> I think the false premise here is that non-experts can build on the work done by experts. I'll take a different side to that argument - it is dependent on th

> I think the false premise here is that non-experts can build on the work done by experts. I'll take a different side to that argument - it is dependent on the expert to establish the bounds and rules of change to a system. How a system naturally grows from people with less context is one of the architectural traits of the system that can be traded on. I have often seen systems which can ONLY be worked on by the so-called "expert" who wrote it. Even when you hire a new "expert" level person to work on it, it takes forever and success is not gauranteed. On the other end - a system that actively works against doing things the wrong way. It has an established core idea with clear boundaries between the roles of interlocking parts. For the intended usecase of the system, it is as extensible as it needs to be, while still remaining within the bounds of the intended usecase. It does now allow a vacuum cleaner to become a microwave oven, but it does let you add custom attachments, modify and swap out drive systems, adjust suction power, and change body style and hoses for the vacuum. Those can absolutely be done by non-experts, and the system is still very malleable. Think of all the code on github for libraries people use every day that is added by non-experts, or hobbyists, or people who are new.