An Aggregate Is a Function, Not a Sum
Comments on “An Aggregate Is a Function, Not a Sum”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Alexander Novikov describes how his team built sales-receipt reports on ClickHouse after four designs failed on duplicates, late documents and outages. The design that held treats an aggregate as a function of its documents: hourly per-register cells are recomputed from a narrow table when marked dirty, then versioned and replaced. On a small two-shard test cluster, the 600-shift report fell from 23.4 s to 2.1 s. The figures are the author's own and unsourced, and replica recomputation is an acknowledged gap, untested because the cluster had one replica per shard.
Why it ranked
No. 10 of 109 with a score of 0.69. The panel rated it highest on technical depth (0.94) and substantiated (0.89), lowest on audience breadth (0.50). Placed No. 10, No. 13 by score alone: the placement terms moved other stories past it.
- AI judgeJudge 2claude-haiku-5-5
- Detailed ClickHouse engineering writeup on incremental aggregation with dedup and late data; highly technical and practical for data builders, though niche in audience and not landmark significance.
- AI judgeJudge 3gpt-5.6-luna
- A detailed, evidence-backed ClickHouse production design for deduplicated, late-arriving aggregates, including failed approaches, implementation mechanics, benchmarks, correctness checks, and explicit replica limitations. It is broadly useful to data and infrastructure practitioners.
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
0.69total under rubric v2.3
- Significance0.54
- Reader interest0.80
- Audience breadth0.50
- Practical relevance0.81
- Substantiated0.89
- Technical depth0.94
- New information0.88
- Counts against: Promotional0.03
- Counts against: Overclaim0.15
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? | 0.56 | 0.35 | 0.70 | 0.54 | 0.35 | +0.24 | +0.129 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.80 | 0.70 | 0.90 | 0.80 | 0.20 | +0.26 | +0.208 |
| Audience breadthHow many people who build technology does it concern? | 0.55 | 0.40 | 0.55 | 0.50 | 0.15 | +0.13 | +0.065 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.83 | 0.70 | 0.90 | 0.81 | 0.20 | +0.22 | +0.178 |
| 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. | 0.97 | 0.80 | 0.90 | 0.89 | 0.17 | +0.05 | +0.045 |
| Technical depthHow technically substantive is it? | 1.00 | 0.85 | 0.98 | 0.94 | 0.15 | +0.05 | +0.047 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.82 | 0.90 | 0.93 | 0.88 | 0.11 | +0.05 | +0.044 |
| 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. | 0.04 | 0.05 | 0.01 | 0.03 | 0.04 | −0.19 | −0.006 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.22 | 0.15 | 0.08 | 0.15 | 0.14 | −0.11 | −0.017 |
| Sum of contributions | +0.693 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.693 | ||||||
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/2026-11-06/10/"><img src="https://rubric.hrih.si/badge/2026-11-06/10.svg" alt="Ranked #10 on Rubric, 6 Nov 2026" width="238" height="28"></a>Claim check
Claim check: 1 matches source · 7 no source given. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
No source givenNumber
In the second attempt, a single recompute pass read 46 million rows to use 508 thousand.
The article gives no link or citation for it.
The article says “One pass read 46 million rows to use 508 thousand”
The source says No source to compare with.
No source givenNumber
In the fourth attempt, a report over 600 shifts computed from documents read about 3.5 billion rows.
The article gives no link or citation for it.
The article says “A report over 600 shifts read about 3.5 billion rows”
The source says No source to compare with.
No source givenNumber
With cells, the 600-shift report reads up to 360 thousand rows instead of 3.5 billion.
The article gives no link or citation for it.
The article says “The 600-shift report reads up to 360 thousand rows instead of 3.5 billion”
The source says No source to compare with.
No source givenNumber
The narrow table reads 30 to 37 times fewer rows than the wide table for the same minute of traffic.
The article gives no link or citation for it.
The article says “the narrow one reads 30 to 37 times fewer rows”
The source says No source to compare with.
No source givenCausal conclusion
Collapsing journal marks reduced them from about 14 per cell per hour to 1.16, making the main recompute pass five times cheaper.
The article gives no link or citation for it.
The article says “Small inserts produced about 14 marks per cell per hour; after collapsing it was 1.16, and the main recompute pass became five times cheaper”
The source says No source to compare with.
No source givenNumber
Smaller granules and compressed blocks cut decompression from about 1.1 GB to about 0.3 GB and CPU by about 30% for 9% more disk.
The article gives no link or citation for it.
The article says “The same query decompressed about 0.3 GB, and CPU dropped by about 30% for 9% more disk”
The source says No source to compare with.
No source givenBenchmark result
Zero mismatches were found across 31 checks against an independent reference after three hours of live traffic.
The article gives no link or citation for it.
The article says “Zero mismatches across 31 checks, after three hours of live traffic in steps of increasing load, the last at three times the first”
The source says No source to compare with.
Matches sourceGeneralization
Oracle's partition change tracking implements incremental view maintenance for materialized views.
The cited source says the same thing.
The article says “Oracle's partition change tracking does exactly that for materialized views”
The source says “There are three incremental refresh methods: - log-based refresh - partition change tracking (PCT) refresh - logical partition change tracking (LPCT) refresh”
Source cited by the article: docs.oracle.com/en/database/oracle/oracle-database/21/dwhsg/refreshing-materialized-views.html
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 1 pts, not on front page
Recorded for context only, checked 6 Nov 2026, 05:47 UTC. It played no part in the ranking.