Content Quality: Well-organized News piece (1,180 words, within the 400-1200 range) that separates the two source posts cleanly: 'Why the Parser Is Being Replaced' and 'A Parsing Bug Fixed With Memoization' draw from the PEG-parser deep dive, while 'The Rest of the v2.0 Preview' draws from the highlights post. The 'What We Don't Know' section appropriately flags that the release date is only 'this fall' and that several pieces (including the runtime grammar-extension API) are previews subject to change. Technical explanations of PEG vs. YACC/Bison, packrat parsing/memoization, and the parser-vs-binder distinction are accurate paraphrases of the source material.
Source Verification: All 3 sources read from local snapshots (not re-fetched): source-0.html.gz (duckdb.org/2026/08/17/duckdb-20-highlights.html, sha256 verified against manifest), source-1.html.gz (duckdb.org/2026/08/20/duckdb-20-peg-parser.html, sha256 verified), source-2.html.gz (github.com/duckdb/duckdb/releases, sha256 verified). All three returned HTTP 200, none used archive_fallback, and manifest.json shows suspicious_patterns: null for every entry (confirmed by inspecting the manifest directly, not just the summary). source-0 confirms the 'Cyanoptera' codename, the fall-2026 timeline, the client/server ('Quack'/CONNECT) mode, triggers, the storage-format bump to v2.0.0, the ICU removal, the C API revamp, and the '10,000 commits since v1.5' figure -- all quotes checked verbatim and matched except one (see Quote Fidelity finding). source-1 confirms the PEG-vs-YACC/Bison rationale, the 2018 first-commit lineage, the Python 3.9 PEG comparison, the parser/transformer/binder pipeline description, both example-query error messages (Parser Error and Catalog Error, checked character-for-character), the parenthesis-benchmark timings (5.303s / 10.640s / 0.001s), and the pipe-syntax extension demo -- all quotes verbatim except the Python 3.9 one flagged above. source-2 (GitHub releases) confirms v1.5.5 is the 'Latest' tagged release, dated 2026-07-22T10:52:48Z (article rounds to 'July 22, 2026'), and that no v2.0 tag exists in the release list, supporting the article's 'no v2.0 tag yet ... consistent with the fall timeline' framing.
Factual Accuracy: All specific claims, numbers, error messages, and code examples trace to the two DuckDB blog posts. One quote (the Python 3.9 PEG comparison) is not verbatim -- it drops the word 'the' and truncates the trailing clause without an ellipsis; substance is preserved but the exact wording inside quote marks does not match the source. This is the sole factual-accuracy issue found and is addressed with a corrections record. No fabricated specifics, no orphan source URLs (all three cited sources also appear in article.sources), no hallucinated quotes beyond the one flagged.
Overall Assessment: Substantively strong, well-sourced technical preview article with accurate technical explanations and correct GitHub-releases cross-check. The single quote-fidelity issue is minor, does not affect the headline/summary/lead, and is honestly coverable with one corrections entry. duckdb.org is added to the source allowlist as an official first-party vendor blog per existing precedent. Verdict: APPROVE_WITH_CORRECTIONS.