<aside> 🐘 We believe cities can be beautiful, safe places with abundant housing, amazing education, and no poverty. Civic Abundance is the desire to make all of those things happen and the belief that all cities can become like that if communities take the right actions towards those goals. We aim to help communities realise civic abundance by building coordination infrastructure that lets communities build consensus on shared goals, track how they are progressing on those goals, and determine the best actions to reach those goals.
</aside>
By Andre Vacha (Design/Data Sci and Frontend) and Finn Macken (Full-stack/Data Sci and Strategy), Founding Engineers, Roote Foundation.
We make what we measure
The goal of this write-up is to provide a high-level overview of CivicDash. Please provide feedback at any level of analysis — concept, UI/UX, aesthetics, mode of delivery — as we aim to start development with this current iteration.
Interface Teaser for CivicDash, a chrome tab extension to realise Civic Abundance.
<aside> 🎓 CivicDash is a community-driven platform for proposing and evaluating solutions to realise civic abundance.
</aside>
At the highest level, this project has one goal: to create a dashboard that surfaces and fosters discussion around metrics that affect San Francisco as a city. The core philosophy of Civic Abundance is that civic change is most effectively accomplished when contextualised through data, and our role is to make that data as accessible as possible.
We are far from the first people to try and resurface data in San Francisco. Most notably, the San Francisco Scorecards track important metrics, framing them against an ontology of “themes”: liveability, public health, public safety, and so on.
Two keyframes from SF Scorecards, an adjacent but passive civic dashboard.
Civic Abundance is not a zero-sum game, but we wanted to differentiate ourselves from existing dashboards as much as possible, both so that we can contribute to the ecosystem meaningfully, and as a proof of concept for the kinds of civic engagement that we want to see in future dashboards.
We did this in three main ways, all of which are discussed in detail below. Briefly:
Many-to-Many relationship between Themes, Problems, Indicators and Solutions
CivicDash is organised around four main categories of information: Themes, Problems, Indicators and Solution Markets (a kind of prediction market — see below). These categories are all interconnected, but at the same time, there’s a clear flow that moves from broader themes, to specific problems, to the indicators associated with them, to the solutions that address them. Our goal is to encourage movement from higher to lower scales of analysis.
<aside>
🌲 Theme: A set of opinionated problems in line with the vision of civic abundance. In doing so, provides a high-level overview of the city.
e.g. (Abundant Housing), (Urban Sustainability)
</aside>
<aside>
🌿 Problem: A specific, opinionated issue facing San Francisco, contained within a Theme.
e*.g. (Abundant Housing → Homelessness), (Urban Sustainability → Waste)*
</aside>
<aside>
🍃 Indicator: A quantitative description of a problem developed and maintained by us as a source of truth.
These descriptions are visualisations of modelled datasets (including geospatial) by default, but can extend to predictive and explanatory models later on.
A Problem (homelessness) can be described strongly (number of people experiencing homelessness) or loosely (number of vacant homes) by Indicators, but it is measured without ambiguity.
e*.g. (Abundant Housing → Homelessness → Number of Vacant Units), (Urban Sustainability → e-Waste → Tons of e-waste Generated Per Month)*
</aside>
<aside> ☀️ Solution Market: a prediction market constrained to Solution Proposals and Solution Evaluation to Problems. We support two forms of prediction market: binary (for solution evaluation), and categorical (for solution proposal).
*Binary Market: e.g. Will the SF government’s new Affordable Housing policy achieve its goal of creating 5000 new single resident occupancies by 2025?
Categorical Market: e.g. San Francisco has a homeless population of 10,000 people, but 40,000 completely vacant units. Why can’t we address homelessness by matching these people with units?*
</aside>
There is an ethical side to civic dashboards that is almost never addressed: they are fundamentally normative. They make judgements about what constitutes an important problem, an effective indicator, a reasonable solution. They often take the philosophical leanings of their creators and present them as objective truth. For some things, this is relatively uncontroversial: homelessness is bad, high education levels are good, CO2 emissions are an important indicator of the sustainability of a city.
Very often in our research, however, we’ve found that it’s not that simple. Is the mean price of an apartment a good measure of affordable housing? Is the emotional fulfilment of citizens something that should be explored alongside issues like public safety and infrastructure development? This is doubly true for solution evaluation. People devote their entire academic and legislative careers to studying the most effective responses to endemic social problems.
Prediction markets, outlined below, offer us a way out of some of this normativity by creating a framework for group decision making. They allow us to aggregate the beliefs of an entire community to create a normative framework.
There is a version of this application that completely delegates the normative parts of the dashboard to San Franciscans. It would ask them to identify the most important themes and problems, the most relevant indicators, and the most promising solutions. There are two issues with this. The first is that we want to create a motivated user-flow with a single call to action. The second is that Roote already has a clearly-defined philosophy that we want to represent. We want to be opinionated.
As a result, we don’t ask the community to identify important problems. We present our worldview explicitly through our categorisation of themes and issues, and acknowledge that our dashboard is not an objective source of truth. The role of the community is instead to identify and evaluate solutions, leveraging their expertise as citizens of San Francisco.
<aside> 🌲 In other words, we are normative (opinionated/top-down) in defining the problems facing San Francisco, but un-opinionated (pluralistic/bottom-up/community-oriented) in our approach to finding solutions for those problems.
</aside>
Prediction markets are a core part of this dashboard, because they provide a systematic, research-supported mechanism for aggregating the insights of many different people.
For a detailed explanation of prediction markets and their effectiveness, you can check out the resources at the bottom of this document. In short, they use financial incentives (often play money, which has similar effects as real money) to turn event predictions into markets.
They have two main criteria:
An example of a binary prediction market, hosted on Manifold Markets. Participants bet YES or NO on the market, specifying an amount of faux money to stake with their bet.
We care about prediction markets because we need a systematic way of polling citizens for their opinion, and market providers like Manifold and Kalshi have a publicly accessible API that we can leverage, as well as research supporting the effectiveness of their frameworks.
<aside> 🌲
</aside>
Example of a binary prediction market related to a political outcome
We expect these to be nuanced prediction markets around existing policy solutions, à la approval ratings, albeit with better incentives for accurate representation.
<aside> 🌲 Categorical prediction markets will be used for solution proposals.
For example: Which of these solutions will most effectively address the problem of Z in the next year?
</aside>
Example of a categorical prediction market used to propose solutions. Users suggest their own categories over the course of the market.
In categorical prediction markets, people state claims about what they believe is right, and voting surfaces the “right” or dominant claims. The current probabilities of the market resurface “metrics of belief” associated with the indicators in focus.
Modern dashboards tend to look like this:
Template examples of passive dashboards, taken from Dribbble. Despite wildly different aesthetic choices amongst these designs, notice the commonality of a confrontational user flow, where the interface broadcasts a set of tightly packed widgets. “What do we do with it?” is a tricky question to answer with these, especially for the ‘average’ SF resident.
They’re pretty, but completely unstructured. There’s no user flow, no call-to-action. They’re crammed with data. We want to build a dashboard that challenges this approach and creates an opinionated user flow. CivicDash should teach people how to engage with it, and should provide a hook to keep people coming back to check it, again and again.
We defined the concept of an affective loop to clarify this behavioural paradigm. Affective loops have three parts: