0:00 The ChatGPT desktop app bundles a full, hidden copy of LibreOffice, and its cache folder weighs in at a staggering one-point-seven gigabytes.0:08 That’s the fact, and it connects to a much bigger story about where the real work of AI is happening — not just in the cloud, but right on your machine.0:18 Last time we talked about the best of Hacker News, we were looking at the big ideas.0:23 This week, it’s all about what’s hiding in the code, and what it means for how we're all going to be working.0:30 So here's the lay of the land this week.0:33 The biggest concrete event is that discovery by developer Simon Willison.0:37 He popped the hood on the official ChatGPT/Codex desktop app and found...0:42 well, everything.0:43 A full Python and Node.js installation.0:46 Native binaries like Poppler for PDF rendering.0:48 And yes, a 430 megabyte copy of the entire LibreOffice suite.0:52 It’s not just a thin client calling an API; it’s a self-contained software bundle.0:58 We're going to come back to exactly why that matters.1:01 The counterpoint to that story comes from a developer who's gone in the complete opposite direction.1:07 Instead of using a bloated official app, they’ve built their own private AI backend on an M4 Pro Mac Mini.1:14 They're running a powerful, 35-billion-parameter model locally, on their own hardware, serving it to all their devices.1:21 Their argument is simple, and you’ll hear it a lot this week: cloud APIs are "rented land." This is about taking back control.1:29 Meanwhile, the cloud providers are fighting back.1:32 Anthropic just dropped new versions of their models, Claude Fable 5.1 and Mythos 5.1.1:38 The headline here is cost and privacy.1:40 They’re promising up to a forty-five percent cost savings on some workloads, and new enterprise safeguards that allow for zero data retention.1:49 They’re trying to make their rented land feel a lot more like private property.1:54 And while the platform war between local and cloud heats up, the tools we use to build everything are starting to creak under the strain.2:03 East River Source Control, or ERSC, just hired Martin von Zweigbergk as its new CTO.2:08 If that name sounds familiar, it's because he created Jujutsu, the modern alternative to the Git version control system.2:16 His new job?2:17 To build the next generation of version control specifically for AI-driven development.2:22 It’s a signal that the fundamental tools of our trade are about to go through a major overhaul.2:28 But not all the news is about building things.2:31 Sometimes it's about things falling apart.2:34 The FBI is now investigating a dark web service called Nexus, which is actively selling digital scans of over 153 million U.S.2:42 and Canadian driver’s licenses.2:44 The data, they claim, was siphoned continuously for over a year from a major identity verification company.2:50 This is a catastrophic software supply chain failure, with consequences that are almost too big to grasp.2:57 And finally, a little dose of reality.2:59 A popular analysis is making the rounds, looking back at some of the more...3:04 pessimistic AI predictions from a couple of years ago.3:08 Specifically, AI skeptic Ed Zitron’s 2024 claim that companies like Meta and Google were "dying products" and "dying companies." The analysis contrasts this with their financial reports through 2026.3:20 Meta’s revenue jumped from 135 billion to 201 billion.3:24 Google’s went from 307 billion to 403 billion.3:26 Turns out, the AI-fueled giants are not, in fact, dying.3:30 They're thriving.3:31 It's a useful reminder to check the receipts on bold predictions.3:35 So what does it all add up to?3:37 You've got this three-way tug of war.3:39 Developers are building their own private AI stacks on local machines.3:44 Cloud companies are slashing prices and adding privacy features to keep them.3:49 And the apps we use are quietly shipping entire office suites to give us the best of both worlds.3:55 This isn't just about features anymore.3:57 This is a fight for the soul of the developer workflow.4:01 Okay, let's dive deeper into that central tension, because it’s the most important pattern shaping software right now.4:08 It's the battle between local control and cloud convenience.4:12 Let’s start with the local side.4:14 A developer going by raybb posted their entire setup, and it’s a masterclass in digital sovereignty.4:20 They took an M4 Pro Mac Mini with 48 gigs of RAM — a powerful machine, but not some unobtainable supercomputer — and turned it into a personal AI server.4:30 They're running multiple models.4:32 A big one, Qwen3.6-35B, for deep reasoning, and smaller, faster models for everyday tasks.4:38 They’ve wired it all together with Tailscale so they can access their private AI from their phone, their laptop, anywhere, without touching a public API.4:47 Why go to all this trouble?4:49 The quote that stood out to me, the one that really nails the sentiment on Hacker News, is this: “The main reason to run local: cloud APIs are rented land.” Think about what that means.5:01 When you use a cloud API, you are completely at the mercy of the provider.5:05 They can change the pricing on a whim.5:08 They can impose usage limits that break your application.5:11 They can, and do, swap the actual model behind the API endpoint without telling you, leading to subtle shifts in performance and output that are maddening to debug.5:22 And that’s before you even get to the privacy concerns of sending all your data to a third party, or the content filters and government restrictions that can block legitimate use cases.5:34 Running your own model on your own hardware solves all of that.5:37 The cost is a one-time hardware purchase, not a variable monthly bill.5:42 Your data never leaves your machine.5:44 The model is the model — it doesn't change unless YOU change it.5:48 It’s predictable.5:49 It's controllable.5:50 It's yours.5:51 Now, where have we seen this pattern before?5:54 This is the mainframe versus the personal computer, all over again.5:58 In the early days, computing was a centralized resource.6:02 You rented time on a massive, expensive machine owned by a corporation.6:06 