© 2020 Strange Loop
Puzzles take many forms depending on many cognitive abilities, like wordplay, lateral thinking, arithmetic, and spatial reasoning. A certain class of puzzles depends on mental model alignment, or internalizing a set of rules for predicting and explaining observations of a system. Leveraging mental models creates a spectrum of experiences between puzzles, which have handcrafted, usually unique, solutions, and problems, which specify what counts as a solution within a wide solution space. Problems whose solutions are algorithms are programming problems. We observe a pattern in successful techniques for building accurate mental models through interaction---specifically, informative failure---and discuss its application to three points along this spectrum: puzzles, problems, and programs.
Chris Martens is an Assistant Professor at North Carolina State University, where she directs the Principles of Expressive Machines (POEM) Lab, a research group concerned with the intersection of logic, creativity, and computation. She borrows tools from type systems, programming language design, and proof theory to enable authorship of generative systems, interactive narratives, and system models. While a postdoc at UC Santa Cruz, she led efforts on the Gemini game generation engine, and during her PhD at CMU, she created the Ceptre linear logic programming language.