 We're both really excited to be here for the record. We created this talk when we were living in the same city and in Houston. And we are friends because we are passionate about data, food, and organizing meetups. And today we want to talk about some of the things we worked on together during Hurricane Harvey last year as we were part of the tech community response to Hurricane Harvey. Before that, just a quick introduction. This is Neeraj Ayulon. I'm a developer advocate at Google working on a product called Google Data Studio. This is where I say this talk has absolutely nothing to do with what I do at work. I am not speaking as a representative of Google, but please come find me if you want to talk about BI or DataVis. OK, I'm done. I'm Neeraj. I'm a principal consultant at January Advisors. We work in the public sector, and we work with large nonprofits. We were interested in projects for the social good. So if you have any ideas or questions, please come talk to me. Cool. And you'll get to hear about some of the things that Neeraj is working on, because they'll be part of the stories that we tell during this talk. And so a quick roadmap. We're going to talk, first half of the talk is going to be lots of stories. What happened during Hurricane Harvey? How did the tech community respond to that? And then the second half of the talk will be some of the project management and the structure around it. How did we make decisions that facilitated bringing people together around a common cause to build solutions for really immediate, rapidly changing needs? And with that. Thanks. Yeah, so everyone here has heard about Hurricane Harvey, was following the news closely when it was happening, was stunned and shocked by the devastation that they saw. But how many of you were in Houston at that time? Great scenes. OK, one person. So I'll just set the scene really quickly for you. We, when we were expecting Harvey to come, we basically, as a city, had a few close calls where we'd had hurricanes coming in. We had this mantra hunker down and just shelter in place. So we didn't want to go and have a huge traffic jam with everybody leaving the city. So our city officials and county officials told us stay where you are, ride it out. So I was like, cool, I get a weekend. I'm going to work on some projects. I went to the hardware store not to pick up supplies for the hurricane, but to pick up supplies to do a little project where I was going to put lights in my cabinet and hook it up to a Raspberry Pi. And I got all my stuff. I was super excited. I was in there. And as I was, that evening, when I was trying to get the lighting to work and it wasn't wiring up and I was getting really frustrated, I started getting texts on my phone. Friends were asking me, hey, are you seeing what's happening? This is devastating. People are in a lot of trouble. Can we do something? The reason they were asking me or reaching out to me about that is because I'm a co-organizer of a few different meetups. One of the meetups is a code for America Brigade in Houston called Sketch City. And we host the city of Houston hackathon where we work on tech solutions to large social problems. And so some of these people had been on teams with me or we had met through there. And they were like, hey, is there anything going on? They wouldn't know if there were any projects. And I was like, there's a project in my kitchen. But that's not what you're asking about. I should get off my butt and start doing something. So a few of us got together and started working on the first most pressing need, which was rescue. So people who are in imminent danger, they're in rising water, need to get out of there. And so the app we started working on was for people to drop a pen on a map, say, hey, I need rescue at this location. And an important step that happened that led to that is what happened during this hurricane was our county judge, who's kind of like the executive of the county, the leader of the county, made a critical decision to ask people to help each other. So a lot of times during disasters, they don't want citizens running around, they don't want citizens running around trying to save each other, because usually those people end up, those civilians end up needing rescue themselves and end up bogging down the system more. But they were completely at capacity. All of the first responders were completely maxed out and they said, hey, if you have boats or high water vehicles, get out there and start helping your neighbors. And so in order to facilitate that, we started building this map. And the idea was just solving one problem at a time. First, where do you find people? Second, how do you delist people once they've been rescued so somebody else isn't going to that same spot? So all these little solutions we had to keep coming up with. There was another tool that rolled out around the same time. There were a few little pockets of developers that all started having these ideas and wanting to work on things. So we wanted to pull all those people together. Another project is another one that really we're gonna spend a little bit more time talking about today. So as soon as I saw that that immediate need was being met, I kind of got onto Facebook, tried to make sure people, everybody knew was okay, look at social media, see who's doing okay. And I saw that one of my friends had posted that he had built a text bot. This friend of mine builds text bots for everything. He's got one for finding your voting location. He's amazing. And so he had built a text bot. So a city official, a city council member had posted a list. It was an image of a list of shelters. So this is the next need that you need. So you have rescue and then once you rescue people then need a place to stay. That's safe. So there were a bunch of shelters that had opened up and this was