| | | Andy Pavlo’s annual databases year-in-review is back. It's equal parts technical insight, industry gossip, and strong opinions. If you care about where Postgres, sharding, MCP, and modern data systems are heading, this is a must-read!
Carnegie Mellon University | Andy Pavlo |
| | |
| | This 2025 recap is a great map of the year in LLMs. Reasoning models, coding agents, CLI workflows, open-weight leaps, evals, safety failures-in-waiting, and the “$200/mo” era, framed with sharp takes and lots of links to dig into.
Simon Willison |
| | |
| | Many of the best ideas in R and Python tooling didn’t originate where they’re used today. This post traces how patterns around UX, docs, and developer tooling move between ecosystems, and what that cross-pollination gets right.
Emily Riederer |
| | |
| | Advent of Code... in SQL? This post walks through solving AoC 2025 using a single, pure ClickHouse query per puzzle: no UDFs, no preprocessing, no temp tables. Equal parts impressive, cursed, and illuminating about what modern SQL engines can actually do.
Clickhouse Blog | Zach Naimon |
| | |
| | | | Box plots break down fast with heavy skew. This post digs into why, then experiments with dot plots, violins, and new “quintile area” strips. The best part: it shows where the alternatives fail too, and why box plots still earn their place.
Xan Gregg |
| | |
| | Why do some charts feel effortless to read? By breaking down Datawrapper figures from a New America report and rebuilding them in ggplot2, this post shows how design details, not fancy charts, do most of the work.
Daniel Roelfs |
| | |
| | | This new book builds intuition for how machine learning works, without heavy math. Clear visuals explain prediction, distance, evaluation, trees, preprocessing, and an end-to-end project helps round it out. Free to read online.
Eliott Kalfon |
| | |
| | This post turns Python performance folklore into concrete numbers. How fast is a dict lookup? How expensive is async? How much memory do objects really use? Includes clear tables, sane comparisons, and an excellent “mental model” resource.
Michael Kennedy |
| | |
| | | This rapid-fire interface began as a joke about the flood of ML/AI papers, but if you skip the timer and advance it manually, it's actually useful for browsing what's new.
Mark J. Nelson |
| | |
| | | |
|
