- a few basic axioms, based on principles well understood at certain scales but not universally accepted to work in combination and on a bigger scale

- a lot of very messy empirical data, such that you need to be well versed in statistical techniques to understand whether it supports a hypothesis or not

- predictions that are on a small scale frequently wrong and little agreement on magnitudes, despite scientific consensus over the general direction of movement

- a lot of debate over whether proposed regulatory or technocratic solutions will actually do more good than harm and relatively little evidence of the size of the benefit they might bring that isn't contingent on trusting said imperfect prediction models. And a fair bit of inertia about doing the regulating because the companies that proposed to be regulated have a pretty good idea about what they stand to lose.

- a lot of arguments over whether it's all a big con, driven far more by politics than understanding of the subject matter