Buildings are made up of bricks and steel, wires and fuses. The internet is also constructed. It is made up of material and immaterial infrastructure — from servers to fiber optic cables to code. The underlying architecture of the internet has an impact on how internet users can engage with the internet as a space or a resource. Importantly, much of this architecture is effectively invisible. We don’t always know where the server that hosts a website is located, and we rarely know what code is running behind the scenes to generate our search engine results. Like the buildings we construct, the internet is the product of human decisions about how it should be. So, it matters who’s making those decisions!
Since we usually can’t see the infrastructure of the internet, any biases built into the system are often similarly invisible. But biases exist. Most servers (that host websites and store data from internet users) are physically located in the United States or Europe, even though three quarters of internet users come from the global south. Some regions or countries are more ‘connected’ than others due to long, complicated histories of either privilege or exclusion, with geographical consequences. Most computer scientists and engineers are (still) men, and many algorithms written to perform some of the most basic functions of the internet demonstrate tendencies toward racism and sexism. As we start to build algorithms that are the foundation to machine learning (so-called ‘artificial intelligence’) — we have to ask how we’re teaching machines to «think» and «learn» and who is doing the thinking and learning.
Questions We Care About:
- What does the geography of the internet really look like?
- Who builds the internet we see and use everyday?
- What biases are built into the invisible architecture of the internet?
- We can’t make code neutral (because humans write it!), but how can we humanize it?
- How can we imagine a more distributed and autonomous public internet infrastructure?