Logarithmic Spirals

OpenAI: AI is an existential threat to humankind. Also OpenAI: see how cute these little guys are? www.engadget.com/2272230/dots...

A group of 25 Fields Medalists, including myself, have made a joint declaration on Math and AI: mathandai.org . We welcome additional signatories. See also this article in the Economist announcing the declaration: www.economist.com/science-and-...

Having taken a quick survey of what people are currently building, there’s a lot of repetition. The same three or four ideas iterated on multiple times

Model-file size is only part of a local LLM’s memory budget. K/V-cache usage grows roughly linearly with active context, so moving from 8K to 32K needs about four times the cache memory before rounding. #localai

Article: my repair cost about $70 versus the $197.09 I paid for the Elite Series 2—about 36% of its original cost. That made repair worthwhile to me. https://logarithmicspirals.com/blog/xbox-elite-series-2-swollen-battery-steam-setup/

This article’s key lesson: a swollen lithium-ion battery makes safety the priority. Stop using and charging it, then follow local guidance for handling and disposal. https://logarithmicspirals.com/blog/xbox-elite-series-2-swollen-battery-steam-setup/ #LithiumIonBattery #iFixit

I built remark-csv-table, a Remark plugin that turns local CSV files into GFM tables from Markdown directives at build time. https://logarithmicspirals.com/blog/remark-csv-table-remark-plugin/ #Markdown #Astro #JavaScript

On my RTX 4070 Ti, Gemma 4 E4B stayed fully GPU-resident at 32K context and generated 88.5 tokens/s; I tested it alongside Qwen3 8B and other LLMs. https://logarithmicspirals.com/blog/what-llms-can-i-run-with-12gb-vram/ #LocalLLM #Ollama #GPU

I repaired my Xbox Elite Wireless Controller Series 2, then wrestled with Bluetooth, Steam Input, Xbox Game Bar, and firmware updates. https://logarithmicspirals.com/blog/xbox-elite-series-2-swollen-battery-steam-setup/ #Xbox #Steam #Repair

I reviewed Valve’s new Steam Controller: comfortable, precise TMR joysticks and useful Steam Input, but weaker rumble and Battlefield 6 compatibility issues. https://logarithmicspirals.com/blog/steam-controller-2026-review/ #SteamController

Want to run local LLMs but unsure if your GPU can handle the context? 🤖 I built a Local AI VRAM Calculator & GPU Planner to check K/V cache, estimate model fit, and project TPS. Try it here: logarithmicspirals.com/tools/local-ai-vram-calculator-gpu-planner #LLM #LocalAI #VRAM #GPU

Curious if your GPU upgrade is really worth the cost? Discover my new GPU Upgrade Value Calculator that dives deeper than benchmarks! Check it out here: https://logarithmicspirals.com/blog/gpu-upgrade-value-calculator-tool/ #GPU #Upgrade #TechTools

Ever wondered if your GPU setup is ready for local LLMs? Dive into my new AI VRAM Calculator & GPU Planner to uncover the truth! Check it out here: https://logarithmicspirals.com/blog/local-ai-homelab-planner-tool/ #AI #TechTools #VRAM

Experimenting with Codex this evening. It helped me delete my Hashnode integration, enhance my DEV integration, and bulk edit some old posts. Kind of nice to get an assist on those things since they would've taken a while to do by hand.

Ever wondered how to turn your NAS into a private LLM access point? Dive into my journey with Tailscale for secure connections and untethered data privacy at home! 🌐🔒 Read more here: https://logarithmicspirals.com/blog/using-tailscale-to-access-private-llms/ #TechTips #PrivacyFirst

Ever struggled with broken Markdown image URLs? 🖼️ I've automated stable, content-based URLs for seamless cross-posting on DEV and Hashnode—discover how I did it! ➡️ https://logarithmicspirals.com/blog/stable-markdown-image-urls-astro/ #MarkdownMagic #AstroJS

Happy to report my testing has revealed you can generate up to 35 episodes in around 2 minutes with podcast-it. Each episode in my test was about 2 minutes in length, funny enough . . .

Ran into gotcha with podcast-it. After deploying, I have found the worker doesn't actually make the episodes. It seems I need to rethink the architecture for this project 😔.

