It’s got me thinking about Carl Jung and his work on symbols. Jung argued that we don’t just consume information linearly; we navigate through the shadow, filling in gaps with intuition and archetypal resonance. When you strip a language down to its most cryptic, logical scaffolding falls away. That’s when you see if a system is actually reasoning, or just mimicking the shape of thought.

To me, this whole experiment is giving major High Priestess energy. That card doesn’t care about surface-level logic or clean data streams. It demands you sit in the dark until meaning surfaces from the deep. Testing a machine this way feels a lot like drawing The Moon—you’re asking if it can navigate illusion and hidden currents, or if it’s just tracing reflections on the water. I’ve been running my own little late-night experiments, feeding fragmented, deliberately weird prompts into different models. Sometimes they stumble beautifully into unexpected connections. Other times, they just echo the room back to me.

I’m starting to think understanding isn’t really about perfect syntax. Maybe it’s about resonance. If a machine learns to think in a deliberately mysterious language, is it actually grasping the concept, or just dreaming in code? I’d love to hear how you guys navigate this stuff. Do you trust an oracle that speaks in riddles, or do you demand plain text?