Post date: May 08, 2021 11:48:27 AM
We presume that computers reduce numerical errors. <<But are there any ergonomic studies that demonstrate this?>> But humans, being humans, still make plenty of mistakes which result in a lot of trouble including financial losses and human fatalities. Humans need help from algorithms to detect and correct their mistakes, and to prevent them in the first place. And they should get more of this help than they currently do.
The attached presentation (below) argues this point using two examples where better computers and algorithms could substantial improve outcomes. The first example is the prevalence of errors and disasters arising because programming languages can’t detect unit mismatches and dimensional errors. The second is the persistent error of 'misunderstimation' of uncertainty. Humane programming languages could fix both of these fundamental problems for us.
See also the general argument about a better computing language. See also human errors research. See also units. See also underestimation of uncertainty (advanced bias correction).