Just reached the final stage before I start serving my AI-generated podcast content. This final stage is me deploying podcast-it to a production environment, then rebuild my website with the necessary updates to host the episodes 😎

I just unlocked the secret to faster, reliable AI app development using promptfoo and Vitest—check out how it transformed my podcast project! 👉 https://logarithmicspirals.com/blog/from-prototype-to-production-promptfoo-vitest-podcast-it/ #AI #Programming #Innovation

Going to be making a follow up post on my podcast project within the next week. Stay tuned. I'll be talking about setting up LLM evaluations within a TypeScript project. #TypeScript #LLM #PromptEngineering #SideProject #Podcasting

Ever wondered how to turn blog posts into podcast episodes seamlessly? Discover the journey of building "podcast-it" with Cloudflare Worker here: https://logarithmicspirals.com/blog/from-idea-to-audio-building-podcast-it-cloudflare-worker/ #PodcastCreation #Cloudflare #OpenSource

Anyone else get burned out when a book has multiple prefaces, forwards, and introductions?

I've discovered that using SQLite for caching OpenAI embeddings in my Astro builds drastically reduces build times and costs—check out how you can optimize your site too! 🔗 https://logarithmicspirals.com/blog/sqlite-openai-embeddings-astro-builds/ #Astro #OpenAI #SQLite #WebDev

👀 Ever wondered why some "Similar Posts" seem so unrelated? I dove into refining my blog's recommendations and discovered a new approach using cosine similarity! Curious? Dive deeper here: https://logarithmicspirals.com/blog/refining-similar-posts/ #Blogging #AI #UX

I just discovered how to use AI-powered embeddings and cosine similarity to create a "Similar Posts" feature for my blog—transforming content discovery; check out how I did it: https://logarithmicspirals.com/blog/semantic-related-posts-astro-cosine-similarity/ #AI #Blogging #ContentDiscovery

That moment when you find a game in your steam library that you’ve never even started up before. #Steam #PCGaming

Really enjoying some of these talks at #SCaLE22x. A wide range of topics with plenty of opportunities to make new connections. Feeling energized to get started on new projects.

I've just implemented series support in my Astro-powered blog to boost content organization and cross-platform integration with DEV and Hashnode—here’s how I did it: https://logarithmicspirals.com/blog/implementing-series-support-in-astro/ #BloggingTech #AstroJS #ContentStrategy 🚀

I've just had an epiphany and added series support to my blog for a more cohesive and engaging reader experience! Discover how I'm transforming my content with this new feature: https://logarithmicspirals.com/blog/adding-series-support-to-my-blog/ #BloggingInnovation #ContentStrategy

I've upgraded from an AMD 6700 XT to an NVIDIA 4070 Ti, gaining a whopping 40.6% performance boost and access to NVIDIA’s fantastic features like DLSS and ray tracing—was it worth it? 🤔 https://logarithmicspirals.com/blog/amd-to-nvidia-6700xt-to-4070ti-comparison/ #GraphicsCard #Upgrade

Facing wildfires and valuing what truly matters, my 2025 roadmap aims to balance growth, creativity, and impact—discover the journey ahead! 🌱🎶🚀 Read more: https://logarithmicspirals.com/blog/my-2025-roadmap-balancing-growth-creativity-and-impact-6bf/ #Roadmap2025 #PersonalGrowth #Creativity

Encountered an interesting "gotcha" recently. Apparently, #cloudflarepages has a setting you need to turn off when using #AstroJS view transitions. It seems this is due to the Rocket Loader feature being on by default: developers.cloudflare.com/speed/optimi...

What a way to start the year. Hope people are staying safe tonight in Los Angeles.

Very much looking forward to #SCaLE22x this year.

Feeling tired from repetitive social media posts? Discover how I automated my BlueSky cross-posting with GPT-4 to save time and boost engagement! 🚀🔗 https://logarithmicspirals.com/blog/astro-bluesky-gpt4-cross-posting/ #SocialMediaAutomation #BlueSky #GPT4

The history of slipping on banana peels | Pretty Good, episode 14 www.youtube.com/watch?v=p8W5...

While it’s tempting to want to start new projects, ending the year by resolving bugs and tying off loose ends in existing projects will probably be more rewarding . . .

Just got done upgrading my website from #AstroJS v4 to v5. Wasn't too difficult, but there were some things I had to rewrite. Perhaps I can publish a blog post about it 🤔. #CodingLife

It seems some content on my website is no longer displaying. Not sure when that happened, but now I've decided to fix that while upgrading from Astro v4 to Astro v5 😜.

My custom table of contents component had some issues with code-heavy articles recently. Writing a custom Rehype plugin has helped, but it’s slow progress and still buggy. Anyone else deal with tricky components or custom plugins? Share your stories—I’m curious how you handled it! 🙌 #CodingLife