a static image of a list of shelters. There were no addresses, it was just names. And so he said, hey, this isn't searchable. You can't actually like click on this and find it. It's gonna be hard, especially for people who are already dealing with a crisis for them to actually get to where they need to be. So he made a database, put it in there, geocoded all the addresses and then built a text bot so people could text their zip code and get their directions to this place. And then somebody else posted on that same thread, hey, how do we make sure this stays up to date? Because shelters were opening, closing, reaching capacity. Things were changing super fast at that time. And so we started a Google spreadsheet. And a phone bank. And a phone bank. So, and this is where the community really stepped up. The second that Google spreadsheet was open, people were on it and you just started seeing like the little boxes that are moving around when people are editing. Everybody was on it. People we didn't even know started calling and filling out data and the data was just being updated in real time. It was really something that no one agency or organization could have taken responsibility for because there was this crowd source information that was just in time. And it was the most up to date list of shelters, information availability and needs that existed during that time, like during Harvey and shortly after. Yeah, and public officials started sending that link out to that spreadsheet. And then people started building tools against it to try to use the data. And so the idea was that, we could actually have community-driven collaboration for people to come together, build tools that make the immediate needs as they evolve. So you have these really acute needs for rescue, then you have needs for shelter. And every time a new need popped up, people were coming up with ideas and coming together to build tools for them. And that was really cool. We started pulling out these different pockets of developers together. We kind of created this headquarters of it. And I think about how it felt for me to be there in that moment because I don't have a boat and I wanted to help so badly. I wanted to be out there. I saw all these people that were up to here in water carrying their children and I was like, what can I do? And so being able to build tools and help with the information problem and really link people to resources was something that I was really grateful to be in the position to do. So that Google spreadsheet eventually evolved. Well, first we maxed it out. This is before Yulon started working there. We maxed it out and we actually reached out to Google and we were like, hey, it's starting to be a little bit unreliable because we had too many people on the sheet. I think we had like 130 people trying to edit the sheet at the same time. And they were like, look, this is just not what this tool was built for. We can't rebuild this whole thing to support this. Here's some best practices, ideas, fragment the list, do things like that. What we found was that that Google sheet was doing a lot for us right out of the box that we didn't want to lose. It provided a place for people to edit data in place. So you could see the context of what you were changing. It provided an opportunity for everybody to collaborate through a web browser. You don't need to download a tool. It also provided that social context. So you're on there calling, sitting in your home, wondering if anybody else is out there and you're seeing all these other boxes moving around the page, working along with you. You feel like you're part of a team. Even though you don't know a lot of these people. And that social context was really important and motivating. So we wanted to replicate that as much as possible. We quickly built a Ruby on Rails app that had a lot of those same features but allowed us to support newer features and start building and expanding on it as new needs came up. One of those needs was the idea of having more complex data management. So we have initially a bunch of people editing and amending data. Then as we started going on, and Yulon will talk a little bit more about this later, we found that there are best practices, their ideas for people who started making mistakes or misunderstanding, misinterpreting field names or things like that. And the data started becoming unreliable. So what we were able to do is say, okay, we're gonna train groups of volunteers as they start coming in. And then we're gonna have, if anybody out there, here's some information or has some idea or has some context of this. We don't wanna lose that information. So we'll capture that in a queue and then we'll have the trained volunteers process that queue and add the data in. So we started moving away from saying, hey, any information is good information too. We want all information. We're gonna validate that information. So we start evolving kind of the process to be a little bit more mature. And that Rails app was backed by a Postgres database that was also, they had an API on it that was serving maps and text bots and apps. And what's cool about it is, we had like three different maps because people wanted slightly different things. And just because the API was available and it was documented almost right away, these people that came out were just so talented. All these talented engineers were coming out and just had this really complex roadmap and all these different ideas and we're working against it and we're working together really well. We had a really nice resource that people could use to build tools off of for emerging needs. Another thing that came up was, so I mentioned that we had rescue, then we had this relief and getting people into shelters and then we had a ton of volunteers. We just had people coming in from other cities