Anthropic Threw Away Git
Comments on “Anthropic Threw Away Git”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Anthropic says it replaced git for its monorepo with Scion, a centralised, server-authoritative system built by seven people in five months. Merged changes rose from about 2,500 a day in May to over 30,000 in September, and CI alone runs 25 million checkouts daily. Anthropic reports a 200x smaller footprint than its old mirror fleet and checkouts two orders of magnitude faster. It is not open-sourcing Scion, and it suggests git forges could expose server-side APIs instead.
Why it ranked
No. 9 of 110 with a score of 0.67. The panel rated it highest on new information (0.92) and reader interest (0.85), lowest on substantiated (0.61). Placed No. 9, No. 15 by score alone: the placement terms moved other stories past it.
- AI judgeJudge 2claude-haiku-5-5
- Detailed engineering write-up of a bespoke VCS built at scale, with concrete numbers and design reasoning, though it is a single company's internal system, not open source, so practical use is limited.
- AI judgeJudge 3gpt-5.6-luna
- Detailed analysis of a novel, technically specific source-control redesign for extreme agent and monorepo scale, with concrete architecture and performance figures. Its broader timeline and general applicability are somewhat speculative.
Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.
Score
0.67total under rubric v2.3
- Significance0.67
- Reader interest0.85
- Audience breadth0.64
- Practical relevance0.65
- Substantiated0.61
- Technical depth0.84
- New information0.92
- Counts against: Promotional0.08
- Counts against: Overclaim0.38
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| SignificanceHow much does this change what people in tech know or can do? | 0.67 | 0.55 | 0.78 | 0.67 | 0.23 | +0.24 | +0.160 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.81 | 0.82 | 0.93 | 0.85 | 0.12 | +0.26 | +0.222 |
| Audience breadthHow many people who build technology does it concern? | 0.62 | 0.67 | 0.62 | 0.64 | 0.05 | +0.13 | +0.083 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.69 | 0.50 | 0.76 | 0.65 | 0.26 | +0.22 | +0.143 |
| SubstantiatedHow well is it backed? Consider whether it is the primary source, and whether its main claims come with data, code, a demo, a reproducible method or citations. | 0.58 | 0.45 | 0.79 | 0.61 | 0.34 | +0.05 | +0.030 |
| Technical depthHow technically substantive is it? | 0.91 | 0.70 | 0.91 | 0.84 | 0.21 | +0.05 | +0.042 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.89 | 0.90 | 0.98 | 0.92 | 0.09 | +0.05 | +0.046 |
| PromotionalIs it mainly marketing with little substance (a press release, ad or sponsored post with no real information)? A launch post that carries real news is not promotional. | 0.08 | 0.12 | 0.03 | 0.08 | 0.09 | −0.19 | −0.015 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.55 | 0.30 | 0.29 | 0.38 | 0.26 | −0.11 | −0.042 |
| Sum of contributions | +0.669 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.669 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-10-29/9/"><img src="https://rubric.hrih.si/badge/2026-10-29/9.svg" alt="Ranked #9 on Rubric, 29 Oct 2026" width="238" height="28"></a>Claim check
Claim check: 7 matches source · 1 no source given. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
Matches sourceNumber
Anthropic built Scion, a centralised, server-authoritative version control system, with seven people in five months and fully migrated its monorepo off git.
The cited source says the same thing.
The article says “Five months, seven people, while the volume of code merged per day grew tenfold underneath them”
The source says “A team of seven people built both Scion and Antique (our code review tool, which we could easily write a whole other blog post about), not to mention half a…”
Source cited by the article: anthropic.com/engineering/redesigning-source-control
Matches sourceNumber
Anthropic's merged changes per day rose from roughly 2,500 in May 2026 to over 30,000 in September 2026.
The cited source says the same thing.
The article says “May 2026: roughly 2,500 changes merged per day”
The source says “In May 2026, Anthropic averaged around 2,500 code changes merged per day; as of September, we merged over 30,000 per day”
Source cited by the article: anthropic.com/engineering/redesigning-source-control
Matches sourceNumber
Anthropic runs about 25 million checkouts per day for CI alone, which fetched naively from git would be over 600 petabytes of traffic per day.
The cited source says the same thing.
The article says “They run on the order of 25 million checkouts per day for CI alone”
The source says “We perform on the order of 25 million code checkouts per day for CI, before counting developer environments, agent sandboxes, and research jobs. Fetched naively, that would be over 600…”
Source cited by the article: anthropic.com/engineering/redesigning-source-control
Matches sourceNumber
Anthropic's main branch history is near 100 GB and a compressed working copy is over 25 GB.
The cited source says the same thing.
The article says “main’s history is near 100 GB, and a compressed working copy is over 25 GB”
The source says “The history of our main branch alone is almost 100 GB, and a compressed working copy is over 25 GB”
Source cited by the article: anthropic.com/engineering/redesigning-source-control
Matches sourceNumber
Anthropic reports Scion's cache tiers sustain a >99% hit rate, its infrastructure footprint is 200x smaller than the old git mirror fleet, checkouts are two orders of magnitude faster, and CI recovered 25% of its CPU capacity.
The cited source says the same thing.
The article says “every tier sustains a >99% cache hit rate; the infrastructure footprint is 200x smaller than the old git mirror fleet in CPU and memory”
The source says “Each tier of this system maintains a cache hit rate of over 99% and does not require long-lived state, so it’s easy to scale up if needed”
Source cited by the article: anthropic.com/engineering/redesigning-source-control
Matches sourceNumber
An early internal Mythos snapshot rewrote Anthropic's handwritten git-to-Scion ingestion pipeline to be 20x leaner on memory and 4x faster, preserving all invariants.
The cited source says the same thing.
The article says “it came back 20x leaner on memory and 4x faster, with every implicit ordering and data-structure invariant intact”
The source says “it produced a version that consumed 20x less memory and was 4x faster, while preserving all of the implicit invariants around ingestion ordering and data structures”
Source cited by the article: anthropic.com/engineering/redesigning-source-control
Matches sourceDate
Anthropic states that this class of problem will become commonplace for others within six to twelve months.
The cited source says the same thing.
The article says “Their stated timeline for everyone else is six to twelve months before this class of problem becomes commonplace”
The source says “we anticipate that this class of problems will become more commonplace over the next six to 12 months”
Source cited by the article: anthropic.com/engineering/redesigning-source-control
No source givenCausal conclusion
The author concludes the Anthropic problem was a design mismatch, not a scale problem: git's distribution costs buy offline autonomy that its agent and CI clients don't need.
The article gives no link or citation for it.
The article says “they were paying git’s entire distribution tax – full local copies, client-side computation, packfile negotiation, repository maintenance – to buy offline autonomy that none of their clients wanted”
The source says No source to compare with.
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 2 pts, not on front page
Recorded for context only, checked 29 Oct 2026, 05:47 UTC. It played no part in the ranking.