Evaluating Insight, Not Just Computation
Evaluating a Knowledge Graph, instead of evaluating computations in higher education.
(untitled)
Personal website of Daniele Bonatto.
Latest posts
Interpolation, Extrapolation, and What Exams Really Measure
The distinction between pattern recognition and genuine discovery, and why understanding this gap matters for both students and machines.
Formal Verification and the Path to Machine Discovery
When machines prove theorems and why exams matter more than ever in an age of increasingly capable AI systems.
The Rise of Autonomous Systems - When Machines Write Their Own Code
How LLM agents are decoupling correctness from elegance, and what this means for the future of software development and human expertise.
Teaching a Graph of Knowledge as a Story
A reflection on teaching complex technical subjects as graphs of knowledge, drawing on narrative structure, threshold concepts, inquiry-based learning, and visual reasoning.
NHST vs. Bayesian Data Analysis, and what about machine learning?
A personal view on why Bayesian analysis feels more natural than NHST, and how it connects to machine learning.
Better Scientific Writing Prompts
A practical writing prompt trick to reduce LLM-generated random words in scientific writing.
Strong Types in Scientific Software - Safety and Pitfalls
How rich, domain-specific types in scientific software prevent subtle bugs and clarify logic, but can sometimes introduce hidden inefficiencies.
A Tribute to Ramanujan’s Genius
A tribute to Srinivasa Ramanujan, one of history's greatest mathematicians.
Rust Workspaces
An exploration of organizing Rust projects using workspaces, with a practical guide for structuring and managing dependencies.