I think this is closer to my point, which was that some ideas can only work at scale. That doesn't mean they will work at scale. But having them fail small does

I think this is closer to my point, which was that some ideas can only work at scale. That doesn't mean they will work at scale. But having them fail small doesn't prove or disprove anything. So you come up with a federated database model for your company with 30 employees, and find yourself in dev hell and wasting resources on an expansion that could have been solved with a slightly larger monolith. That means you chose the wrong tool for the local job. It doesn't mean the other tool couldn't be effective for a larger job.