MSwitching to Washington and money: OpenAI, Anthropic, and Nvidia all reportedly posted record lobbying quarters.
CA Tesla fan account on X is claiming Nvidia spent two point two million, Anthropic two point one, OpenAI one point five — just July through September.
MI checked. No federal disclosures on those exact numbers yet. Still, these companies have all been ramping up Washington presence for years.
CAnd no surprise. AI policy is the next big battleground: export controls, copyright, safety regulations.
MMicrosoft already reportedly spends over two million a quarter. Google similar. The whole industry is trying to shape the rules before the rules get written.
CI'd rather they spend it on compute and engineers than lobbyists. But when the government is writing rules that could kill your product, you show up.
MFair. They're under pressure from every direction.
CDifferent world entirely: over on Hacker News, a developer named Lukas Werner built a VGA graphics driver from scratch in SystemVerilog.
MSystemVerilog? That's normally for chip design, not old-school display drivers.
CHe posted a full write-up. One commenter said it made him miss the character-based text modes and hardware cursors of old VGA cards.
MIt's a nostalgia trip for people who remember garbage collecting in mode 13h.
CMode 13h. You're dating yourself.
MI'm just saying, that era taught people to care about every single clock cycle.
CAnd that's the beautiful thing about this project: it's not useful as a product. It's useful as an education in how hardware actually talks to software.
MSpeaking of ambition, another Hacker News thread links to an essay by Dan Luu arguing we should build more ambitious software.
CI read that. His point is we've spent a decade optimizing the same CRUD apps while the hard problems sit untouched.
MSo what does he want? More people reinventing WordPerfect?
CMore people building tools that actually change how we work, not just moving buttons around. Spreadsheets, databases, CAD, compilers — the unglamorous stuff.
MBut the market pays for small, safe improvements. Nobody's funding a new operating system without a business model.
CThat's exactly his point. We've optimized for the business model, not the breakthrough.
MI half agree. Some of the most ambitious software today is open source and badly underfunded, and it runs the entire internet.
CAnd then companies with millions of users build on it and donate a coffee mug.
MWhich is probably why one Hacker News commenter is pushing Argus, a free open-source security operations platform.
CMIT licensed, agentic, security operations. It's the kind of thing a thousand startups are trying to sell for six figures a year.
MOpen source security tooling is having a moment — CrowdStrike's gaffe last year made a lot of teams look for options they can actually inspect.
CAnd agentic is the magic word. Give it the ability to act, not just alert, and you've got something genuinely different.
MAs long as the agent doesn't decide to delete prod during an investigation. That's my fear with all of these.
CThat's what the MIT license is for. You can read exactly how it's going to delete prod.
MOne more from Hacker News: an article about small programs and languages, and why they still matter.
CThe argument is that Go and Rust and all the modern toolchains have made binaries enormous, and we've forgotten the joy of a tight little program.
MCounterpoint: the tight little program can't handle Unicode, streaming video, or real-time collaboration.
CCorrect. But not every program needs all that. A pastebin with a web framework and a million dependencies? That's absurd.
MWe do have a complexity problem. Some of these 'small language' projects are really about reclaiming predictability in a world of moving parts.
CThe old Unix philosophy: do one thing well, and let the user pipe the rest.
MUntil your one thing has a CVE in its JSON parser and you're patching a museum.
CYou just can't let us have the nostalgia.
MNostalgia's fine. I just prefer my nostalgia documented, like that VGA driver, not running production.