My small hometown of Greenwich, CT has the largest municipal government in the United States. 260 elected members across twelve districts act as a collective legislative body to ensure a high standard of living for all residents. Local governments such as this date back to this country's founding and are a cornerstone of American democracy. Unfortunately, they are also the most forgotten.
In 2015, Dallas saw a less than two percent voter turnout in adults aged 18-34 for their mayoral election. In Portland, Baby Boomers and the Silent Generation (aged 65 and older) were found to have 33 times the voting power as Millennials. It's time we revolutionize civic involvement and bring a voice back to America's youth and disenfranchised. It's time for Collective.
Collective is a secure and transparent portal for town residents to participate in local ballots. Unlike Doodle poll, these votes are real and can transpire in concrete legislation.
In order to register to vote, residents simply take a picture of themselves and their license as well as add a username and password to their account. Collective will ensure a match between the live headshot and driver license photo and read the user's license for their address and date of birth. This information is then used to identify a user and determine their eligibility for district voting.
As a registered voter, Collective users are able to participate with a real-time feed of district agenda items. They may sort proposed bills by the responsible district boards as well as analyze the results of past items for a holistic view of their local government in action. Our vision is that a user may access the Collective portal anywhere in the world to efficiently educate themselves and vote on the issues they care about in a matter of minutes.
How we built it
User authentication was implemented using Kairo’s facial recognition and Google Cloud’s machine learning API. To verify a person’s identity, we required the user to take a selfie, which was then cross-referenced to a database of eligible voter’s pictures. Kairo accurately performed the cross-reference and provided a confidence level based upon similarities to their actual headshot. We resolved that 80% confidence would be stringent enough for the user to be confirmed.
Next, to check voter eligibility, we had the user take a photo of their driver’s license which was then analyzed using Google Cloud’s machine learning API. We had Google Cloud’s API extract Name, Address and Date of Birth and photo. We again used Kairo to ensure that the photo the user took before and the driver license photograph matched. Then, we used the extracted values to ensure that the user was over 18 and lived in the district of voting, thus ensuring the user is eligible to vote. In addition, the values extracted from the drivers license are used to populate the registration form that the user must fill out. Once the user is verified, we used Cloudfare technology to improve general security and enhance performance.
Challenges we ran into
Our biggest challenge was integrating all of our respective workflows into a complete, clean and intuitive user interface within 36 hours. Accurate computer vision for user verification and account creation was instrumental to the app's practicality as a voting portal and posed unique challenges in it's implementation. Specifically, we encountered and resolved issues configuring the uploaded images and connecting them with the Google Cloudvision API. Incorporating blockchain hashes as a vessel for secure voting was another difficult yet worthwhile hurtle for our group. We knew the application was possible and luckily had some helpful sources for constructing the final JS framework.
Accomplishments that we're proud of
The Collective team is proud to have built a functional political portal without the clutter and dreary UX of a typical governmental website. The face and text recognition for registration and sign in also work, this was simply the biggest technical hurdle. We passionately believe Collective can be implemented by municipalities and student governments for legislation and see it as a critical step to maintaining a healthy democracy. This is a platform that will have especial value to college students who may feel disheartened by today's political climate and unaware of the action they can take. It is an application that will inform and empower individuals and rebuild faith in America's democracy.
What we learned
On the backend, our team learned a lot about Google Cloud's suite of APIs and data storage tools and how to use them for a image recognition-based application. Collective offers implementation of cutting-edge Blockchain technology and demonstrated to our team its true value for secure voting. Lastly, it was shocking to read some of the recent statistics on disparities in civic involvement between different demographics in the United States. It is clear that wealth, race and access to voting resources play a unjustifiable role in who has a voice in municipal government.
Applications in combatting FAKE NEWS
There are two core principles behind battling "fake news" in politics: i. The need for credible and traceable sources ii. Transparency in the democratic process Collective removes the noise around policy with unbiased and credible information straight from Town Hall. This is the last thing our users see before they vote and it emphasizes the objective community-centered purpose of the legislation being proposed. Additionally, by dramatically increasing voter participation and education, we reduce the influence of the less-moderate, outspoken activists that tend to dominate the current local government ecosystem.
What's next for Collective
Municipalities are a great beachhead for revolutionizing civic engagement, however, we see the ultimate goal as expanding to the state and federal stratums. Scaling a secure blockchain voting system will be paramount to this expansion.
Other future applications of Collective include a portal system specialized for student government, a mobile version for Android and iOS, and incorporation of a natural language processing API to parse bills and extract keywords for an optimized user feed and cross platform notifications based on interests.