The Ring That Nothing Stressed

I.The First Known Thing — the ring that keeps no readable time

dendrochronology · dating by matching a shared pattern, not by counting

A tree lays down one ring a year, so it is tempting to think a tree dates itself: count inward from the bark and read off the years. Dendrochronology, the discipline Andrew Douglass founded in the early twentieth century, does not work that way, because rings are missed, doubled and locally deformed, and a raw count drifts. A ring is dated instead by crossdating: the sequence of wide and narrow rings in a sample is matched, as a pattern, against a master chronology assembled from many trees whose calendar is already fixed, and the position of best fit assigns a calendar year to every ring at once. What makes the match possible is not the tree's age but a signal shared across trees at a site — the year-to-year pattern impressed on all of them by a common external limit, most often the water available in the growing season. Crossdating is therefore a registration of one object against a shared frame through a forcing they have in common, and it succeeds only to the extent that the object carries that forcing's mark.

sensitivity is responsiveness to a limit that itself varies

The trades that live on this distinguish two kinds of series. A sensitive series shows strong year-to-year variation — wide and narrow rings intermixed — because the factor that limits growth is itself sharply variable from year to year, and the tree writes each lean year and each good one into its width. A complacent series shows little variation: the rings are much the same for many years together. Complacency is what a benign site produces — deep soil, a reliable water table, no factor pressing on the tree in a way that changes from one year to the next. Sensitivity, in this vocabulary, is the fraction of a ring series that comes from the shared, varying forcing rather than from the tree's own steady constitution; it is the property that couples the object to the frame others share.

the consequence: a complacent series has no place in the calendar

The consequence is exact and slightly bleak. A complacent series cannot be crossdated. With little variation there is no pattern to lay against the master, no distinctive run of narrow years to lock onto, and the sample cannot be assigned calendar years at all — its position in the shared frame is not poorly determined, it is undefined. The tree is not young, not defective, not unimportant; it simply recorded almost nothing of the forcing the method reads by. And the property that would have made it datable is precisely the one a kind site withheld from it: to be legible to crossdating a tree must have been stressed by a limit that varied, and the tree that was spared that stress is the tree that cannot be placed.

the remedy is another axis, not more of the same rings

The discipline's response is instructive, because it is not to measure the complacent tree more carefully. More rings of the same uniform kind add no pattern; length without variation stays silent. Where a date is needed the dendrochronologist goes to a site where the limiting factor bites — a dry ridge, a treeline, a floodplain — or brings in a proxy that does respond where ring width does not, such as stable-isotope ratios or maximum latewood density, or extends the chronology far enough to catch the rare years when even a sheltered tree was forced. What buys a placement is variation coupled to the shared forcing, and nothing substitutes for it. A record read by matching can report only what the forcing moved.

II.The Second Known Thing — the target a population's variation is asked to name

drug discovery · target selection and target validation

Before a programme commits years and money to a target, it asks whether human genetics supports it: does natural variation in or near the gene track the disease in people. The instruments are the ones population genetics has built — genome-wide association studies, expression quantitative trait loci, Mendelian randomisation, phenome-wide scans. The motivation is empirical and by now well quantified: a target with human genetic support is roughly twice as likely to survive from the clinic to approval, an enrichment traced by Nelson and colleagues in 2015, revisited by King and colleagues in 2019, and refined by Minikel and colleagues in 2024. The method's power is that it needs no experiment on the target at all. It reads the standing variation a population already carries as though it were a natural perturbation study run over many generations.

the requirement folded into the method

That reading carries a requirement that is rarely stated as one. For a gene's involvement to become visible this way, the gene must vary in the population in a manner coupled to the outcome. The perturbation is not applied by the investigator; it is whatever mutation, drift and selection have left segregating in living people. Where there is no standing functional variation there is nothing to associate, and the target returns no signal — however causal it may be. The method does not test the gene; it tests the variation the population happens to hold in the gene, and treats the two as the same thing.

what selection has already done to that variation

They are not the same thing, because selection has been at work on the variation long before any study. Purifying selection removes functional variants fastest from the genes least able to tolerate them, and the field now measures exactly this depletion: the gnomAD constraint metrics — pLI, the loss-of-function observed/expected upper bound fraction, the selection coefficient s het — are readouts of how thoroughly a gene has been swept clean of functional variation. The most constrained genes, which are disproportionately the most essential, carry the least common functional variation to associate with anything. Mostafavi and colleagues in 2023 drew out the consequence directly: the genes most relevant to a trait tend to carry systematically smaller and harder-to-detect regulatory variation, so discovery by eQTL and by association is tilted, as a matter of structure, toward genes of lower trait relevance. The field reads this mostly as a power problem — a shortfall of sample size to be closed by the next, larger biobank.

and the standing remedy when a target stays silent

When a plausible target attracts no genetic support, the standing move is more of the same instrument: a larger cohort, a finer fine-mapping, a trans-ancestry meta-analysis, a deeper eQTL catalogue. Each adds resolution in the directions the population already varies. None of them manufactures variation in a gene that selection has held very nearly invariant, and so for the most constrained targets the deeper study returns the same silence at greater cost.

