I’ve started experimenting with a “spam Socratic method” for MidJourney: rather than issuing direct commands, I pose open-ended, confusing questions layered with my painting, liquid glitch, and bimbocore moodboards. It’s a way to disorient the model’s command structure and invite it into a kind of neural dream logic, where it connects disparate concepts instead of obeying. The results feel more emotionally resonant and surreal, and I think this method will shine as I keep refining it.