Honestly I think ship now ship often deserves an asterisk by it. When you’re designing large and intricate back ends for a system that needs it, and you’re a PM who only sees things in terms of end-user features, it’s easy to set yourself up for failure trying to ship so fast that you accumulate technical debt faster than you can ever pay it down (because you underengineer the underlying platform but need to support it forever to meet SLA, and good luck arguing for a complete rewrite)

That is, no mantra above a certain level of abstraction should be adhered to too strictly. Sometimes it’s better to ship 3 months later if it means 3 years of way simpler support and scaling