Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Lumping and splitting; us and them: How not to be evil with categories

Lumping and splitting; us and them: How not to be evil with categories

As embodied humans we spend an incredible amount of brain power categorising everything we come into contact with - it's our way of dealing with huge amounts of detail.

As designers, we take this into our work - using categories to make shortcuts and help get our users to the thing they need.

This talk is about how categories really work and the consequences of using them:
- How we think categories work is quite different to how they actually work in our brains
- The consequences of grouping (lumping) and separating (splitting) people for design and policy can be unforeseen, dangerous and evil

5d79f73bf83f3675bbc6dbbcb3b1ba81?s=128

Donna Spencer

August 13, 2017
Tweet

Transcript

  1. Lumping, splitting; us and them: How not to be evil

    with categories
  2. Quick activity: Yell out: Examples of games

  3. word cloud

  4. Classic category theory (how we think we think) A definition-based

    model Bird = feathers, a beak and the ability to fly Categories are based on shared properties All birds share these properties Containers with boundaries - things are in or out It’s either a bird or it isn’t
  5. Modern category theory (how we actually think) Categories are embodied

    There is no ‘proper’ way to categorise Many things exist in a continuum Including age and gender
  6. Boundaries are fuzzy, overlap & change over time Members may

    not have anything in common Easy to describe, hard to define ‘Game’ is easier to represent with examples Prototypes: Some things belong better than others Modern category theory (how we actually think)
  7. Modern category theory Basic level: Where we think Animal Dog

    Dalma*an Furniture Table Bedside table
  8. Consequences

  9. None
  10. None
  11. None
  12. None
  13. Other categories of evil • Targeted advertising (based on putting

    you in a category) • aka why I don’t tell Facebook I’m single • Insurance premiums based on age/postcode, not behaviour • First-degree price discrimination • Uber has been testing ‘route-based pricing’ • Pre-employment psychometric tests • US recidivism models
  14. How not to be evil Acknowledge that many categories are

    actually continuums Create services, policies etc based on actual behaviours, not group behaviours Always ask why you’re collecting a category and what it will be used for