because everybody had seen how devastating this disaster was and they were lining up and there was this, the biggest shelter was the George R. Brown Convention Center. This is our big convention center and everybody was hearing about the George R. Brown, so the line of volunteers was all the way around this massive convention center building to even get into volunteer and there was nobody in some of these other smaller shelters, some of the other community centers and churches and things like that were completely understaffed. So to get information out about that, we were using that tool that I mentioned a little while ago. We added from saying here are shelters, also here are the shelters' needs. They need volunteers, they need supplies, they need these things and those things changed fast too because somebody says a shelter needs toilet paper or diapers, somebody loads up a truck full of diapers and takes it there, they don't need diapers anymore and you don't want another three trucks coming with diapers there when maybe somebody else needs them. So having that real-time data was really good but coming back to this, once we had all these volunteers putting in all this time, it was really important for the city to be able to account for all these hours and why that was important is because the city can apply to FEMA to get reimbursement for all the volunteer hours that were performed. So they were really interested in having this built and we built this really fast. This was like a two-hour project but it allowed them to go from having pieces of paper scattered around the table where people are like, look, I don't care, I don't want to build out this form, I just want to help to being able to really quickly go on there, log your hours and move on and having it all centralized instead of having to collect the data and do entry. So this solved an immediate problem, we logged over half a million hours of volunteer effort and the city was able to use that to try to apply for reimbursement. After Harvey, so right after Harvey, I don't know if you all remember this time when it was like, wow, this is the apocalypse, there's this storm, then this storm, then this storm, especially if you lived in this region, it really felt that way. We had Hurricane Maria, Hurricane Irma, then we had that major devastating earthquake in Mexico. Because we all had this team together, all these people together, we had all these tools, we said, hey, can we be proactive about this? So we took all these tools, we jumped onto somebody else's slack or we found whoever was starting to coordinate a tech response to any of these disasters and said, here's our tools, take it, do what you want with it. And we had some interesting politics come up where some of our engineers were really attached to what they were building and had a roadmap and were trying to do features and the people who were in the locations that were about to be affected were like, no, we want to be able to build whatever we want, we want to rebuild features if we want to. And so there's a little bit of that negotiation that had to happen and at the end of the day we were like, look, you're there on the spot, you need the same flexibility that we had to build the things that you need as you see them. So they branched the code and they started building stuff and then they could pull features in as we were building them, as their needs emerged. But that was really cool to be able to contribute to these other efforts. And because a lot of these areas have large Spanish speaking populations, that became a really important need that was kind of a difficult thing to do, but we could say, hey, we could deal with this on our end while you're working on some of the more front-end features. The language support became a big thing too. And so now I'm gonna talk a little bit about what's going on now. So over a year after Hurricane Harvey, lots and lots of people still aren't back in their homes. Another application that came out was the MUC map. So this is a similar peer-to-peer type of effort. So what happens is when you have a lot of water in your home, the walls need to be completely torn out. The walls, the floors, everything that's damp needs to be completely torn up. It's a health hazard to keep that in there. So we had these points being dropped on a map. Hey, my house needs mucking. You'd have a group of volunteers descend on the house, take all your personal stuff, all your furniture, all your belongings, your whole life out, throw it into the trash on the street. They'd come in, tear all the walls out of your house and then they'd leave. They'd be onto the next house. And we have a lot of people in Houston today who still don't have walls in their home. They haven't replaced all of that property. And so to be able to help them to get access, there's a lot of funding that came in. There's a lot of social services out there. They're supposed to help them get back on track, but help them connect to those social services. We're trying to use some of what we learned from that project that we told you about earlier to build a database of resources that provides a system for referrals. So as soon as you go into one place to meet one need, they can help you find all the resources to meet all your other intersecting needs. We're working with the Houston Immigration Legal Service Collaborative and the United Way of Greater Houston on that project. And the hope is that these organizations, the United Way will take on this as a long-term thing because a lot of the needs that people had were there before the hurricane hit and they'll be there long after they recover from the disaster itself. So we're really hoping this will have a big impact. In other tool, this was recently deployed for Hurricane Florence. This was the peer-to-peer rescue app. And this one is still