Another complementary idea: if CDC typically issues "recommendations" and not "regulations", I assume that these measures are outside the typical federal regulatory agency requirements for cost/benefit analysis and notice/comment process to justify regulations. (I admittedly haven't done detailed research into the topic to verify that's the case, but it fits with the framework of how federal agencies operate.)
So that would imply the development of a culture and processes at CDC with meaningful differences from other agencies. I'm not saying that cost/benefit and notice/comment are perfect. Cost/benefit depends heavily on assumptions that can be changed to get the desired answer, for example. But it does at least force an agency to have some mindset of doing the work to acknowledge and evaluate trade-offs (cost/benefit) and to justify measures against criticism (notice and comment).
Assumptions can be changed to get desired outcome, but you at least have to model them out and document them - it them becomes an exercise of "what would we need to believe for this to be true?"
Good points here about the CDC culture.
Another complementary idea: if CDC typically issues "recommendations" and not "regulations", I assume that these measures are outside the typical federal regulatory agency requirements for cost/benefit analysis and notice/comment process to justify regulations. (I admittedly haven't done detailed research into the topic to verify that's the case, but it fits with the framework of how federal agencies operate.)
So that would imply the development of a culture and processes at CDC with meaningful differences from other agencies. I'm not saying that cost/benefit and notice/comment are perfect. Cost/benefit depends heavily on assumptions that can be changed to get the desired answer, for example. But it does at least force an agency to have some mindset of doing the work to acknowledge and evaluate trade-offs (cost/benefit) and to justify measures against criticism (notice and comment).
Assumptions can be changed to get desired outcome, but you at least have to model them out and document them - it them becomes an exercise of "what would we need to believe for this to be true?"