweets or sending information into 00:33:35.380 --> 00:33:39.520 the world that there really are a lot of victims, you can use that as a source. 00:33:39.520 --> 00:33:43.970 What was interesting about his response was that he didn’t really address 00:33:43.970 --> 00:33:48.559 what this platform would be. He seemed to view it as a black box 00:33:48.559 --> 00:33:53.710 solution, and he didn’t talk about how it would address the problems 00:33:53.710 --> 00:33:58.500 that he had found with other – interacting with people even on 00:33:58.540 --> 00:34:02.880 the ground, and their information not being reliable. 00:34:03.800 --> 00:34:10.140 Then we talked to someone else who was working in damage assessment. 00:34:10.140 --> 00:34:14.540 We didn’t ask about virtual reality, but she was very excited to talk about it. 00:34:14.550 --> 00:34:18.419 She said, I threw around ideas with people when I came back about some 00:34:18.419 --> 00:34:22.340 kind of virtual reality – some kind of tool which would allow anyone with a 00:34:22.340 --> 00:34:27.020 smartphone to take a useful, informative photograph to start to get images in the 00:34:27.020 --> 00:34:31.030 locations that you need to know to help inform the guidance that you’re going to 00:34:31.030 --> 00:34:33.680 give people who are going to rebuild. 00:34:34.460 --> 00:34:39.220 So she seemed very enthused about the potential of this emerging technology. 00:34:39.220 --> 00:34:44.260 Really optimistic about creating a tool that harnessed virtual reality 00:34:44.260 --> 00:34:50.300 for rebuilding, but didn’t have any operational experience with that. 00:34:51.900 --> 00:34:58.080 And then we saw some more tempered reactions to emerging technologies. 00:34:58.080 --> 00:35:02.960 We found evidence that people are already using emerging technologies 00:35:02.960 --> 00:35:06.500 as a kind of check on their tried-and-true solutions. 00:35:06.500 --> 00:35:13.829 So one loss estimation specialist said of drone satellite imagery that, 00:35:13.829 --> 00:35:17.460 we actually just check against what we see on the ground to 00:35:17.460 --> 00:35:24.119 see if their estimates are reasonable. And then other interviewees – 00:35:24.119 --> 00:35:27.900 multiple ones emphasized the importance of new information 00:35:27.900 --> 00:35:32.460 products regardless of the emerging technologies on which they are based. 00:35:32.460 --> 00:35:36.760 They should fit into existing workflows. So we threw around a few ideas about 00:35:36.760 --> 00:35:40.130 crowdsourced building damage assessments, and the reaction 00:35:40.130 --> 00:35:42.210 was pretty much, well, if you could add that into 00:35:42.210 --> 00:35:45.620 our tools chain, that would be really, really useful. 00:35:47.520 --> 00:35:54.460 So in terms of responders’ information use, what we found is that their use of 00:35:54.470 --> 00:35:58.560 building damage information depends on many factors that can’t be captured 00:35:58.560 --> 00:36:03.760 by simple availability and suitability of the information for their work. 00:36:03.760 --> 00:36:08.310 Those factors, again, are impediments to information sharing, implications 00:36:08.310 --> 00:36:12.740 of information adoption, their varying understandings of disasters, 00:36:12.820 --> 00:36:16.500 and also their attitudes toward emerging technologies. 00:36:17.560 --> 00:36:21.340 To wrap up, I’d just like to acknowledge the many people that have contributed 00:36:21.400 --> 00:36:25.560 to this work – Robert Soden, David Lallemant, Karen Barns, and 00:36:25.560 --> 00:36:29.680 Sabine Loos contributed to the work. Greg Deierlein, Jack Baker, and 00:36:29.680 --> 00:36:32.360 Anne Kiremidjian have been advisers on this project. 00:36:32.369 --> 00:36:35.630 And then we’ve had great collaborations with the University of 00:36:35.630 --> 00:36:39.720 Colorado-Boulder, the Humanitarian OpenStreetMap Team, the University 00:36:39.720 --> 00:36:43.630 of Heidelberg’s GIScience Research Group, and also the 00:36:43.630 --> 00:36:48.119 World Bank Global Facility for Disaster Risk Reduction. 