going strong and they were starting to find other ways to repurpose the peer-to-peer system for rescue. And they're starting to address other needs like privacy and things like that. These are things that we didn't think about at the time, right? Because we were like, there are acute needs. I'll put my address, I'll put my name. Now as organizations start taking ownership of this, these questions start to have to be answered. But these projects are going really strong. I'm really proud of them. And I'm really proud of everybody that worked on all of these tools. And so if someone was curious about how things were implemented or deployed, where can somebody find that code? So if you go to the Sketch City GitHub, you'll see a lot of the code for a lot of these projects. And please feel free to check it out, fork it, or contribute to it. Thanks, Neeraj. Thanks. Thank you guys so much. We have time for... We're not done. Keep going. Okay. Hi. So before I launch into this next section of like some of the ways we chose to work and some of the things we were thinking about with project management, I want to tell you that I was not in Houston during Harvey even though I lived there. I was actually visiting friends in Pennsylvania and then my flight kept getting canceled. And then I just like couldn't get a flight back. And so I was there. I was trying to like work remotely, but then also trying to jump on some of these projects. I jumped onto the Shelter's project. I was initially phone banking. I was like, hey, like let me call, I can call from wherever I want to and like update the spreadsheet. And then as I saw the scaling problems, I saw people having trouble figuring out like, well, we can only have so many editors or like this is getting messy or having data management problems or even onboarding new people. I think this is where some of the process and understanding like how do we rapidly create processes that are appropriate to the situation came in. And this is the part of the talk that I wanna focus on because this is the side that I saw the most of. And to set the stage, we had people on the ground in Houston. We had people like Neeraj who were in meetings with community leaders and potentially like only responding to two texts a day, but they had all the information. And then we had like developers who were in co-working spaces in Houston over like lots of coffee and pizza. And then we had developers from all over the country who were on our Slack saying, how can we help? We have some skills, we know you need them and trying to match all of those together and coordinate that into an effort that would help people on the ground. And so the three key things I wanna talk about is kind of how quickly we had to communicate and the trust that supported that communication and also caring for people because it is in the midst of a disaster. So context of our communication, we had rapidly changing needs, right? We had as soon as people found shelters or as soon as people found were rescued or as soon as a shelter had a certain need filled like something else was coming up. And so we had to make sure that there was a place, there was a single source of truth for everything. And initially that was that massive Google spreadsheet and for different projects we had different single sources of truth but the establishment of this is a single source of truth was really important. And so that's one thing. The second thing is it was chaotic in terms of people resourcing. We had people coming in and out of the projects because they had their own things to deal with, right? Some of them were impacted by the hurricane in different ways. And so we had to be cognizant of that, like I couldn't just be like, I'm gonna treat you like someone who's working eight hours a day, 40 hours a week on this thing because that's just not the nature of the situation. And then like I mentioned with community leaders being in meetings and maybe not able to constantly communicate with the developers in Slack, we have these human bottlenecks of communication. And so some of these considerations that we use, what are tools that people are already familiar with? If I'm in a crisis situation, I don't wanna have to train people, right? I don't wanna have to have some elaborate dev setup or some like fancy tool that has an incredibly high onboarding. There's just not time for that. What is easy to scale? And it's not just like what will support more things but it's also like from a human perspective like how can I get more volunteers on it? And it's very closely related to like what is the familiarity of people with this tool? And then finally, what are tools that empower people to do things on their own, right? What are ways that we create the groundwork, create a framework for people to communicate for people to manage projects such that things can be organically driven and still discoverable. And so these are some of the considerations that fed into us using tools like Slack. We were coordinating on the Sketch City Slack and that's a Slack that was already in place to coordinate civic tech things. Because people already knew how to use it, we just prefixed all the channels with Harvey Dash. And so then we have discoverability, people already know the tool, people already know the community. Using Google Docs. Google Docs is incredibly low friction in terms of onboarding. We created a master project management document that said these are the projects, these are their current skill needs, these are the people you can talk to because you have people with all different skill sets and abilities coming in who want to help. Some people want a phone bank, some people want to build, some people are really great at project management and having the central source of truth. So somebody hops on Slack and says, I want to help, how can I help you