Content Quality: Clean, well-structured News piece with Overview / What We Know / What We Don't Know / Context sections. Technical JEP summaries are accurate and appropriately terse for the News category (597 words, within the 400-1200 range).
Source Verification: All 3 sources were fetched successfully (HTTP 200) and read from the gzipped snapshots on disk; sha256 of each decompressed file matches manifest.json, confirming snapshot integrity. (1) source-0.html.gz (InfoWorld, 'JDK 27: The new features of Java 27', Jul 29 2026) — confirms nine JEPs, JEP 523/527/531/532/533/534/536/537/538 descriptions, non-LTS six-month support, and (pre-delay) 'final release candidate stage for JDK 27 follows on August 20, with general availability scheduled for September 15' — this pre-delay GA figure (Sept 15) is NOT used anywhere in the article; the article only cites this source for feature descriptions and the pre-delay Aug 20 final-RC date, both of which it verbatim supports. (2) source-1.html.gz (InfoQ, 'Java News Roundup ... Aug 3, 2026', published Aug 10 2026) — confirms the RC1 delay from Aug 6 to Aug 20, the Aug 18 CPU alignment, that Build 34 was originally meant to be the RC1 build, and both Reinhold quotes verbatim ('This reduces the time available for feedback on RC builds by two weeks, from five weeks and four days to three weeks and four days...' and the following sentences through 'This suggests that the risk is tolerable.') — these long quotes are word-for-word matches. One shorter phrase in the same paragraph, 'due to an increase in frequency of Critical Patch Updates (CPUs),' is presented in the article as part of a direct quotation but is actually InfoQ's own paraphrase in the source (verified via raw HTML: 'due to <a>increase in frequency</a> of Critical Patch Updates (CPUs)', no quote marks around it in the source, and no 'an' before 'increase') — flagged as a finding above and covered by a corrections record. (3) source-2.html.gz (InfoQ, 'OpenJDK News Roundup ... May 18, 2026', published May 25 2026) — contains the official JDK 27 release schedule table with 'General Availability: September 14, 2026' stated explicitly and literally, directly supporting the article's claim ('JDK 27 remains on track for a September 14 general availability date, according to InfoQ') with no arithmetic or inference required. This also confirms JEP 531/532/533/534/536/537/538 target status and preview/incubation round numbers.
Factual Accuracy: Investigated the apparent Sept 14 vs Sept 15 discrepancy between InfoWorld (Sept 15, in a pre-delay July 29 article) and InfoQ (Sept 14, in the official schedule table from the May 18 roundup) flagged for extra scrutiny. Resolution: the article never uses InfoWorld's Sept 15 figure — it cites InfoWorld only for feature descriptions and the (accurate, pre-delay) Aug 20 final-RC date, and cites InfoQ's explicit schedule table alone for the Sept 14 GA date. That is a legitimate editorial choice (the InfoQ roundup is the primary report of Oracle's approved release schedule, directly attributed to Mark Reinhold; InfoWorld's Sept 15 appears to be that outlet's own rounding/error) rather than a fabrication or an undisclosed pick between two conflicting numbers attributed to the same claim. No arithmetic was needed or performed in the published body text to reach Sept 14 — it is a direct, literal quote from source-2's schedule table. Separately, cross-checked Reinhold's 'five weeks and four days' / 'three weeks and four days' figures against the calendar (Aug 6 + 39 days = Sept 14; Aug 20 + 25 days = Sept 14) and both resolve consistently to Sept 14, corroborating InfoQ's figure independent of the schedule table. All nine JEP numbers, preview/incubation rounds, and descriptions check out against the two InfoQ roundups and the InfoWorld feature rundown. The one substantiated issue is the misquoted CPU-frequency phrase documented above.
Overall Assessment: Substantively accurate, well-sourced, neutral News piece. Headline, summary, and lead are all directly supported by the InfoQ Aug 3 roundup. The Sept 14 vs Sept 15 GA-date question was investigated in detail and resolved: the article correctly and exclusively relies on InfoQ's explicit schedule table, not on arithmetic and not on InfoWorld's differing pre-delay figure. The single recoverable issue is a misquoted short phrase that does not affect the headline, summary, or the article's central claims. Approving with a corrections record.