The act of translating the world's complexity into a computable format and structure inherently leads to compression: in which the territory is compressed into a map. This is why Facebook presents us with such a tragically lame abstraction of the richness of human social life.

At small scales or with narrowly scoped "translations", this is actually fine. Accounting software is a good example of this, as is music production software. Even email seems to be ok.

But at scale, any computing system that interacts with the wider world - such as a global transportation system (Uber), a home-based hospitality company (Airbnb), or egalitarian content delivery system (YouTube) will be forced to see its map interface with the territory. The results are never good.

Have you seen what happens when you take a screenshot, print it out and scan it back into your machine? Maybe not much at first, but over time more and more artifacts - noise - will appear. It's called generation loss and eventually, as entropy gains an ever-increasing foothold, nothing will be left but this digital gibberish. This is what happens with these computing systems at scale. As these systems enter into a cycle of ingesting and translating (compressing) the world's complexity, they then output their distorted model for the world to deal with.

Global Generation Loss: Compression & Complexity in Computing Systems
Alex Singh
Actions
Connections