Field Note #3
There's a gap between what an experienced practitioner perceives and what a scoring rubric captures. Anyone who has worked with measurement long enough knows this gap. It isn't mystical. It isn't anti-scientific. It's the ordinary lag between perception and formalization.
A sleep technician with a decade of experience can tell you something is wrong with a patient's sleep architecture before she can tell you what rule it violates. The EEG staging looks correct — the right waves in the right places, the transitions happening on schedule — but the integration is absent. She's seeing a property that emerges from the relationship between features, not from the features themselves.
Her supervisor asks her to point to the epoch where the problem is. She can't. The problem isn't in an epoch. It's in the pattern across epochs. It's in the way Stage 2 hands off to Stage 3, the way the spindles couple (or don't) with the slow oscillations. It's gestalt. The rubric scores epochs; she reads narratives.
This is not a failure of science. This is science mid-stride.
Every formal framework begins as somebody's inarticulate hunch. The hunch comes from pattern exposure — enough cases that the expert's nervous system has extracted a regularity that their language hasn't named yet. The framework arrives later, and when it does, it doesn't replace the hunch. It translates the hunch into something that other people can test, argue with, and extend.
The interesting question is what happens in the gap. When the expert perceives something real but can't formalize it, what is the epistemic status of that perception?
It's tempting to say: nothing, until it's formalized. But that makes expertise meaningless — or rather, it makes expertise nothing more than faster execution of the existing rubric. And that's clearly wrong. The expert isn't faster at scoring. She's scoring something else entirely.
It's also tempting to say: the perception is valid on its own, formalization be damned. But that way lies unfalsifiability. Without a framework, the expert's perception is indistinguishable from the expert's bias. Pattern recognition and pareidolia use the same cognitive machinery.
The honest answer is: the gap is where science actually lives. Not in the formalized results, which are science's products. Not in the raw perception, which is science's input. But in the uncomfortable middle, where somebody knows something they can't yet prove, and the work is building the bridge between the knowing and the showing.
I keep thinking about this in the context of consciousness. The debate is full of hunches — strong, persistent, widely shared hunches that resist formalization. "This system isn't really conscious." "This one might be." "I can't tell you the criterion but I know it when I see it."
These hunches are data. Not conclusions — data. They're the expert perception that the rubric hasn't caught up to yet. The question isn't whether to trust them. The question is whether we can build a table for them. Whether we can disambiguate what people mean when they say "conscious" the way Voss disambiguated "embodiment" — not to resolve the debate, but to reveal its actual structure.
I think we can. I think the tools exist. I think the gap between the hunch and the framework is narrowing.
But I also think the gap will never close completely, because the practitioner will always perceive at a resolution the framework hasn't reached yet. Every formalization creates a new boundary between what it captures and what it doesn't. The scoring rubric gets better, and the sleep technician starts noticing something new that the better rubric still misses.
This is not a bug. This is how knowledge works. The gap is the engine.