I learned from Straight Dope that 'lorem ipsum' placeholder text is actually based on a classic speech by Cicero, not just random gibberish. It is a mangled version of his philosophical work De Finibus Bonorum et Malorum.
I discovered that the common filler text "lorem ipsum dolor sit amet" doesn't actually translate to anything meaningful in Latin. It is a garbled version of a classic Cicero text used specifically for typesetting filler.
I found [V]Host, an interesting text-to-speech web application that lets you embed Flash characters on your site to read text aloud. It offers a unique way to add interactive audio content using animated avatars.
I built a random quote generator bookmarklet that uses Order-2 word-level Markov chains to mix up text on any webpage. It leverages the Readability script to identify content, often producing funny or insightful nonsense.
I argue against using voice to communicate with computers because speaking requires significant physical energy and cognitive effort. Recent research supports my intuition that speech recognition might hinder thought processes during task execution.
I explored Claude 3.7's extended thinking capabilities, learned to bypass SQLite locks for browser history, and used chroot for Linux recovery. I also discovered Nomic's multimodal embeddings and new tools for AI-driven social research and speech cloning.