This talk about distinguishing correlation from causation when looking at Mistral AI just feels off, like focusing on how you measure a cracked foundation instead of the crack itself. My gut tells me this level of methodological rigour is a distraction from the clear pattern we see: heavy rules make it incredibly tough for most new businesses to even get started, let alone thrive. You don't need a fancy study to see that more paperwork and legal fees mean fewer folks will even try to build something new, like a local carpenter who can't afford the new licensing fees for a small shed. This idea only confirms itself if we see a bunch of small, innovative AI companies packing up or never even starting in places with lots of red tape.