libe blogging AMRO
The Evolution of Data Christo Buschek
He does data driven investigation. They won the Pulitzer prize last year.
Data has been using data since the 70s. They used to mostly do statistics. The new idea is data-driven where they use data to discover things.
Human rights research is also often data driven investigation. This is a collaborative process involving technology, design, etc
The process is preservation, exploration, verification and narration
He and some collaborators tried to discover how many people are imprisoned in China in Xinjang. They found camps big enough to hold a million people.
Theystarted knowing that some camps existed, but not much else. Journalists had no access to the region.
Baidu Maps is the Chinese mapping service like google maps. They censor some areas by loading the image and then loading a white box over it. So the journalists decided to first look at censored squares.
They had three types of tiles. Satellite pictures. Watermark images that have no data. And censorship images.
he wrote a script that emulated a human noodling around and ran it across a cluster.
Of 50 million tiles. 5 million were masked. They decided to look for infrastructure that could support camps. This reduced the tiles to 800k.
The next step was verification. This is a human process that takes time. Verification subjects data to due process. It is a creative process.
He wrote a tool for researchers to annotate metadata. They built a rubrick as they looked at the data.
428 locations bear the hallmarks of prisons and detention centres. Around 2017ish structures became permanent.
they had three categories: certain, like and unsure. Certain ones had external verification.
Their output was articles and the data set which was then also used by others. A local activist/ journalist travelled to some of the camps and filmed them.
Christo built the method and tools. He says that humans shapes tools and tools shape humans. He says technology is not neutral. Toolsencode systems of values. He notes that Silicon Valley is extremely ideological. Their tools do not reflect our values. We must make our own tools.
Our tools and systems reflect our social structures. Our communities are organised as open-ended gatherings. Collaboration is central. Tools are.nodes in networks.
Any tool is a positive choice and also a negative trade off of what it can’t do.