 Great. Thank you so much. And yeah, thank you to everyone attending and especially to my society for continuing to host this conference online Stacey and I are actually first-time tic-tac attendees and We're excited to share some lessons we've learned from building democracies database So I'll start with a number 519,682 I was going to ask for a show of hands for anyone who knew what this number represented given that we are remote I will just tell you all that this is the total estimated number of us elected officials That includes local municipal governments town and townships counties school boards and special districts Which are a fun thing that I could talk about with anyone who's interested But how many elected officials are there in the entire world, so that's just the United States How many officials are there in the world's democracies at all levels including local? The real answer is that no one actually knows exactly because there is no definitive single source of that information So we've been working on building a database of the world's legislative districts and elected officials for a little over a decade now And we've learned a number of lessons along the way But first because we assume that most of you probably aren't familiar with us. We want to briefly tell you who we are So we work for an organization called as a via We are a mission driven geospatial software company located in Philadelphia, Pennsylvania and Our mission is to advance the state of the art in geospatial technology and apply it for civic social and Environmental impact and we stick really closely to these three areas I don't know how familiar folks are with the corporations We have always operated as a social enterprise since our founding in 2001 But in 2010 we also became a certified b-corp which essentially means that as a business We strive to balance our purpose with the need to make profit So how does this relate? I'm going to take a moment and tell you about Cicero's origin story Which is a true tale of civic engagement So in 2006 a nonprofit organization local to our office called the greater Philadelphia cultural alliance Came to us with a problem the city of Philadelphia was about to pass a new budget that slashed funding for arts programs and The cultural alliance wanted their members to reach out to their city council representatives to let them know that this would negatively impact their lives and the city The problem was that most people didn't actually know who their local representatives were or how to get in contact with them And I see Mark nodding at the time there were services that were available that really focused just on contacting your federal representatives Studies have shown that fewer than 20% of Americans can even name their state legislators and that's not just a problem in the US, though Other countries fare slightly better worldwide. There's still a gap in common knowledge of civics government and representation. We really wanted to support this organization and they didn't have very much funding. So for about $2,500 we built for them the first iteration of what we called Cicero It was a database with an architecture that followed the structure of real governments within individual countries. And we populated the database with legislative district boundary shapes, which would enable us to match full addresses to legislative districts rather than just relying on postal codes. We felt really strongly that if we were going to help inform people about who their representatives were we had to make sure that we're providing the most accurate matching possible Nothing deters civic participation more than directing someone to the wrong elected official and we didn't want to be responsible for that. So after that we added information on elected official names and contact information at the local and state levels in Philadelphia, which we gathered by hand. Then we created an API that would match addresses to the correct district and return elected official information and you're looking at a very early iteration of Cicero here. We then turn that over to the organization to use for their advocacy efforts. It actually enabled more of TPCA members than ever before to reach out and contact their elected officials at the local level. And ultimately the city did not pass the budget that would have resulted in funding cuts for arts programs. So TPCA's efforts were overwhelmingly successful. We had just witnessed a successful advocacy campaign come to fruition, which was super exciting. And it was led entirely by Philadelphians who cared about a common issue and just needed more information to help them take action. So this kind of led us to ask ourselves if we helped one small nonprofit in this way, could we expand Cicero to help others. So our mission became to be the global resource for elected official information and geographic data on elections and democracy. As we've expanded our data coverage from one US city toward the ambitious goal of covering every democracy in the world, which spoiler alert, we have not yet met. We've learned some valuable lessons along the way and I'll turn it over to Stacy now to talk about some of those. Thanks, Stephanie. Can everyone hear me? Okay, cool. Yeah, so the first lesson we learned fairly quickly after looking past the city of Philadelphia is that political geographies are complicated. We've been really thoughtful with our database model. So all we really needed to do was collect the data and keep it up to date. There were two types of data we'd need, legislative boundary shapes and elected official names