Flaw Reading Order: Fatal First

Severity weighting

Order beats completeness whenever a seller sends a gallery rather than a single frame. Twelve photographs can hold one fatal defect and eleven that flatter the item, and a reader who walks the set front to back meets the flattering ones first and forms a view before reaching the frame that matters. This desk therefore sorts every observation into three severity bands before judging quality at all: band one ends the purchase, band two is negotiable, band three is recorded and then absorbed. The bands are a reading order rather than a rating scale, and nothing inside them says an item is good or bad in general.

Weight is assigned by reversibility rather than by how dramatic a fault looks at full magnification. A bonded joint that has already separated, a stamped code cut into a surface that cannot be recut, a bezel that stops short of seating: none of those improve with time, and none can be repaired for less than the difference in price. A loose thread that trims cleanly or a fingerprint on a lacquered panel carries less weight however large it appears on screen. The test reduces to one question, asked of each observation in turn: after six months of ordinary use, will this still be there?

The weighting table further down is a desk construction rather than a platform rule, and it exists because one sequence error kept appearing in reader reports. A hairline surface mark was read first, a displaced midsole was read last, and the decision to accept had already formed by the time the structural frame was opened. Sorting by band first costs about ninety seconds per listing and reverses the outcome on roughly one listing in five in the recorded sample. The saving is not the ninety seconds. It is the refusal that never gets issued on an item whose fatal fault was sitting in frame six.

Two habits keep the sort honest. Every supplied photograph gets read once before any band is assigned, because a surface mark in one frame is often the visible edge of a structural fault in another. And bands attach to observations rather than to sellers, so a listing carrying one fatal fault is not dismissed wholesale; the remaining frames still decide whether that fault is isolated or part of a run. A seller with one bad frame out of twelve is a different proposition from a seller whose four frames all show the same lifted edge, and only a full read separates those two cases.

Fatal defect ranking

Fatal observations share a single property: nobody downstream can correct them, and the item cannot be used as intended while they persist. In footwear the recorded set covers a midsole silhouette that differs between left and right by more than one stitching-line width, an outsole bond that lifts under thumb pressure at the toe or heel edge, and a heel counter that folds under light hand pressure. Each of those three can be tested by hand in under ten seconds once the frame has been taken, which is why they are the first three the desk looks for.

Watch evidence forms a second cluster. A dial index sitting off the printed minute track, a bezel that stops short of the pip at twelve, and a caseback engraving whose depth varies across a single word are all errors of form, and form errors do not settle with wear. A bracelet whose end link leaves a gap wider than the neighbouring link pitch puts the piece in the same band before accuracy is ever discussed. Movement questions, which need a timegrapher and a week of wear, are deliberately left out of this band because they cannot be settled from photographs at all.

Ranking inside the band matters less than placing an observation correctly, but the desk orders fatal findings as bonding first, symmetry second, identity third. Bonding failures shorten an item's working life fastest. Symmetry errors are the ones visible at arm's length and the most common stated reason for refusing a parcel. Identity failures, such as a stamp that disagrees with the printed box code, affect resale value more than use, and they rank third for that reason alone. A reader selling nothing has little reason to care about the third group at all.

Counting before ranking keeps the process repeatable. State how many observations landed in the band, name the part each one concerns, and stop there. A set yielding two fatal findings is a refusal whether or not a partial credit is offered, because both faults still arrive in the parcel. Where an observation could sit in either of two bands, the desk files it higher and writes the reason for the doubt beside it. Recording the doubt matters more than resolving it, since the next reader will otherwise re-open a question that was already answered once.

Cosmetic defect ranking

Cosmetic observations occupy the middle ground, and the deciding question is whether a mark stays visible in ordinary use. A scuff along an outsole edge vanishes the first time the item meets pavement; the same scuff on a front-facing toebox panel does not. Position sorts this band far more than size does, so a five millimetre mark on a named front panel outranks a twenty millimetre mark under the arch. The same rule applies to bags and outerwear, where a mark on a corner outranks a longer one along an inner seam.

Negotiable marks carry a second property worth using: they can be described precisely enough to price. A reader who records the panel, the distance from a named seam and the length of the mark has produced something a seller can answer with a credit or a replacement. A reader who writes that the upper is scratched has produced nothing, and the exchange usually stalls on which photograph is being discussed. Three fields, written in the order the eye moves across the frame, turn a complaint into a claim with a number attached to it.