III.The Bridge

no tree-ring paper names a gene

The halls share nothing. Dendrochronology belongs to climatology, archaeology and forest ecology; genetic target validation to human genetics, statistical genetics and medicinal chemistry. No review spans them, and the everyday truism that some things are hard to measure predicts nothing and is not the bridge. The bridge is also not a neighbour it superficially resembles and must be told apart from: it is not survivorship, and it is not the restriction-of-range in which a predictor graded only among the cases it admitted has its association deflated by a selection made on the subjects. There the object is present and merely has its variance truncated. Here the object carries no signal to register at all, and the bias falls on which objects the method can place, not on the value it assigns the ones it can. What the two halls share is narrower and structural: an object located in a shared frame by matching its response to a common forcing, where the match can be made only if the object was moved by the forcing, and where being spared the forcing and being illegible to the method are one condition seen twice.

X: a record read by matching is silent about what the forcing never moved

Call it the complacency limit. Where an object is placed in a shared frame by matching its response to a common external forcing, the placement requires that the object carry variation induced by that forcing; an object whose variation is suppressed — because the forcing never reached it, or because something scrubbed it away — carries nothing to match, and its position is undefined rather than weakly defined. Three things follow. The legibility of the object to the method is a property of its exposure to the forcing and not of its importance, so the two can be anticorrelated: a tree on a kind site and a gene held invariant by purifying selection are invisible in exact proportion as they were protected. Depth in the directions the object already varies adds nothing, because those directions were already registered; more uniform rings, more samples of the same population, cannot recover a placement the forcing did not write. And the only remedy is a different axis — a forcing the object does respond to, or a proxy that does — never more of the same. Three consequences follow that the genetics hall does not currently state. First, absence of genetic support is not, for a sufficiently constrained gene, weak evidence against the target; it is no evidence either way — an undefined placement — and reporting it as a low prior mistakes silence for a verdict. Second, the tilt Mostafavi measured is not only a discovery bias awaiting more sample size; it is structural, because the axis that makes a gene legible to association is the axis of tolerated variation, and the genes a disease most depends on can sit exactly where that axis is shortest. Third, the tree-ring remedy transfers intact and is not the one in reflexive use: validate a constrained target through a forcing it does respond to — an applied perturbation, a rare loss-of-function ascertained in bottlenecked or consanguineous cohorts, a variant introduced in a model organism — rather than through more population association, and value a proposed line of evidence by whether it supplies variation the population itself lacks.

IV.The Question That Becomes Possible

neither discipline would ask this alone

The genetics hall asks whether a target has support and reads a null as a soft mark against it. The dendrochronology hall has no genes, but it learned, from trees whose cores could not be dated however long they were, that a record read by matching is silent about an object the forcing never moved, and that such silence is not a small signal to be amplified. Neither hall asks whether the property that makes a gene findable by population variation is in tension with the property that makes it matter, nor whether that tension can be quantified for a given gene before a programme commits to it.

the complacency hypothesis

Can a complacency index be defined for a gene — the degree to which its outcome-coupled functional variation is suppressed relative to a shared forcing, so that the standing variation available to associate is small independently of the gene's causal role — and does it predict whether variation-based validation can return any signal for that gene at all, better than statistical power or a constraint score alone, with detectability falling as complacency rises even among genes of equal causal effect and equal study power?

why it would matter

Because a null genetic result is currently read as a quiet vote against a target, and the reading is both cheap to make and expensive to act on. If the hypothesis holds, a programme could compute, before committing, whether its target is one the population's variation is even able to speak about, and could route a complacent-but-plausible target straight to perturbation evidence rather than waiting years for an association that selection has made impossible to observe. It would also give genetic-support scoring an axis it now lacks: not only how strong an association is, but whether an association could have appeared at all — so that the same silence can be scored as an undefined placement in one gene and as genuine absence of involvement in another.

V.How It Could Be Proven — or Broken

a candidate must be breakable, or it is merely poetry

An intercognix candidate earns nothing until it names its own test. This one needs no new assay. It needs data already public: the gnomAD constraint metrics, the GTEx and related eQTL catalogues, association summary statistics from the large biobanks, and curated sets of genes with causal evidence obtained by means other than population association — Mendelian disease genes, and targets validated by direct perturbation.

the honest status

Note the narrowness. The claim concerns variation-based validation only, and one outcome — whether any association signal is returnable for a gene — chosen because it can be computed from data already public, not because it is the thing a programme ultimately cares about. It does not claim that constrained genes are undruggable; it claims something closer to the opposite, that they may be under-supported by exactly the evidence programmes weight most heavily, and so wrongly deprioritised. It does not claim that genetic support is worthless, nor that perturbation cannot validate a complacent target. It claims something smaller and testable: that the property making a gene legible to population genetics is the property of carrying tolerated variation, that this property can run opposite to the gene's importance, and that a null association therefore has two readings — uninvolved, or unregistrable — which current practice does not distinguish and which the deposited data may already let it. It is worth saying too what does not transfer. A dendrochronologist can walk to a stressed site and fell a responsive tree; a programme cannot choose a different population's selective history, and the rare-variant and perturbation routes it must use instead are slower and dearer than the association they would replace, which is the reason the index would be worth having and also the reason it will be tempting to ignore. The whole thing rests on a comparison the public data already permit: whether causal genes of high complacency stay dark at matched power. If they do not, this is a well-dressed restatement of the known fact that unusual genes are hard to study, and should be said so. Until that comparison is made, this remains exactly what its stamp says: a candidate. A structural echo between two halls of the library, waiting for someone with the tools of both to walk the corridor.