Gradient descent is the preferred way to optimize neural networks and many other machine learning algorithms but is often used as a black box. This post explores how many of the most popular gradient-based optimization algorithms such as Momentum, Adagrad, and Adam actually work.
“Keep thy airspeed up, lest the earth come from below and smite thee.” - William Kershner
Four years after my post on doing network captures on macOS with Process ID, Wireshark 4.6.0 has been released which includes support for parsing this…
Writing a Hypervisor in 1,000 Lines
As we saw in the opening post, SQLite stores metadata about tables in a special "schema table" starting on page 1. We've been reading records from this…
A look at the experiences of activists in East Germany (GDR) dealing with Stasi informants and infiltration before the fall of the Berlin Wall in 1989.
Here’s a Mastodon thread from a year ago. Just a quick check on how ChatGPT’s getting on … Me: Who reassigned the species Brachiosaurus brancai to its own genus, and when? ChatGPT…
I define "intelligence" as being capable of knowing or understanding, at least within some domain. ChatGPT cannot know or understand anything, so it is not intelligence. It does not know what its output means. It has no idea that words can mean anything.
Transform Your Time Management Forever! Discover how switching to UTC eliminated time zone confusion, boosted productivity, and simplified global scheduling for remote work.
How building edge-case first led to bloated, overly-granular libraries and what we can do about it
Within free open source software (and free content more generally), there are two major categories of copyright licenses:
Released today, a report of a study investigating the project’s feasibility will serve as input for the European Strategy for Particle Physics and be assessed by the CERN Council in the coming months
A friend called me last week. Someone who’d built web applications back for a long time before moving exclusively to backend and infra work. He’d just opened a modern React codebase for the first time in over a decade.
At one level, this question is not very fair, because an answer to it as stated would be that one simply does not use doubly linked lists. They have been popular in introductory computer science lectures because they are a neat way to explain pointers in a data structure that's easy to draw on a whiteboard, but they…
The Matrix Conference 2025 was a huge success, and the energy and enthusiasm was just incredible!
The second birthday of ChatGPT was only a little over a month ago, and now we have transitioned into the next paradigm of models that can do complex reasoning. New years get people in a reflective...
An extensive guide with over 40 tips to learn how to reduce your screen time. Understand the causes and meet your objectives with these tools and techniques.
Using LLMs doesn't make a product special anymore
The gophers nested too greedily and too deep.
DAVIS, California – November 18, 2025 – The Connectivity Standards Alliance (Alliance) announced today the release of Zigbee 4.0 and Suzi, the new brand for Zigbee’s Sub-GHz feature…
React2Shell (CVE-2025-55182) is a critical RCE vulnerability in React Server Components. Learn which versions are impacted and how to mitigate.
Timeline & Responsible Disclosure
Recently a colleague mentioned building NixOS images that run unchanged on multiple architectures. Given the past adventures on this blog with systemd-repart and cross-compiling NixOS, I decide to give this a go.
The one constant that I have observed in my professional life is that people underestimate the need to move fast. Of course, doing good work takes time. I once spent six months writing a URL parser. But the fact that it took so long is not a feature, it is not a positive, it is … Continue reading Why speed matters
Stephen Wolfram explains the rich ruliology of lambdas, made particularly significant by their connection to practical computing. Covers basic computations to undecidability to multiway graphs and evaluation strategies.
Asynchrony, fast approximate exponents, and 10x more efficient softmax.
I have a spicy git alias that allows me to "archive" old git branches by converting them to tags, making them less visible across git tooling. # ~/.gitconfig…
To start off, we can leverage the STL for this calculation. std::count_if() more precisely:
Pydantic’s JSON loading uses a huge amount of memory; here’s how to reduce it.
Jeffrey Yasskin recently pointed out an interesting security bug: The idea is, if you had registered googlelogoligature.net then Chrome on Android (and possibly other Google products) would have displayed it as Google.net, potentially tricking users into thinking they were really interacting with Google. To see how…
