content(newsletter): publish issue 403 and soften two selection rules to suggestions - #144
Merged
Merged
Conversation
… to suggestions Issue 403 covers GPT-6 Astra, Netflix on the LLM-as-a-judge lifecycle, Databricks on agentic GPU kernel generation, Google's TimesFM-3 and the Polars 2.0 release candidate. Every source was grounded from the page being linked, with Astra and the Netflix post needing a logged-in browser because both return 403 to a plain fetch, and the Polars quotes re-read directly since only one fetch tier reached that domain. Vendor figures are attributed to whoever claims them rather than asserted, which is why the Astra section carries both the saturated benchmarks and the two places OpenAI report losing to Claude, plus the monitorability regression they disclose. The article sections separate each sentence with a blank line. This departs from every issue since 340, which ran one paragraph per section, and renders as separate paragraphs on the site and in the email. The events block gains two corrections the scaffold could not make on its own, since it copies the previous issue forward blind: Economic Forum 2026 was in events.yaml as a speaking slot but missing from the list, and World Summit AI Europe was listed as September when it runs 7-8 October. MLSys 2027 joins events.yaml on the owner's call, overriding the scout's paper-venue rejection. The organiser confirms Santa Clara in May 2027 but has not announced program dates, so the start date is a placeholder carrying an '# unverified' marker. AI_dev Europe was also approved but could not be added: the Linux Foundation folded the brand into Open Source Summit and no 2027 edition exists to point an entry at. Selection rules 10 and 13 become explicit suggestions. Both were descriptive statistics written as prescriptions: rule 10 asked for a named-individual opinion piece every week from a corpus measurement of 0.60 per issue, which is a quota that rule 15 already bans, and rule 13 applied a cross-issue tag weighting as a per-issue composition check. Neither should have counted against a shortlist.
Visual parity — dark theme 🌙1440x1000 — 7 of 40 routes moved.
420x900 — 7 of 40 routes moved.
Differing pixels locate a change; the peak channel delta sizes it. A large count with a Informational: label the PR |
Visual parity — light theme ☀️1440x1000 — 7 of 40 routes moved.
420x900 — 7 of 40 routes moved.
Differing pixels locate a change; the peak channel delta sizes it. A large count with a Informational: label the PR |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Issue 403 covers GPT-6 Astra, Netflix on the LLM-as-a-judge lifecycle, Databricks on agentic GPU kernel generation, Google's TimesFM-3 and the Polars 2.0 release candidate.
Grounding
Every source was grounded from the page being linked. Astra and the Netflix post both return 403 to a plain fetch and needed a logged-in browser; the Polars quotes were re-read directly because only one fetch tier reached that domain. Vendor figures are attributed to whoever claims them rather than asserted, which is why the Astra section carries the saturated benchmarks alongside the two places OpenAI report losing to Claude, plus the monitorability regression they disclose.
Formatting change worth a look
The article sections separate each sentence with a blank line, so each section renders as several paragraphs rather than one. Every issue since 340 ran one paragraph per section, so this is a deliberate departure and the reason to review the rendered page rather than only the diff.
Events
Two corrections the scaffold could not make on its own, since it copies the previous issue's block forward blind:
events.yamlas a speaking slot but missing from the list.MLSys 2027 joins
events.yaml. The organiser confirms Santa Clara in May 2027 but has not announced program dates, so the start date is a placeholder carrying an# unverifiedmarker. AI_dev Europe was also approved but could not be added: the Linux Foundation folded the brand into Open Source Summit and there is no 2027 edition to point an entry at.Selection rules
Rules 10 and 13 become explicit suggestions. Both were descriptive statistics written as prescriptions. Rule 10 asked for a named-individual opinion piece every week from a corpus measurement of 0.60 per issue, which is the kind of quota rule 15 already bans; rule 13 applied a cross-issue tag weighting as a per-issue composition check. Neither should have counted against a shortlist.
Verification
npm run check,check:voiceand prettier are all clean. The style linter reports two warnings that are expected and deliberately left standing: 1 of 5 colon-ledes against a corpus rate of 58%, and 1 semicolon against ~3.4. Both are frequency tells rather than errors, and individual issues swing hard around those averages.Not included
Post images, candidate-pool bookkeeping and backlink syncing all happen after publication.