I did not start Oneirox because I hated dream websites. I started it because every tool I tried answered a question I had not asked.
I would wake with chest pressure, a chase that dissolved in thirty seconds, and a body that still felt “on.” The site would ask for a keyword. I typed snake or falling. It returned five meanings. Sometimes fifty. None of them asked what my amygdala was rehearsing, whether the cold on my skin was thermoregulatory residue after sleep, or whether the image was a threat-simulation output rather than a metaphor waiting for translation.
If a tool starts with “snake means…”, it has already abandoned the brain.
The premise I will not ship
Dream dictionaries predate REM research by roughly a century. Digitizing them with a modern interface — or wrapping them in a language model — does not make them neuroscience. It makes folklore searchable.
I spent years reading the opposite stack: sleep neurophysiology, somatic markers, threat simulation, emotional consolidation. The practical conclusion was simple enough to write on a sticky note:
Images are outputs. Mechanisms are the question. Chase, fall, invasion — often limbic rehearsal without motor output (Revonsuo). Recurring people and unfinished conflict — often consolidation still in progress (Walker, Cartwright). Pressure, cold, jaw tension on waking — often more honest than the plot (Damasio).
That is why Oneirox pages commit to SIGNAL · BODY · MORNING — one committed reading — instead of a menu of interpretations. When every meaning is offered, none is yours.
What this means for the Lab
It also means I will keep building instruments that do not need an API to be useful: the Sensory Dream Mapper for interoception, the Dream Phase Calculator for chronobiology context, the Sleep Cycle Calculator for ultradian planning, and the somatic and dream-mechanism libraries for people who want mechanism before they open Lab Search.
The Lab is the everyday layer — Search, instruments, mechanism pages. Same refusal: no symbol tables dressed as insight.