Engineers at heart wouldn't let a single point of failure in in the design of it to let a pin drop unexpectedly. Is the vehicle insured for a do-over should it fail to launch unspectacularly and land in the water?
If the HARD requirement is "you have this much mass and volume to make a thing to a particular confidence score" you make the damn thing and hope you are on the correct side of the dice.
No single engineer gives a shit about the entire system because no single person is equipped with the human ability of comprehending each individual part that has whatever confidence rating that is required.
The solar shield is like rolling out five layers of tightly packed tinfoil from several points without tearing. Claiming that you would do away with single points of failure if you were the one designing it is too easy to say on HN
There's also the whole "what if we could avoid mass and volume constraints" tangent, that was discussed in the yesterday's Starship article, but it's more of a theoretical consideration for now.
So, what I learned is that single points of failure on initial deployment -- which by definition only has to run once -- is a drastically different cost vs risk calculation than something that happens during normal operations.
You are right that a magical rocket could solve many of NASAs problems (just launch ten Webbs and use the ones that don't work as sunshades for the rest!) but I think NASA is being pretty practical in not waiting for one to be invented.
What was the original budget for JWST? Where's it at now? From your assessment and the numbers, it sounds like there's not just double but triple redundancy now.
Still though, if everything works and we learn more about the universe, I think it'll have been worth it.
(Ok, "weeks" isn't quite applicable in that case... Unless you're counting the time for updates to propagate through all upstream dependencies.)