How many changes did you merge in your latest dev build? Iterative engineering where the iterations are too long, and the changes too numerous and too risky <cough>"shower head"</cough> means you get masking bugs and don't converge on a shippable product. If the launch mount fails again, the subsequent iteration will be two years, not 8 months, and the cost will be 5X.
Compare this to SLS which needs about two years between launches to make changes.
Will be interesting to know how many of those 1k changes are tested.
That's what the next launch is for.
So just like Microsoft: testing in production.
It's a test flight
The key to spacex's success is that they are not afraid of having rockets blow up during development. They launch, learn, iterate, launch again. I'm not sure how any sane person can argue with their results.
Both approaches are valid- NASA is the “we have one shot to get this right” approach and SpaceX is the “launches for test data are cheap” approach. Both are valid