I think you are wrong. The reality is that It is an order of magnitude more efficient and safe to take the shotgun approach. We can stay here and try and push the half-life of our civilization out with ever decreasing returns or we can send out a _lot_ of bad attempts to colonize asteroids, moons, and planets and see much stronger gains to our species' half-live via redundancy.

A decent metaphor is that trying to make 1 system more reliable quickly becomes less cost effective than making a network that increases reliability via redundancy. While we have all of our life on one planet all our failures are essentially total. Have life on many planets limits the impact of catastrophic failures/random accidents.

Suppose your entire civilization was running in a single Docker container. Being a SPOF, that's not ideal. Now, consider scaling out horizontally by running on a cluster of a dozen planets. That provides a bit more breathing room for your devops team.

The website highscalability.com is a good resource on this topic of fault tolerance and availability. It turns out that a lot of what applies to keeping services like Google running applies to the entirety of the human race.

Species != civilization. TO treat humanity as a monoculture is to skate over some rather yawning chasms of philosophical disagreement between different groups of people and to assume they've been resolved in your favor.