I totally agree. Kekulé's discovery of the structure of benzene was in a dream and purely visible. When I still did chemistry (post-doc), much of my thinking was not verbal but "Intuition" (whatever that is). Explaining in words my discoveries was secondary.
The fault lines between unconscious understanding and reality come into focus when writing code. Intuition is fantastic for one big idea, or a few medium-sized ones, but struggles to remember lots of details.
There weren't tons of details to pin down. IIRC, the question was the structure of benzene, which was known to be C6H6, but nobody could figure out how to make a carbon backbone six atoms long that would take only six hydrogen atoms. Hexane, for example, is C6H14.
Kelulé's dream was of six snakes in a circle or hexagon, each eating the next snake's tail. He turned that into the (correct) idea of the benzene ring - six carbon atoms connected in a circle. The only remaining detail was that it had to have alternating single and double bonds (and even that was wrong - the double bonds get smeared out, so that it's more like one and a half bonds between each atom).
So I'm not sure what you mean by "all the details". There were very few details. "Connected in a loop" was the only thing that mattered; it explained everything.
