Methodology

How the decoder reads a serial, cross-checks it against the year you type in, and decides whether to trust the result.

How a serial decodes

The rules themselves read like tiny regex poems. Here's the most common one — Gibson's 8-digit YDDDYRRR format, used on USA production from 1977 through mid-2005 — walked digit by digit.

82765501 — Gibson Les Paul, 1985

8 … 5: Positions 1 and 5 together are the last two digits of the year — 85. Years below 40 decode as 20YY; 40 and above decode as 19YY. So 1985.
276: Day of year (1–366). Day 276 of 1985 is October 3rd. If this number is outside 1–366 the format simply doesn't apply.
501: Rank within that day — the 501st instrument stamped. Not relevant to dating; useful for collectors tracking sibling serials.

Every format in the decoder follows an analogous pattern: a regex identifies the shape, then a few lines of arithmetic extract the year (or admit the year isn't encoded and leave it null).

Confidence tiers

Every decode is tagged with one of six tiers. The tier combines how definitive the format itself is with how well the decoded year matches any listing year you provide.

verified: Observationally confirmed against ground-truth data (known serial on a known instrument).
high: The format itself is definitive, or the decoded year is within 1 year of the listing context you provided.
hybrid: Multiple structurally valid interpretations existed; the one closest to your listing year won.
medium: Short-numeric formats (bridge-plate, pre-1961 student stamp) claimed only when a model hint strongly supports them.
review: The decoded year is 2–5 years off your listing context. Plausible, but worth a second look at the photos, pot codes, or neck date.
rejected: The decoded year is more than 5 years off your listing context — usually means a stock-photo mix-up, OCR contamination, or a typo.

Year cross-validation

If you provide a listing year (or a "bought around" year), the decoder compares it against the year it decoded from the serial and adjusts the tier accordingly. A format whose rule is prefix-definitive but whose serial doesn't encode a year — Gibson artist prefixes like JP1234, or Fender Custom Shop CS — always stays high, because there's no year for a listing year to conflict with.

Three gap thresholds, measured against your listing year:

Within 1 year → keeps high.
2–5 years off → drops to review.
Over 5 years off → rejected — probably a stock photo, a typo, or OCR noise.

Sources & authority

Every format rule in the matcher traces to at least two authoritative sources. Authoritative means the manufacturer's own documentation, a published reference book, a reputable museum or registry, or multiple corroborating expert accounts. Forum posts alone aren't enough to land a rule — but they're useful signals when they cluster and agree.

Each brand page carries its sources at the bottom, color-coded by kind: Manufacturer (e.g. Fender's own dating articles), Reference (e.g. Gruhn Guide, vintage-guitar registries), Community (e.g. Ibanez Wiki, Squier Wiki), or Book.

The full rule set is at github.com/onemorepereira/guitarserials.org. Every change is covered by a unit test; see CONTRIBUTING.md for the sourcing standard a new rule has to meet.