Japan's Space One rocket explodes shortly after launch

SpaceX is only "flight testing" half of their rockets. The upper stages, which are equally as complex albeit smaller than lower stages, remain single use devices. And SpaceX has blown up plenty of rockets learning how to launch them. That a Japan has lost one doesn't speak ill of anything or anyone.

Boeing isn’t a great comparison for your point because their planes are reused for each flight and their recent drop in safety is not due to staff retention (or at least and drop in staff retention is another symptom, like their safety record, of the larger management problems behind Boeing)

Boeing isn't hitting generational turbulence, it's hitting the consequences of serious mismanagement over the past 2 decades. Right now is the least turbulent point in terms of availability of rocket engineering talent in the US, considering all the different companies out there that have built or are building rockets.

Or it could just be the low interest rates means it’s profitable to waste money like this for most parties.

Everyone managing durable technology deals with this issue, from government leaders deciding on nuclear-powered submarines to each person with their phone: What has greater ROI: building/buying a new thing or repairing and maintaining an old thing?

Both terms of the ROI (R/I) expression differ for the new thing and old thing: the return, new and old will have different features and performance; and the investment, build/buy vs repair/maintain.

Reliability is one cost - the cost of building or of repairing the machine to perform at reliability level X. Certainly, we manufacture very many reliable new things - most people expect their new car or phone or house to have lower reliability costs (i.e., greater reliability) than an old one. Old servers have higher maintenance costs than new ones, generally speaking. Famously, when you drive your brand new car off the lot and its status flag switches to 'used', its value drops by half.

Is there evidence that reliability in particular costs less on old rockets, or that there is any consistency in those numbers across rocket types? Rockets vary greatly between vendors.

And yes, I know about the cost advantages of reusing rockets - how could anyone on HN miss it? That doesn't mean that the used rocket is more reliable than the unused one, only that its reliability costs are low enough that the overall costs net out in favor of the reused first stage - and then only up to a certain number of uses. Will we ever have discussion of anything space-related without this endless SpaceX-is-better stuff?

You can't really test a rocket until you fly it. You can ground test subsystems and components all you want, but that's not a system-level test of a rocket. The only true system-level test of a rocket is to fly it. When you deal with volatile hypergolic substances subjected to extreme conditions all flowing through a complex system-of-systems that have never been through integrated testing together, you are going to have some spectacular failures. Doesn't matter how much money you have or how clever your engineers are or what a quality organization you've implemented.