Design in the Era of the Algorithm

Favor accuracy over speed

Getting it right matters more than getting it fast. Rushed predictions cause real harm.
Allow for ambiguity

Not everything can be categorised cleanly. Design systems that handle uncertainty gracefully.
Add human judgment

Algorithms should inform decisions, not make them alone. Keep humans in the loop for consequential choices.
Advocate sunshine

Be transparent about how systems work. Users deserve to understand what’s happening with their data.
Embrace multiple systems

No single algorithm has all the answers. Combine approaches and allow for diverse inputs.
Make it easy to contribute (accurate) data

Good outputs require good inputs. Design for data quality, not just data quantity.
Root out bias and bad assumptions

Algorithms inherit the biases of their training data and creators. Actively seek and correct these.
Give people control over their data

Users should be able to see, correct, and delete their data. Control builds trust.
Be loyal to the user

When business interests conflict with user interests, side with the user.
Take responsibility

You can’t blame the algorithm. Designers and organisations are accountable for what their systems do.

Favor accuracy over speed