say, go look at the document, talk to a project lead. And that's probably how I spent all of my time is creating that document, maintaining that document and then directing people there and saying, let me connect you to a place where you can plug into your skills. And then Google Hangouts. What does that look like to just be able to hop on a call or on a video call with somebody I need to camera something out real quick with? And then I didn't mention this in the slide, but Google Sheets was really powerful for us as a collaborative tool, as a database, and in a lot of other ways. And so just kind of creating that framework and saying, hey, take this framework, run with it. Another thing we had to think about with that master project management document was who had access. Nierge and I created that document in a 90 minute call the night before a flood of volunteers and things were happening. And so we had to think about, should we give everyone who's interested in working on these projects edit access? And the answer is no, because we wanted the document to maintain a sense of readability and usefulness. And so trying to figure out, so who gets edit access? Because I can't be the bottleneck for that. And then trying to figure out, okay, we have these project managers. How do we do handoffs? How do we say this is, I used to own this document but I'm going to let go of that control because I know you now know more about this project than I do because you are following it on the ground and so many things are happening, whether it's the needs are changing, whether it's the code-based changing, whether it's the team changing. I'm not able to keep track of that for every project. Rapid communication requires trust. When Nierge disappeared for half a day, I needed to trust that the processes we had built and the things that we had said were still valid. But also within the community, right? A lot of what we were able to do was build on people who organized meetups, like the sketch city meetup. We also co-organized the Houston data visualization meetup. I was a co-organizer of the Houston girls coding meetup. And all of these communities meant that we had some kind of prior trust with each other. We knew each other's communication patterns. We were comfortable working with each other. Maintaining trust, it's in a chaotic environment and some engineers really just wanna build a thing. And so there's a difference between coming into a project and saying like, I'm just gonna start coding. Versus, hey, where can I plug in? I have XYZ skills, right? And so I saw people do both and tried really hard to shift people towards the second because the first was unhelpful, would lead to duplicated effort. But also break trust with people who are on the ground and knew the needs more closely and we needed those relationships. And with the idea of who decides what to build, how do we make sure that that's a localized decision, that it's not built by what feature would be cool but what feature is needed? People care, it's a disaster. So actually the folks from Hurricane Sandy were really helpful. They would pop on our Slack and just yell at us, go take a break, walk away from your computer. And I ended up doing that after a couple of days and that's just kind of the nature of it and because there's just so much going on and people are whole people, they're not just faceless things behind a screen. Working this way isn't unique to disaster response. We're talking about this in the context of Hurricane Harvey but this happens in every organization. There's some kind of crisis or something where you need all hands on deck and so you'll have to be rapidly communicating. Have to build on existing trust and also communicating in ways that maintain that trust so that people don't just get pissed off and walk away. And people care. People you work with, you yourselves are people that exist outside of the code you write and the things you maintain. And so how do we make sure to account for that and treat each other like humans because that's what we are. Neeraj mentioned this. Harvey is not over for everyone. It no longer looks like this in Houston. There's not debris everywhere on the streets but there are still people who are deeply impacted by that. And so I think as we're telling some of these stories about the just the tech community, I do wanna zoom out and say like, there are a lot more people that were impacted by this and still are to this day. Finally I wanna say, here's some takeaways. One is just we wanted to share this story. We wanted to tell you about some of the ways the tech community responded to Hurricane Harvey, what it was like to be part of that effort. Wanted to talk to you about kind of the trust and the communication and the choices we made that allowed us to build the projects we did and highlight the importance of having a community, building a community well around technical things, not only because you come together and learn really great things but also because you can really deeply impact the lives of your community. And I wanna encourage you all to get involved. Find a local code for America Brigade, find people who are working on responses to other disasters. Each of you, if you are in this room, has skills that are deeply needed. And with that I wanna thank you for your time and invite you to keep in touch. Nirij and I will be hanging around the conference. Thank you so much. We do have time for one question. Can I have like one last two sentence thing? You may give two sentences while we ask for ops view if you're in the house to come find me. Okay, so we are the ones giving the talk. There are like thousands of people who worked on these things, not all of whom we have met in person. And so I wanna give a shout out to them because they really made it all happen. Yeah. Thanks. Thank you.