> If you want a better system, it's better to make as many changes as possible to the system so that problems are impossible and/or unlikely. I disagree with t

> If you want a better system, it's better to make as many changes as possible to the system so that problems are impossible and/or unlikely. I disagree with this premise. If you want to make a system better, you need to enable as many information flows as possible, shorten feedback loops, etc. The classic Meadows leverage points. The proposal of making cars noisy has the opposite effect, by sweeping problems under the rug. Just "doing as many things as possible" is a great way to paint oneself into a local optimum.