Socrates Bourbon

Under the hood

How Socrates builds a pour’s profile, and what the Confidence Score means

Before your guide says a word, the pour you name is researched, checked for era and identity, and graded for how settled its record is. Here is what that involves, in plain terms.

1. The retrieved record comes from the public record

Socrates does not invent a pour’s profile. For a named pour it retrieves what has actually been published: the producer’s own tasting notes for that exact bottling first, then the independent reviews that agree with each other and with the producer. A note only one source mentions does not make it into the profile.

For vintage and discontinued bottles — the dusties — the search is pointed at the archives the collecting community trusts: bottle-dating and distillery references such as WhiskeyID and Bourbon Veach for identity, and long-running review archives and forum records for the tasting notes. Retail and auction sites are excluded on purpose.

Each retrieved record is kept permanently, so a pour is researched once and every sipper after that reads the same retrieved record. Different spellings, acronyms and year formats resolve to one record, never several.

2. Era is arithmetic, not opinion

A bottling year minus an age statement is a distillation year, and the big houses’ eras are hard dates. A 1998 twelve-year Elijah Craig was distilled around 1986, at the old Bardstown distillery, before the 1996 fire. Socrates computes that before the search runs, so the retrieved record is built for the right distillate, and a modern shelf bottle is never read against a dusty profile, or the reverse.

For the documented dusty families — pre-fire Heaven Hill, Stitzel-Weller, National Distillers, dusty-era Wild Turkey — the eight tasting axes come from the founder’s own dataset of 400+ documented blind tastings, and the web fills in the prose.

3. A curated master list sits above the web

The most-poured bottlings, modern and dusty, carry a curated profile built from several independent research passes and reconciled axis by axis, with the founder’s review as the final word. When you name one of these, no search runs at all: the retrieved record is on file, and the Confidence Score starts at the top of the scale.

4. Identity is checked, and never assumed silently

If the retrieved record could be a different release than the one you named — a batch, a vintage, a limited edition — Socrates says which bottling it is reading as and gives you a one-tap “Not my pour” to correct it. Until you do, the retrieved record’s numbers are given as ranges rather than stated as fact.

5. The Confidence Score is evidence, not a guess

Profile Confidence is arithmetic over things that can be counted, and the card behind the score lists every one of them. It goes up with:

It goes down only for a real reason: the research itself was unsure, no independent sources could be found, or the retrieved record may be for a different release than the one you named.

High — a documented profile, usually with a settled public score. Every stage is confirmed against it.
Medium — documented, but a release detail is unsettled or the score is thin. Socrates gives proof and age as ranges.
Low — little published intel on this exact pour. Socrates reads it by its type and speaks in ranges; the reveal grades against typicity. The two of you work the mystery together.

The score describes the retrieved record, not you. Your own reads are scored separately, at the reveal, against that retrieved record.

6. When the retrieved record is wrong, it gets fixed

Every reveal, and every guide message, carries a way to report an error. Reports go to the founder. When one is right, the correction becomes part of that pour’s retrieved record permanently, and the sipper who reported it hears back.

What Socrates never does


Name a pour and see it work

Your first three sessions are free. No app store hassle ever.