and contact information. Seems pretty easy, right? At the federal level in the United States, it is fairly easy. At the state level, it gets a little bit complicated because data on the 7000 plus state elected officials is not made available in bulk in a machine readable format. And GIS data lives in a variety of different state agencies, depending on where you are, but we can and did sort of deal with all of that where it got really complicated was at the local level. Just to show an example, these are the shapefiles of Pasco County, Florida voting precincts and they are colored by Board of County Commissioner District. Just to get everyone on the same page quickly to briefly explain what a shapefile is. It's just a digital file format that displays a geographic area such as a legislative district like here and we need these for every legislative district in order to match a person to an official. So if we wanted to add this county commission to the Cicero database, we'd need to know the bounding lines for all 11 districts here in order to know which commissioner represented any given constituent. If we take a look at the west side of the county, you'll notice issues with the digitization of these district files. All these white spaces in between the pink and purple. In reality, there are no white spaces. Those should be full color. There are no gaps in coverage of these commissioners, but now it's our team's job to find supplementary documentation to get a more accurate reading on where the district lines fall. And this is a common issue across localities. They lack the technology or they don't have the need to digitize or any number of reasons why their files might look like this. Often a small town's district map only exists in the form of a hand drawn document and sometimes you're lucky and they have uploaded a PDF of this online, but even then you have to digitize those shapefiles to make it functional. As Stephanie mentioned before, there are over 500,000 elected officials in the US alone, and expanding the Cicero database to cover all of them would mean tracking down and cleaning up shapefiles for hundreds of thousands of districts. And as we've learned, political geographies are complicated. Second lesson we learned again fairly quickly after beginning to expand our work is that election data changes frequently. In setting up our processes for monitoring and running QA, QC checks on the data we were keeping, we quickly learned that this data changes frequently and whatever tooling we built would need to work with that understanding. What information is changing? Well, elected officials change their Twitter handles, for instance, from one campaign to the next. Their office numbers change without notice, and elected officials are constantly stepping down for all sorts of reasons, such as this State Senator in Texas. We decided early on to have a full-time team dedicated to keeping our information up to date. Part of election data changing frequently includes URLs changing and breaking all the time. Therefore we couldn't rely consistently on scraping and effectively scale our efforts. Having a brittle process based on a small town's website is not going to be consistent or stay the same forever, and maintaining all of these scrapers for all of these small sites would take a lot of maintenance itself. So we put some smart processes and built some intuitive tools to help our small team be more efficient at the change detection process. We set up Google Alerts, we wrote custom SQL scripts, we added the ability for users to report errors to us directly, and we employed a service called Visual Ping that detects changes on the websites we track as shown here. It's not a perfect system, but it helps us keep up with frequently changing election data. When an election happens, we rely on official outlets to tell us what happened. We need to know election rules and calendars. We need published and updated election results, and we need a clear understanding of what happens when a race is too close to call or results in a tie. Government websites and local news outlets are our go-tos. But what happens if your government website is literally just an online PDF of contact information? Or your official election results page doesn't have clear documentation about the rules of a near tie. For instance, will these races in Spokane result in run-offs or recounts, or are these margins large enough to call a winner? Or what if the local newspaper is the only source publishing election results for an entire county, and you hit the paywall before you finish gathering the information about all the races happening in that county? If local news outlets are the only sources publishing election results across hundreds or thousands of localities, do you have to start subscribing to hundreds or thousands of newspapers? The third lesson we learned was that your data is only as good as a source. Typically when our usual data sources are insufficient, we can email a legislative secretary or the county board of elections to get the information we need. Most are kind and are able to respond, as seen here, but the work is often held up until they have the time to do so. In rare cases when we get no response, we have submitted FOIA requests. Lesson number four is that change won't happen without this. Non-profits and advocacy organizations need this data, specifically at the local level, to do their jobs. And I'll share a couple of stories from some of our users accessing this data to create real change. The Animal Defense League of Arizona is a small grassroots 30-year-old organization