4 ms·
My way of expressing the same thing would be something like this (I'm not sure if it is an improvement, though): Old methods are not automatically bad because
by Natanael_L 10y ago
My way of expressing the same thing would be something like this (I'm not sure if it is an improvement, though):
Old methods are not automatically bad because you don't understand them or why they are used. New methods are not automatically good.
If you want to change something, try to prove both that your change isn't worse in any important way and that is notably better in some meaningful way. Understanding where the original method came from helps you prove the right things.
- baldeagle 10y agoAnd that is the root of the article's thesis: the current regulator has basically said that if we had technology X back in 2007, none of this would have happened; this idea is a dangerous presumption against the adaptability of the system based in partial models.