00:36:48.120 --> 00:36:51.540 And this work was sponsored by the National Science Foundation. 00:36:51.540 --> 00:36:53.600 And I’m happy to take any questions that 00:36:53.600 --> 00:36:56.540 you might have about the presentation. 00:36:57.000 --> 00:37:02.080 [Applause] 00:37:02.140 --> 00:37:04.800 - That was – that was [inaudible]. 00:37:07.000 --> 00:37:10.080 That was a great talk. Thank you so much. 00:37:10.080 --> 00:37:12.400 Does anyone have any questions for Gitanjali? 00:37:12.400 --> 00:37:15.760 I know Anne will have many questions. Hold on. 00:37:15.760 --> 00:37:17.220 - You can go to someone else [inaudible]. 00:37:17.220 --> 00:37:19.180 - No, no. Does anyone else – before Anne? 00:37:19.180 --> 00:37:21.020 Because she’s going to ask a lot of questions. 00:37:21.020 --> 00:37:23.440 - No, no, no. Just ask one. - Just ask one. Really? 00:37:23.440 --> 00:37:25.160 - Just for now. - Okay. 00:37:25.170 --> 00:37:29.540 - Can you go back to your slide with the graph – the figure of time 00:37:29.540 --> 00:37:34.180 and building resolution? Yeah. That was interesting. 00:37:34.180 --> 00:37:40.980 So I wondered if you missed out the block- and city-level uses of 00:37:40.980 --> 00:37:44.260 the information because of the people you spoke to. 00:37:44.260 --> 00:37:48.960 So you’ve got national, mostly. Whereas, a lot of recovery decisions 00:37:48.960 --> 00:37:53.340 are made at the local level. And did you tap in – did you have 00:37:53.340 --> 00:38:00.180 any people that were – that were representing that role in your sample? 00:38:00.180 --> 00:38:05.840 - We had not many people representing work that’s done during the recovery 00:38:05.840 --> 00:38:10.620 process, so that’s entirely possible that we didn’t represent those tasks well. 00:38:10.620 --> 00:38:14.190 - Yeah. Anyway, that would be my hypothesis on that because I think, 00:38:14.190 --> 00:38:18.069 in general, the concentration of damage – the pattern of damage 00:38:18.069 --> 00:38:23.460 in an area on buildings is quite important for how you recover. 00:38:23.460 --> 00:38:24.950 - Okay. - Yeah. 00:38:24.950 --> 00:38:27.180 - Okay. Thank you. 00:38:27.180 --> 00:38:30.960 - I have a couple of questions. For those of us that aren’t 00:38:30.960 --> 00:38:34.530 qualitative researchers, can you please walk through your methodology a little 00:38:34.530 --> 00:38:36.790 bit more and explain the method? - Sure. 00:38:36.790 --> 00:38:40.100 Let me grab a slide on this. 00:38:43.520 --> 00:38:48.460 Sure. So we did a qualitative analysis. What that meant in practice is that 00:38:48.470 --> 00:38:51.920 we got all of the interviews that we’ve done transcribed. 00:38:51.920 --> 00:38:55.710 We sat down and read them multiple times. 00:38:55.710 --> 00:38:59.569 And we tried to see what themes were emerging from the interviews 00:38:59.569 --> 00:39:03.390 about how people were specifically using and interacting with building 00:39:03.390 --> 00:39:08.360 damage information. So that was a process of inductive coding. 00:39:08.360 --> 00:39:11.680 And that process, which had many, many iterations, 00:39:11.720 --> 00:39:15.500 led us to identify those four codes in particular. 00:39:16.300 --> 00:39:19.880 - Great. - Does that answer your question? 00:39:19.880 --> 00:39:22.380 - Any other questions for Gitanjali? 00:39:24.360 --> 00:39:27.340 [Silence] 00:39:27.720 --> 00:39:32.520 All right. Well, we might end the seminar a little bit early today. 00:39:32.520 --> 00:39:37.920 Before we say goodbye, I just want to have everyone thank Susan Garcia for 00:39:37.920 --> 00:39:42.480 helping us set up on – every week. So thanks, Susan, for all your hard work. 00:39:42.480 --> 00:39:45.760 [Applause] 00:39:45.760 --> 00:39:48.900 And we’re going to go and take Gitanjali out to lunch now 00:39:48.900 --> 00:39:52.390 in the café on campus. Anyone is welcome to join us. 00:39:52.390 --> 00:39:54.790 And thank you all for coming, and hopefully we’ll see you 00:39:54.790 --> 00:39:56.760 here again tomorrow. Thanks. 00:39:57.120 --> 00:40:01.040 [Applause] 00:40:06.140 --> 00:40:11.740 [inaudible conversations]