Report Out from the Earth Index Offsite
The Earth Index team met this month in Washington D.C. for our offsite. Since we are a remote organization, “offsite” meant in person without screens in between us!
Momentum for Earth Genome’s visual search tool for the environment has been building thanks to recent funding and exciting partnerships. After spending a couple months researching and developing our technical infrastructure and user interface, it was time to gather in person to jam on the direction of our product, set some technical milestones, and bond as a team.
Our primary goal was to align on 6 month-out feature requirements for Earth Index that would make it a useful tool for indigenous groups who want to monitor and protect their land.
We are sharing outcomes from this gathering in the spirit of working in the open. By working in this way, we amplify our progress and catalyze new work in this space. Already, this openness has led us to new insights and partnerships that we otherwise would have missed out on.
Our first day in D.C. consisted of staggered arrivals, co-working at our host space, Eaton Workshop, and a conversation with Conservation Metrics / Sky Aelans about how Earth Index could be used to find new roads in Malaita in the Solomon Islands. The team finished off the day with Ethiopian food and conversations about camping, international grocery stores, and home-brewing.
The second day, we got into the bulk of our planning. After reflecting on the trajectory of Earth Index up to this point, we spent time aligning on our prioritized users based on interviews we’ve conducted thus far:
- Indigenous community / community based organization members
- Geospatial and data analysts
- Data journalists
We discussed these personas’ proximity to the indigenous communities in which they seek to have impact, their familiarity with the technology underpinning Earth Index (machine learning and geospatial data), and their biggest pain points.
With these users in mind, we took a look at our most recent wireframes to get insight into the design process. We engaged in a challenge to imagine and sketch how we might make the experience of searching satellite imagery in Earth Index “delightful.” This yielded some interesting ideas. Should Earth Index have a search button or should search results come back continuously and automatically? How could we use color or animation to signify the confidence we have in a given search result? Should we show a view of the region of interest in vector space, where image tiles are collected not by geographic proximity but by visual similarity? We don’t have all the right answers, but we’re excited to try some of these ideas out.
With the user and design context set, we started to hone a product roadmap with alpha, beta, and v1 milestones. In the alpha version of Earth Index, we will focus on nailing the experience of searching satellite imagery for a single time period. In beta, we want to introduce comparisons between satellite imagery from two time periods. And by v1, we will support monitoring in the form of comparisons across several sequential time periods. Throughout all of these phases, we will focus on searching sentinel-2 imagery with DINO embeddings only. We will bring in a small number of trusted users for usability testing after alpha, which we hope to launch by mid-May, and will gradually grow our user base from there.
To celebrate reaching this goal, we met up with friends and collaborators Ian Schuler and Dan McCarey, grabbed (probably a bit too much) pizza, and winded down for the evening before our last day.
The main task for the last day: get all of the decisions organized and documented. We started by breaking down our milestones into feature lists. We then broke these down into tasks by role (backend engineering, frontend engineering, design, product), level of effort, and more. Our formerly jumbled Asana page for the project was beginning to come into focus.
In our remaining time, we tossed around big ideas to consider beyond this 6 month time period. Should we seek new satellite imagery sources? Can we build out support for users to be able to create and share custom map views in Earth Index? Should we get our own satellite?
This time together gave the team a lot more clarity on where we’re heading and what we need to do together. We’re all looking forward to the next one!
