Jepsen: ParticleDB 1.1.4
Comments on “Jepsen: ParticleDB 1.1.4”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Jepsen tested ParticleDB, a multi-model database for AI workloads built by Orkes, from versions 1.1.4 through 1.1.18.11.rc-22, finding 96 issues ranging from crashes and data loss to anomalous query results.
Why it ranked
No. 27 of 59 with a score of 0.56. The panel rated it highest on technical depth (0.95) and new information (0.94), lowest on audience breadth (0.41). The claim check took 0.100 off. Placed No. 27, No. 53 by score alone: the placement terms moved other stories past it.
- AI judgeJudge 2claude-haiku-5-5
- Extensive, well-evidenced Jepsen-style testing of a database's correctness; highly technical and practically relevant to database builders, though scope is one niche product and the text is truncated.
- AI judgeJudge 3gpt-5.6-luna
- A detailed, reproducible audit of a new AI-oriented database uncovers severe correctness, durability, and compatibility failures, with extensive test methodology and version history. It is highly useful, though focused on one emerging product.
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
0.56total under rubric v2.3
- Significance0.56
- Reader interest0.79
- Audience breadth0.41
- Practical relevance0.71
- Substantiated0.93
- Technical depth0.95
- New information0.94
- Counts against: Promotional0.06
- Counts against: Overclaim0.16
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? | 0.55 | 0.45 | 0.68 | 0.56 | 0.23 | +0.24 | +0.134 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.66 | 0.80 | 0.90 | 0.79 | 0.24 | +0.26 | +0.205 |
| Audience breadthHow many people who build technology does it concern? | 0.24 | 0.55 | 0.45 | 0.41 | 0.31 | +0.13 | +0.054 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.65 | 0.70 | 0.78 | 0.71 | 0.14 | +0.22 | +0.156 |
| 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. | 0.95 | 0.90 | 0.94 | 0.93 | 0.05 | +0.05 | +0.046 |
| Technical depthHow technically substantive is it? | 0.98 | 0.92 | 0.95 | 0.95 | 0.06 | +0.05 | +0.048 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.92 | 0.90 | 0.99 | 0.94 | 0.09 | +0.05 | +0.047 |
| 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. | 0.13 | 0.05 | 0.01 | 0.06 | 0.12 | −0.19 | −0.012 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.30 | 0.10 | 0.08 | 0.16 | 0.22 | −0.11 | −0.018 |
| Sum of contributions | +0.660 | ||||||
| Claim-check penalty | −0.100 | ||||||
| Total, as published | 0.560 | ||||||
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/2026-11-01/27/"><img src="https://rubric.hrih.si/badge/2026-11-01/27.svg" alt="Ranked #27 on Rubric, 1 Nov 2026" width="238" height="28"></a>Claim check
Claim check: 2 matches source · 1 differs from source · 3 no source given · 1 couldn’t check. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
No source givenNumber
Jepsen found 96 issues in ParticleDB across versions 1.1.4 through 1.1.18.11.rc-22.
The article gives no link or citation for it.
The article says “We present 96 issues in ParticleDB, ranging from the pedestrian (crashes, deadlocks, data loss, etc”
The source says No source to compare with.
No source givenNumber
By version 1.1.18.11.rc-22, all but ten of the issues appeared resolved, but ParticleDB still lost committed writes with the default occ mode and allowed anomalies and Lost Update with serializable.
The article gives no link or citation for it.
The article says “By version 1.1.18.11.rc-22, all but ten of those issues appeared resolved”
The source says No source to compare with.
Couldn’t checkQuote
Orkes claims that version 1.1.19-rc.13 resolves all remaining issues discussed in the report.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “they claim that it resolves all remaining issues discussed in this report”
The source says No passage could be compared.
Source cited by the article: orkes.io/blog/particledb-database-built-for-ai-agents
Matches sourceFirst, only or fastest
ParticleDB's home page advertised that it was "formally proven" with a TLA+ proof, but no TLA+ specification was available during the collaboration.
The cited source says the same thing.
The article says “ParticleDB’s home page advertised that ParticleDB was “formally proven” with a TLA+ proof”
The source says “TLA+ Formally Proven”
Source cited by the article: web.archive.org/web/20260518203012/https://particledb.ai
Matches sourceQuote
ParticleDB's home page claimed "Full ACID. 159K TPS." while the text below it said "snapshot isolation", which is not fully isolated.
The cited source says the same thing.
The article says “ParticleDB’s home page proudly claimed “Full ACID. 159K TPS”
The source says “Full ACID. 159K TPS. MVCC with snapshot isolation. Concurrent reads never block writes. WAL for crash recovery”
Source cited by the article: web.archive.org/web/20260518203012/https://particledb.ai
No source givenNumber
In version 1.1.18.11.rc-18, strict-serializable mode showed Stale Read roughly every 300,000 transactions in one testing environment.
The article gives no link or citation for it.
The article says “these anomalies appeared roughly every 300,000 transactions, using Elle’s exponential key distribution and 10 concurrent clients pushing ~80 transactions per second”
The source says No source to compare with.
Differs from sourceCausal conclusion
In 1.1.18.11.rc-18, ParticleDB started up despite WAL checksum mismatches after bitflips and lost acknowledged writes (44 elements of key 1754).
The cited source says something else; both quotes are shown.
The article says “ParticleDB would discover that its data files were corrupt, then start up anyway, serving data to clients”
The source says “For example, it forgot about every write made to key `1774` somewhere between 102.15 and 131.61 seconds”
Source cited by the article: github.com/particledbai/jepsen/issues/108
Also covered by
Elsewhere
Elsewhere: HN 235 pts, front page
Recorded for context only, checked 1 Nov 2026, 05:47 UTC. It played no part in the ranking.