I like the mental model of "innovation points". You have a certain budget of innovation points that you can spend while building your project. If you spend too

I like the mental model of "innovation points". You have a certain budget of innovation points that you can spend while building your project. If you spend too much (i.e. use new shiny things throughout the stack), you're going to be chasing after bugs everywhere and won't get very far. If you spend too little (i.e. use only old stuff), you'll have a hard time implementing real innovations. The idea is to spend innovation points strategically, i.e. use new shiny components (like a new programming language or a new DB) only where it actually brings you a tangible benefit, while using tried and trusted components for everything else. (The concrete budget of innovation points varies per project. A government project will have way fewer innovation points than a weekend side-project.)