Has a lot to do with Goodheart's law as well.<p>But basically anything moderately complicated allows for this mathematical bullying. "It's complicated, the experts have said xyz, don't you try to deny it!"<p>Anything moderately complex is also going to be infused with judgements that vary in how well justified they are. VAM models like they are talking about in the paper are like this as well, and require discussion to understand. The risk is you make everyone think the evidence is a lot more certain than it is.