Introducing our state of the art video generation model Veo 3, and new capabilities for Veo 2.

Researchers at the University of Helsinki and Cambridge attempted to build a faster SQLite using modern programming paradigms like io_uring and disaggregated storage. They demonstrate up to a 100x reduction in tail latency. These are my notes.

In 1843, Ada Lovelace published the first nontrivial program. How did it work?

Longtime WordPress contributors ask: “Where are the adults in the room?”

“I do fear that the pressure to make better TVs will be lost…”…

On a mission to index all Ruby conferences. Your go-to place for talks and events about Ruby.

This post has made it to the Hacker News front page, see the discussion there. If you like this, you might also like my Instagram channel Nerd on Bars @nerdonbars where I calculate the power output of various athletes and myself.

Popeye the Sailor and the Belgian boy reporter Tintin lead the class of characters and works of art becoming public domain in 2025.

I decided to roll out my own LLM benchmark: how well can different models render an SVG of a pelican riding a bicycle? I chose that because a) I like …

We are adding a new generated code API to Go Protobuf.

Explore this post and more from the watchmaking community

As a child, Deirdre Sullivan learned from her father to always pay her respects at funerals. Now, this NPR listener believes those simple acts of human kindness are as important as the grand gestures.

Analysis of a 2022 GRUB2 fwsetup feature that causes a bootloop when loading a new config on an old GRUB2, and how it bit me two times

The European Union has signed a $11B deal to build a sovereign satellite constellation to rival Elon Musk's Starlink.

In 2024, we watched American life unfold beneath the golden arches—from Trump’s drive-through stunt to Luigi Mangione’s final hash brown.

OpenAI recently published a public writeup for an incident they had on December 11, and there are lots of good details in here! Here are some of my off-the-cuff observations: Saturation With thousa…

Universal Transformer Memory uses neural networks to determine which tokens in the LLM's context window are useful or redundant.