browser · idea · project

Nicky Case's explorables

Hand-built playable explanations of systems (Parable of the Polygons, The Evolution of Trust).

inspireevidence: mixedby Nicky Casencase.me ↗

Nicky Case’s explorables

  • Maker: Nicky Case
  • URL: https://ncase.me
  • Status (10 October 2026): an independent creator funded by Patreon. Best-known works are Parable of the Polygons (2014, with Vi Hart), The Evolution of Trust (2017; ncase/trust has 6.3k GitHub stars, CC0) and To Build a Better Ballot. Recent work has moved to AI-safety writing (“AI Safety for Fleshy Humans,” 2024 to 2025) and comics; the most recently pushed public repo is civic-ai-comics (July 2026). They also made Nutshell, a tool for expandable explanations. We couldn’t confirm a new explorable in 2025 or 2026.

What it is

Playable explanations of systems: segregation from small biases, trust as an iterated game, anxiety, voting methods, epidemics. The reader runs a simulation, changes the rules, and sees the result, with a narrative guiding them.

The problem it’s solving

Systems behaviour (emergence, feedback, equilibria) is hard to grasp from text or a static diagram. Letting people play the system builds intuition that prose can’t.

Its path / bet

Craft: one person writing, drawing and programming each piece by hand, released free, often public-domain. Case also curated the explorable explanations hub that collected the genre.

How it works (concretely)

Hand-built HTML, JS and canvas pages with simulations and step-by-step narration. Each takes months.

Strengths

  • Proof that interactive explanation can reach millions of people.
  • A pattern worth copying: narrate, then hand over a control, then ask the reader to predict.

Weaknesses / limits

  • Months per piece, by a rare multi-skilled author. The output rate is a handful a year at best, and has slowed as the author moved to other forms.
  • No controlled evidence that the pieces teach better than good text; their reach is the evidence.

Relation to fictty

Inspire. The pattern (narrated steps, a simulation, a control, a prediction) is what a good agent-made teaching screen should follow. The cost per piece is what agents change.

Could fictty adopt it instead of building?

No. These are finished artefacts, not a toolkit. Nutshell is a small browser tool for expandable text.

What fictty should take from it

  • Predict, then reveal. Ask the person to guess before showing the result; read the guess back. It’s cheap in fictty (a prompt node, watch) and it is the part of explorables most supported by learning research.
  • Narration plus one control beats a dashboard of controls. Keep teaching screens to one thing the person changes at a time.

Sources