charged with protecting and defending animals across Arizona. Through legislation, wildlife protection, animals and entertainment advocacy, fair neuter clinics and other efforts. They mobilized their members using our data during the legislative session to stop bad bills and encourage the passing of good bills. And in 2017, they employed this tactic to get the landmark hot car bill in Arizona passed and effectively saved the lives of many animals. In the 2019 legislative session, Washington Environmental Council and their partners were working hard to pass a law to transition the state of Washington to 100% clean electricity. They knew that a few legislators were swinging votes on the issue and they needed their constituents' support to be able to vote for this bill. So they sent out an action alert to their members in those specific legislative districts using Cicero's data. And their plan was a success and ultimately they were able to garner enough support to pass 100% clean electricity across the state. In our own state of Pennsylvania, the State Education Association ran a very successful Twitter campaign a few years ago to showcase poor classroom materials. They posted pictures and used Cicero's social media data to tag the appropriate state representatives and garner attention. The campaign did do that. It got a lot of attention and it actually resulted in more funding for materials across the state schools. Another state education organization, this one in Kansas, took a different approach and generated this visual using our data for their annual state advocacy day to show legislators how much aid money is being spent in their districts. This campaign was successful in winning more funding for the Kansas Comprehensive Grant whose recipients are 12% more likely to graduate than their peers. And since our first success with the GPCA, countless organizations focused on local advocacy such as the Chicago Coalition for the Homeless, Bike Walk Kansas City, PFLAG New York City, the Philadelphia Parks Alliance, the Scholarship Foundation of St. Louis, Seattle Metro Chamber of Commerce and many more have utilized Cicero to advocate in their communities. Take the London Cycling Campaign for instance who use Cicero to match their members to greater London Assembly members and advocate for better cycling regulation in the city. Campaigns like these underscore the persistence of local advocacy and the importance of our commitment to cover all democracies around the world. Change won't happen without it. So it's been over a decade since we started Cicero and we still aim to be the global resource for elected officials and democracy. We're currently covering nine countries and we are still expanding daily and we do so based on requests of our users and folks that are actively using the data. We encounter new challenges every day and with every election cycle. Right now for example many election dates are moving, being postponed or paused due to the coronavirus outbreak and we are scrambling to track all that information and make sure that we know when those new elections are being scheduled. We're pretty lucky to have users that are in many cases willing and able to pay for our data which certainly helps us offset our costs but we've always been committed to making our data freely available to any individual and to our local community even when it results in a loss for us. Looking to the future, our biggest challenge is being able to scale our coverage while maintaining the same levels of data quality which is crucial. We don't believe we can do that without leveraging new technologies. As Stacy mentioned, a lot of our data collection is necessarily manual. We think that humans will always need to be involved in the process to some extent but we believe there are some stages that we can better automate. So to that end we've dedicated some time to using machine learning and natural language processing to help increase our efforts. We've also applied for grants to help fund research in this area. To our knowledge no one has used natural language processing to collect elected official contact information from websites. So far we've been unsuccessful with our funding efforts but we're definitely planning to keep at it. We know the scope of the problem is big and we don't want to try to do this by ourselves which brings us to our last lesson. We can't do this alone. There are already, we know several of us tackling this issue from different angles and we want to give a special shout out to my society's every politician effort which we know is currently paused but we took great inspiration from. In 2017 we partnered with Ballot Ready which is a nonprofit in the United States seeking to better inform voters about what's on their ballots. In order to do that they also need to draw upon local legislative district shapes. So we had shared interest in that area. We funded their collection of some of the local city council districts in the United States and similarly we're open to relationships with others now and in the future. We believe pretty strongly that this data can and should be made available globally especially at the local level and we'd love to talk to anyone else working in the field or anyone who wants to collaborate or partner. We were hoping to be able to meet some of you face to face but in lieu of that please please do reach out. Here's our contact information. We are happy to have any conversations after this conference ends with anyone who's interested. And that's it. Thank you to everyone and I think we actually finished a minute or two early.