Completely agreed. Here's another way to look at it from an economic perspective: It's usually best to focus on propping up your differentiation and competitiv

Completely agreed. Here's another way to look at it from an economic perspective: It's usually best to focus on propping up your differentiation and competitive advantages, going stock with everything else. If you rely on standard components for things that aren't your special sauce, then you get to ride market forces of commoditization which will drive the cost of those standard components towards zero. Can you build your own backend database in Haskell instead of using Cassandra or PostgreSQL? Yeah, that would probably be really cool and do well on these boards. Then look back at how many hours you spent getting that to just understand simple SQL, and how many customer-facing features that could've been. Is it cooler to build that database than to set up an email auto-mailer? Yes! Special functional algorithms that operate on B-trees are far more exciting than simple state machines that send emails. But the latter is what the customer cares about, so therefore that's what you care about. One interesting exception to using cutting-edge stuff may be making it easier to hire cutting-edge people. It doesn't matter how smart you are. The opinion of the market is the only one that matters, and it expresses it in the form of repeat sales.