daily mirror — 30/07/2026

CHANGELOG.md

@@ -2,6 +2,12 @@
 
 All notable changes to this dataset. The data itself is dated and append-only; this log records **structural** changes (new files, schema, coverage).
 
+## 2026-07-30, errata mechanism
+
+### Added
+- **`ERRATA.md`** (trilingual). Records known defects in already-published dated files. Policy stated explicitly: **files for closed dates are never rewritten**; the correction is published in the errata instead, so the record stays exactly as distributed and the defect stays discoverable. The file for the *current* date may still be regenerated during the day, since the mirror runs more than once daily.
+- **First entry, ERR-2026-001:** `data/divergence-2026-07-29.csv`, row `Renan Santos`, `polymarket_pct` published as `8.6` where the site and the daily synthesis published `8.70%` (`divergence_pp` `0.8` instead of `0.90`). One row of seven on that date; 0.10pp. Cause and prevention documented in `ERRATA.md`.
+
 ## 2026-06-30, June poll wave + coverage refresh
 
 ### Added / Updated

ERRATA.md

@@ -0,0 +1,79 @@
+# Errata
+
+**This dataset is dated and append-only. Files for past dates are never rewritten, including when they contain a known error.** Corrections are published here instead, so the record stays exactly as it was distributed and the defect stays discoverable.
+
+Scope note on "append-only": the file for the **current** date may be regenerated during the day, because the mirror runs more than once daily. **Files for dates already closed are never modified.**
+
+---
+
+## EN
+
+### ERR-2026-001 · `data/divergence-2026-07-29.csv` · Renan Santos
+
+| | Value |
+|---|---|
+| **Affected row** | `2026-07-29,Renan Santos,8.6,7.8,0.8` |
+| **Published** | `polymarket_pct = 8.6` · `divergence_pp = 0.8` |
+| **Correct** | `polymarket_pct = 8.70` · `divergence_pp = 0.90` |
+| **Magnitude** | 0.10pp |
+| **Rows affected** | 1 of 7 that date. All other rows and all other dates are unaffected. |
+| **Detected** | 2026-07-30 |
+| **Status** | **Not corrected in place, by design.** The historical file keeps the value as distributed. |
+
+**Cause.** The source file `polls-data.json` stores the same market price in three fields: a display string (`polymarket`), and two numeric fields (`odds`, `value`). A mid-day rebaseline on 2026-07-29 updated the display string to `8,70%` and left the numeric fields at `8.6`. The exporter reads the numeric field first, so the numeric one is what reached this dataset.
+
+**What the site and the published daily synthesis said that day:** `8.70%`, with the market-versus-poll distance stated as `0.90pp`. The narrative and the dataset therefore disagree by 0.10pp for this single row.
+
+**Prevention.** Since 2026-07-30 a check (`checkPrecoTriplo`) fails the build if the three fields disagree by more than 0.005. It runs in the schema validator and in the pre-commit hook.
+
+---
+
+## PT-BR
+
+### ERR-2026-001 · `data/divergence-2026-07-29.csv` · Renan Santos
+
+Este conjunto de dados é **datado e aditivo**. Arquivos de datas já encerradas **nunca são reescritos**, mesmo quando contêm erro conhecido. A correção é publicada aqui, para que o registro permaneça exatamente como foi distribuído e o defeito continue descobrível.
+
+| | Valor |
+|---|---|
+| **Linha afetada** | `2026-07-29,Renan Santos,8.6,7.8,0.8` |
+| **Publicado** | `polymarket_pct = 8.6` · `divergence_pp = 0.8` |
+| **Correto** | `polymarket_pct = 8.70` · `divergence_pp = 0.90` |
+| **Magnitude** | 0,10pp |
+| **Alcance** | 1 linha de 7 naquela data. Nenhuma outra linha e nenhuma outra data são afetadas. |
+| **Detectado em** | 30/07/2026 |
+| **Situação** | **Não corrigido no lugar, por decisão.** O arquivo histórico mantém o valor como distribuído. |
+
+**Causa.** O arquivo de origem `polls-data.json` guarda o mesmo preço de mercado em três campos: uma string de exibição (`polymarket`) e dois campos numéricos (`odds`, `value`). Um rebaseline no meio do dia 29/07/2026 atualizou a string para `8,70%` e deixou os numéricos em `8.6`. O exportador lê primeiro o campo numérico, e foi ele que chegou a este conjunto de dados.
+
+**O que o site e a síntese diária publicaram naquele dia:** `8,70%`, com a distância entre mercado e pesquisa declarada em `0,90pp`. A narrativa e o conjunto de dados divergem, portanto, em 0,10pp nesta única linha.
+
+**Prevenção.** Desde 30/07/2026 uma checagem (`checkPrecoTriplo`) reprova a publicação se os três campos divergirem mais de 0,005. Roda no validador de esquema e no gancho de pre-commit.
+
+---
+
+## ES
+
+### ERR-2026-001 · `data/divergence-2026-07-29.csv` · Renan Santos
+
+Este conjunto de datos es **fechado y aditivo**. Los archivos de fechas ya cerradas **nunca se reescriben**, ni siquiera cuando contienen un error conocido. La corrección se publica aquí, para que el registro permanezca exactamente como fue distribuido y el defecto siga siendo detectable.
+
+| | Valor |
+|---|---|
+| **Fila afectada** | `2026-07-29,Renan Santos,8.6,7.8,0.8` |
+| **Publicado** | `polymarket_pct = 8.6` · `divergence_pp = 0.8` |
+| **Correcto** | `polymarket_pct = 8.70` · `divergence_pp = 0.90` |
+| **Magnitud** | 0,10pp |
+| **Alcance** | 1 fila de 7 en esa fecha. Ninguna otra fila ni ninguna otra fecha están afectadas. |
+| **Detectado** | 30/07/2026 |
+| **Situación** | **No corregido en el lugar, por decisión.** El archivo histórico mantiene el valor tal como fue distribuido. |
+
+**Causa.** El archivo de origen `polls-data.json` guarda el mismo precio de mercado en tres campos: una cadena de exhibición (`polymarket`) y dos campos numéricos (`odds`, `value`). Un rebaseline a mitad del día 29/07/2026 actualizó la cadena a `8,70%` y dejó los numéricos en `8.6`. El exportador lee primero el campo numérico, y ese fue el que llegó a este conjunto de datos.
+
+**Lo que el sitio y la síntesis diaria publicaron ese día:** `8,70%`, con la distancia entre mercado y encuesta declarada en `0,90pp`. La narrativa y el conjunto de datos divergen, por lo tanto, en 0,10pp en esta única fila.
+
+**Prevención.** Desde el 30/07/2026 una verificación (`checkPrecoTriplo`) rechaza la publicación si los tres campos difieren en más de 0,005. Se ejecuta en el validador de esquema y en el gancho de pre-commit.
+
+---
+
+*No other errata recorded. · Nenhuma outra errata registrada. · Ninguna otra errata registrada.*

