Every Data issue
87 issues since 26 May 2026, newest first. Each one is the day's Data news, with links to the original sources.
August 2026
- DuckDB joins AWS, 4-bit model beats bfloat16, dashboards lie
- MetaRoCE rewrites Ethernet rules for AI clusters
- Netflix kills its own autoscaler, open source wins
- The reproducibility trap
- Hudi cuts Uber's DSAR scans, 1914 viz rules still hold, ASR benchmaxxing measured
- Rare books destroyed for AI; Google buys Spirit data
- Polars 2.0 RC, Iceberg v3, and Netflix's Flink ad pipeline
- DataFusion Comet 1.0, Depop slashes Spark costs 52%
- One Swapped Digit Beats All Hallucination Detectors
- Snowflake overflows, DuckDB outruns Spark, and a 16-year-old bug
- Netflix retires batch queueing; Fory JSON 10.9x Jackson
- Autodesk splits 100-TB index as Tinder migrates Elasticsearch
- AI agents auto-heal data pipelines but need two-layer ontology
- NATS 2.12 atomic batches, Netflix Maestro, Salesforce cuts 97%
- IKEA’s 20,000‑item index and why early‑stopping A/B tests are broken
- Ontologies for AI Agents, Warehouses Fail, Postgres+ClickHouse
- Flock’s spy net, Netflix’s 8B‑node query, and the semantic‑layer trap
- AI tutors over-help: new framework reveals LLMs' blind spot
- Airflow hits seconds, new energy math flips fossil shares
- Etsy open-sources Kafka skills for Claude Code
- Polars scales 16B rows from laptop to cluster
- Lawn-mowing game solves TSP; Google pulls AI satellite
- Meta Doubles Training Efficiency, Datatype Font Charts in Text
- Databricks slashes query latency 30x, Halodoc cuts CPU 30%
- Netflix dumps Kafka; prompts shrink 74%, hit 75ms
- State Dept AI map misplaces all African countries
July 2026
- EvoLib & Google Science One: Self-fixing LLMs, verifiable AI
- Iggy drops Kafka, Atlassian adopts, Airbnb slashes failures 80%
- Open standards fail data lakes double-count sensors
- Knowledge spines, lower costs, president’s billions
- Zero-copy isn't free: Inside Hotstar's real-time ad engine
- On-disk ANN indexes cut vector search costs by 90%
- Startups bet on DataFusion, AI tracing, agent runtime
- Iran war slashes gas supply, fertilizer up 30%
- DuckDB Iceberg CRUD, Text-to-SQL Only 10% Accurate
- Fix RAG Hallucinations, Simulation Fuels Physical AI
- Data management is the gatekeeper of generative AI success
- Comet cuts latency, Microsoft's Ontology, AI agents need context
- Self-serve BI is real, AI agents face CFO hurdle
- Satellite paint flaw, and why classical ML still beats LLMs
- Hudi, Iceberg, Delta Lake bring ACID to object storage
- Prefect acquires Dagster, Salesforce queries S3 data lakes
- Meta slashes ad latency 28% with BPF, Snowflake helps win gold
- AI code flood threatens open source and context rot
- Subject: Airflow 3.3 auto-fixes drift, Iceberg eats Hive, medallion cracks
- LinkedIn slashes data traffic 99% with homomorphic encryption
- Persistent neural state, GraphRAG, and AI hallucination fixes
- NIST bans FF3, Kafka 4.3 linger.ms, 92°F trick
- Iceberg v3 slashes JSON queries, graph analytics on a laptop
- SQL’s analytics win hides deeper infrastructure truth
- Iceberg Lakehouse Cuts BigBasket Latency, CO₂ Storage Stalls
- StreamFusion rewrites Flink SQL, Norway opens 23 APIs
- RAG hallucinations fixed, prompt bloat cured, Neon DB scaled
- Glue and Iceberg Cut Log Query Times, Athena Gets Serverless Spark
- OpenSearch slashes log costs 70% as Hudi indexes live
- TabFM predicts any table without training
June 2026
- Memora cuts AI context tokens by 98%
- Kafka stalls, Parquet speeds up, CDP hits 30ms
- Cheap inference crushed satisfaction; logistic beats XGBoost
- Netflix kills Kafka, MCP reshapes BI, Sail 0.3 speeds Spark
- Zalando hits 1M RPS, Context Graph beats vector RAG
- BC's permanent DST shift will break Postgres timestamps
- Half of adults can’t afford healthcare
- Short-video data proves Polymarket payouts are fake
- PyCanopy trounces DuckDB spatial; Zepto boosts ClickHouse 45%
- Chukei Proxy 90% Snowflake savings, Databricks €3k waste
- Python 3.14 JIT, CUDA 8x speedup, Redis Streams challenge Kafka
- Claude audit catches data leaks LoRA falls short
- Warehouse-native MDM solves AI's identity crisis
- AI token reality check, 100M points in-browser, MCP unifies tools
- DataFusion gets SQL superpowers, Iceberg cuts latency
- IceStream kills Iceberg stale data faster
- Cloudflare Reports Bot Traffic Surpasses Human Web Traffic Globally
- PostgreSQL 19 Beta Ships Autoscaling I/O
- Deepfake Detectors Lose 20% Accuracy on Social Media
- Polars Distributed on K8s, Dagster Hikes 10x, Lakebase Branches
- Multi-Query Engines Cut Costs as AI Agents Flood Data Stacks
- dbt Core 2.0 Alpha, Ray Data, Bodo GPU Iceberg Writer
- ClickHouse launches CostBench; Iceberg 1.11 adds registerView
- Databricks Zerobus, Google Zero-Trust Analytics, Iceberg Routing
- Delta Lake Standard, Claude Builds Fabric Pipeline