amongst other misperceptions I think the entire aerospace industry has been colored in profound structural ways by high cost to launch and geostationary satellite design.
every engineering process was designed around creating multi-layered protections against component failure because there was such a high fixed cost to lofting those components.
If you're spending a $20k per kg to put the thing up, and in the event of failure your next chance to loft the asset will be year's away, then it makes sense to massively over-engineer the asset.
In low earth orbit, as well, you need thousands of duplicated assets with rated lives measured in years not in decades.
with starship and starlink, most of the cost is not fixed cost of launch, its in the payload bay.
Cost to launch should get below $100 per kg and you may be able to relaunch next week (if something goes wrong.) Cost to manufacture starlink v3, we think will start at ~$1,000 per kg. Associated groundstation capex for the incremental bandwidth is also much more meaningful relative to the cost of launch.
Net you are fine trading off reliability vs cost since you have to design for failure anyway (given thousands of assets) and sacrificing 1% reliability for 5% cost savings is a huge win.
This runs counter to everything that the industry has internalized over 50 years, so it has allowed SpaceX to continue to run orthogonally for nearly a decade.
With starship and v3 starlink we should begin to see the fruits of that labor really come to pass.