The Following text is an article written by OpenAI’s GTP-3 Ai. It was trained on my previous blog posts “A Pattern Language” and “Architecting Data”:
My favourite book is “A Pattern Language” by Christopher Alexander. The book is an entire system of architecture. There are 253 patterns. Each one describes how to create a great experience in buildings and cities. Each one is both generically applicable and specific to a particular context. Each one is a simple idea that can be used to create something very complex.
In the world of data and information we see many of the same themes – the tension between the generality and the specificity of things, the tension between the complexity of the real world and the simplicity of a representation, the tension between how things are and how we wish them to be.
Data are both generically applicable and specific to a particular context. The problem is that these two ways of thinking are fundamentally at odds with one another. Data can be used to make simple claims about the world, for example that ‘users who take this action are more likely to take this other action’, but this is at the expense of complexity. Data are not the world, but a representation of it.
Data are both complex and simple. It is possible to use data to predict the future or to understand the past. However, this is at the expense of context. Data are not the world, but a representation of it.
Data are both simple and biased. The problem with the generality-specificity tension is that you can’t have both. If you want to make a grand claim about the world based on a set of data you need to make a simplifying assumption. Any simplifying assumption that you make will inevitably lead to bias.
Data are both biased and neutral. This isn’t to say that there is no place for generality and specificity in data. It’s to say that data is an inherently complex thing. It’s to say that data are both generically applicable and specific to a particular context. It’s to say that data are both complex and simple.