Thanks jppope. I like the idea to talk about 'results of the work' and you can ask probing questions about why it worked. If they can describe logically what th

Thanks jppope. I like the idea to talk about 'results of the work' and you can ask probing questions about why it worked. If they can describe logically what they did and how it linked to those results, describe what they are certain and uncertain about it will probably be hard for a liar to fake all of that (the liar will be confident of everything). "Results" is a funny thing to define though. For example say I know nothing about coding, and I hire someone to make a website. Based on their "results" they make good looking fast websites without bugs. I hire them. I get my site. I need a change and suddenly they are not available or got expensive. I ask someone else to do it and the say "maaan, wtf they have used Perl and a weird JS library for the front end, and purescript and I don't know how to maintain that. I'll need to rewrite it". With machine learning, to evaluate the results, I think you'd need to know a bit about machine learning and what P(good work by this person | company made money from the model) is vs. P(null | company made money from the model)