Stitching belongs in this band with one exception. A wandering row, a line that crosses a panel boundary, or curve spacing that tightens along its length are cosmetic because the seam still holds. Thread ends beyond roughly three millimetres, a stitch count that falls by more than two inside a ten millimetre window, or a seam that has pulled the surrounding material into a pucker all move the finding up a band. The distinction is whether the seam has failed or merely looks untidy, and a tape laid beside the row answers it in one frame.

Colour and finish drift also sits here. Mixed lighting widens an apparent difference, so the comparison has to run between two frames shot in one session rather than between a seller gallery and a photograph found elsewhere. Where a panel reads as one colour across every frame of a set, the finding is a finish difference; where it shifts between frames, the lighting is doing the work and no band should be assigned yet. Two frames of the same panel under the same light is the entire test, and it costs the seller nothing to supply.

See current listings

Tolerable defect ranking

Tolerable findings are the ones that survive a full reading and change nothing. Adhesive residue inside a lining, a protective film still in place, a shallow wrinkle in an inner sole that flattens under a thumb: each is written down and then set aside. The band records that the observation was seen, so a later dispute cannot rest on the claim that nobody noticed it. Recording a tolerable finding costs one line and removes an argument, which is a better return than most of the work described in this dispatch.

Packaging lands in this band for most categories. A dented outer box, a label applied at a slight angle, a dust bag with a frayed drawstring: none of those touches the item, and each has been used to open a claim the photographs could not support. Recording them as tolerable in advance costs nothing and keeps the fatal band free of noise. Two exceptions are worth naming here: a box whose printed code differs from the stamp on the item, and packaging that has been opened and resealed before dispatch.

Two families of finding look tolerable and are not. Anything that will change a measurement, such as a strap that has already stretched or a cuff that has lost elasticity, belongs higher even when it currently looks acceptable. Anything that removes a documented accessory, such as a spare lace set or a sizing card shown in the listing, belongs in the middle band, because the listing itself created the expectation. Both families share a trait: the item as supplied no longer matches the item as described, and that gap is what the bands exist to measure.

A separate tolerable band pays for itself in arithmetic. A reader who has already filed four findings there can read a fifth without re-arguing the first four, and the message to a seller narrows to the two items that matter. Long messages mixing all three bands let a seller answer the cheapest point and leave the expensive one untouched, which is exactly the outcome the ordering exists to prevent. Short messages that name one band and two parts get answered more often, in the desk's own reading of the exchanges it has seen.

Dissenting views

Readers do not weight the bands identically, and the disagreement is worth recording rather than smoothing away. Resale-oriented readers place more weight on identity marks and packaging than a wearer does, because a box code that disagrees with the stamp costs money at resale and nothing on the wrist. Wearers place more weight on bonding, since a separated sole ends the item's useful life whatever the box looks like. Neither weighting is wrong; they answer different questions about the same parcel. The desk records which weighting a verdict used, so the two positions can be compared later instead of argued in the moment.

A second disagreement concerns missing evidence. Some readers treat an absent angle as a fatal finding, reasoning that a seller who will not photograph a heel seam has something to conceal. Others treat it as a request rather than a result. The desk keeps both positions visible: an absent angle is logged as an open question, and no refusal is issued until the frame arrives and is read. Refusing on absence alone burns the strongest position available, which is a refusal issued after the missing frame has been supplied and read.

Third comes the question of partial credit. Accepting money for a fatal fault turns a refusal into a purchase at a lower figure, which is a fair trade for a reader who intended to use the item hard. It is a poor trade for a gift purchase, where the fault stays visible to somebody else for as long as the item is kept. The band does not move; only the decision taken afterwards does. Writing which of the two situations applies, before the offer arrives, is what keeps the decision from being made by the size of the number.

Where dissent cannot be settled, the disagreement is written into the record with the reasoning on each side. Disagreements that recur across a month of reading usually mean the band definitions are too coarse, and the repair is a new row in the table rather than a new rule in the prose. A reader who reaches a different order should be able to see which weighting produced it. That is the whole function of publishing the table: an ordering that cannot be inspected cannot be argued with, and an ordering nobody can argue with stops improving.