Spot on! I’ve certainly drawn a lot from Taleb… You’re right in that I should reframe this a bit: a distributed system may end up looking inefficient, but in its avoidance of failure, is actually optimal. The dichotomy I present is more about perception than outcome. That is, we perceive spending on resilience as “inefficient”, but to your point, systems that don’t take the tails into account can be “efficient” for a decade, and then turn out to be wildly inefficient and costly due to a catastrophic failure.