Month: December 2020
5 min talk on “What model data does one need to preserved to ensure replicability”
Good practice: TRACE – the equivalent of a lab notebook for modellers
Abstract: The acceptance and usefulness of simulation models are often limited by the efficiency, transparency, reproducibility, and reliability of the modelling process. We address these issues by suggesting that modellers (1) “trace” the iterative modelling process by keeping a modelling notebook corresponding to the laboratory notebooks used by empirical researchers, (2) use a standardized notebook… Read More Good practice: TRACE – the equivalent of a lab notebook for modellers
Podcast by Tim Harford on the need for algorithms to be effectively critiquable
Starting with a nice illustration as to the difference between alchemy and chemistry (the former was conducted in secret), leading to a critique of a similar situation with Big Data and how this is used by big internet corporations. Available at: https://www.bbc.co.uk/programmes/m000q3sr
Nature: 5 rules for communicating evidence
A nice short piece basically advocating complete honesty in public pronouncements: “Five rules for evidence communication. Avoid unwarranted certainty, neat narratives and partisan presentation; strive to inform, not persuade.“ https://www.nature.com/articles/d41586-020-03189-1