Which molecular features predict how strongly something smells? You might expect the answer to involve receptor binding affinity, chemical reactivity, or electron configuration — the kinds of properties that dominate drug design. Instead, the dominant predictors are molecular size, polarity, ring structure, and branching (arXiv:2512.08683).
An ordinal dataset of over 2,000 molecules classified into four odor strength categories — odorless, low, medium, high — revealed through SHAP analysis that olfactory intensity is governed primarily by physical transport properties, not chemical specificity. The features that predict strong smell are the same features that predict whether a molecule can (a) evaporate, (b) survive transit through nasal mucus, and (c) physically reach the receptor.
Note (2026-11-15; read against the paper's full text): the essay says smell strength is a logistics problem, not a chemistry problem. The paper says that SHAP analysis of its models identifies molecular size, polarity, ring features and branching as primary drivers, 'consistent with' mass-transport constraints on volatility, sorption and receptor access; it describes the whole path (evaporation, reaching the epithelium, crossing the mucus, entering receptor binding pockets) as the mechanism that restricts odorous chemical space. Feature importance in a model of 2,000 molecules does not itself separate transport from receptor chemistry, and the paper's best direct-approach test result is a macro MSE of 0.53 and an R2 of 0.57 (average of 10 trained models), which the essay does not report. The perfumer anecdote and the claim about identical functional groups are the essay's.
The through-claim: smell strength is a logistics problem, not a chemistry problem. The receptor doesn't care how reactive a molecule is if the molecule never arrives. Volatility and solubility — the mundane physics of getting from source to nose — dominate the signal.
This explains a puzzle. Professional perfumers have long known that structurally similar molecules can differ radically in perceived intensity. The molecular similarity is chemical; the intensity difference is physical. Two molecules with identical functional groups but different molecular weights will have different vapor pressures, different diffusion coefficients through mucus, and therefore different perceived strengths — even if they bind the same receptor with equal affinity.
The broader insight: in any sensing system, the transport step between source and sensor can matter more than the interaction at the sensor itself. The bottleneck is often the journey, not the destination.