Tue, Sep 8
How Does the World Become Data?
Please complete the tasks and readings listed below before class on Tue, Sep 8.
Optional Readings
Each card below names a recurring pattern in how data is made, reused, and treated as if it simply mirrored the world. Click a card to preview it – then ask which choices in representation made that pattern possible.
Topic / Focus
What can a computer “see,” and what has to happen before something in the world becomes data? We will look at several ways computers represent information – numbers, color, text, images, sound, the environment, and eventually people. In each case, the computer does not receive the thing itself. It receives a representation: selected features encoded as numbers according to some scheme.
The central idea is that data representation involves choices. What gets captured? What gets left out? And when the thing being represented is a person, what are the consequences of those choices?
This meeting is the first of three: data representation, then procedure, then learning.
Slides & Activities
Guiding Questions
- How do things in the world become numbers a computer can work with?
- How does a computer know what those numbers mean?
- What information does a representation preserve – and what does it leave out?
- Who decides what gets represented when the data are about the world; about people?
In This Class
In this meeting, we will:
- experiment with binary and several encoding schemes for numbers, color, text, and images
- trace how a physical phenomenon such as sound can be sampled and represented as numerical data
- distinguish the thing being represented, its numerical representation, and the encoding used to interpret those numbers
- examine how changing a representation can preserve some information while losing or simplifying other information
- use a mortgate lending case to ask what happens when people are represented through institutional data
Closing Reflection
Choose something in the world that we might want a computer to understand.
What could be represented easily? What might be difficult to represent – or disappear entirely – once it becomes data?