README.md

@@ -121,6 +121,7 @@ Full column-level definitions for every file are in **[`DATA_DICTIONARY.md`](DAT
 - **Editorial stance:** AFOS reports *divergence* between sources rather than a single blended average — the spread is treated as signal, not noise.
 - **Demographics:** *sample-design* demographics — the declared composition/weighting of each poll's sample (layer A) — are included (`polls/sample-demographics.csv`). *Vote-by-demographic crosstabs* (layer B — e.g. vote share by sex/age/income) are **not** part of Brazil's TSE open data; institutes publish those separately, so they are intentionally absent here rather than partially scraped.
 - **Scale caveat (market vs poll):** Polymarket prices *probability of winning*; polls report *vote share*. The two divergence files keep both raw values side by side and flag the naive gap accordingly — they are not a like-for-like error metric.
-- **Updates:** dated and append-only; each daily commit preserves the full history natively (see `CHANGELOG.md`).
+- **Updates:** dated and append-only; each daily commit preserves the full history natively (see `CHANGELOG.md`). The file for the *current* date may be regenerated during the day; files for dates already closed are never modified.
+- **Known errors:** see **`ERRATA.md`**. When a defect is found in a past date, the file is **not** rewritten — the correction is published in the errata instead, so the record stays as distributed and the defect stays discoverable. *Erros conhecidos em `ERRATA.md`. · Errores conocidos en `ERRATA.md`.*
 
 **Sources / Fontes / Fuentes:** Polymarket (live USD markets) · TSE-registered institutes · 400+ press outlets. Method & source code (Apache 2.0): [github.com/AFOS-Analytics](https://github.com/AFOS-Analytics).