The PC revolution was about putting that power, or a version of it, on every desk.6:11 It decentralized innovation.6:13 It created entire new industries.6:15 People wanted control over their own tools, their own data.6:19 The analogy is powerful.6:20 The cloud today feels a lot like the mainframe era.6:24 A few giant companies — Amazon, Google, Microsoft, Anthropic — own the "means of computation," and we all rent access.6:31 The local LLM movement is a direct echo of the Homebrew Computer Club, of people wanting to build and own their own tools.6:39 But here’s where the analogy breaks down, and you have to be honest about this.6:44 Steve Wozniak could build an entire Apple I in his garage.6:48 You cannot train a 35-billion-parameter model on a Mac Mini.6:51 The local setup we're talking about is for inference — for running the model.6:56 The training still happens in data centers the size of small cities, consuming megawatts of power.7:02 So this isn't true independence.7:04 It's more like downloading a massive, incredibly complex piece of software that was created in the cloud, and then running it locally.7:13 You’re still dependent on the cloud giants for the initial creation of the artifact.7:18 It’s a step toward sovereignty, but it’s not a declaration of independence.7:23 And the cloud providers are not standing still.7:26 They see this desire for control, this frustration with "rented land," and they are responding.7:32 Look at Anthropic's announcement for Claude Fable and Mythos 5.1.7:36 It reads like a direct answer to the local LLM movement's complaints.7:40 Complaint one: It's too expensive.7:42 Anthropic's response: Fable 5.1 is twenty-five percent cheaper for typical workloads, and up to forty-five percent cheaper for the complex, multi-step agentic work that really runs up a bill.7:55 They are aggressively competing on price to make the cloud option more palatable.8:00 Complaint two: I don't want you to have my data.8:03 Anthropic's response: Enterprise Frontier Safeguards.8:06 This is a big deal.8:07 It lets enterprise customers store and manage the data that informs the AI’s guardrails within their own cloud infrastructure.8:15 They're also offering zero data retention options.8:18 They are trying to give you the privacy benefits of a local model, but with the scale and power of the cloud.8:25 They want to make the rented land feel so luxurious and secure that you won't even think about homesteading.8:32 So you have this fascinating dynamic.8:34 The grassroots movement toward local models creates pressure.8:38 The cloud providers respond by making their offerings cheaper and more private.8:43 The end result is that you, the developer, get better options all around.8:48 The competition is working.8:50 Now let's talk about the other side of this coin.8:53 The hidden infrastructure.8:54 Because while the local-versus-cloud debate is happening, a quieter, arguably more important shift is taking place in the very structure of our software and tools.9:05 This brings us back to that one-point-seven gigabyte ChatGPT app.9:09 Simon Willison's discovery that it bundles a full copy of LibreOffice is the perfect example.9:15 Why on earth would an AI chat app need an entire office productivity suite?9:19 The answer is, it's a brilliant, if brute-force, solution to the privacy problem.9:24 Let's say you want to ask ChatGPT a question about a complex Word document or an Excel spreadsheet.9:31 In a web-based world, you have to upload that document.9:34 Your potentially sensitive, proprietary data leaves your machine and gets processed on OpenAI's servers.9:41 A lot of people, and a lot of companies, are not okay with that.9:45 By bundling LibreOffice, the ChatGPT app can do something different.9:49 It can use the local, sandboxed copy of LibreOffice to open and parse your .docx or .xlsx file right there on your machine.9:57 It extracts the text and relevant data locally, and only then does it send the necessary information to the AI model.10:04 Your file never leaves your computer.10:07 This is a return to the idea of a "fat client." For years, the trend in software was toward "thin clients" — lightweight applications, mostly web browsers, that did very little processing themselves and relied on a powerful server.10:21 It was efficient.10:22 It saved disk space.10:24 But in the age of AI and privacy concerns, the pendulum is swinging back.10:28 That 1.7 GB isn't just bloat; it's a privacy feature.10:32 It's a deliberate architectural choice to do more work on the edge, on your device, to keep your data safe.10:38 It’s OpenAI building its own version of a "local first" experience, not by asking you to set up a server, but by hiding one inside the app.10:47 And this theme of re-architecting fundamental layers goes even deeper.10:52 Think about the announcement from ERSC hiring Martin von Zweigbergk.10:56 This is not your typical tech news.10:58 It's about the plumbing.11:00 Version control.11:01 Git.11:01 Git was designed by Linus Torvalds to manage the source code of the Linux kernel — a massive project, but one written by humans.11:09 It's optimized for text files and for a workflow where humans review changes.11:14 Now, fast forward to 2026.11:15 AI is writing code.11:17 It's generating not just lines of code, but entire functions, files, and projects.11:22 And a lot of "AI development" isn't even about code; it's about managing huge datasets, massive model weights, and complex prompt chains.11:30 Trying to manage this new world with a tool designed for human-written text files is like trying to haul freight with a sports car.11:39 You can do it, but it's not what it was built for, and it's going to struggle.11:44 Martin's quote from the announcement is so telling.11:47 He says, "Jujutsu improves the part of version control that sits on your laptop.11:52 But the remote server is still Git, which has a ceiling that comes fast for products at scale.11:58 We think the storage layer has to change to match the model." That's it, right there.12:03 The tools we take for granted, the absolute bedrock of modern software development like git push, are hitting a ceiling.12:11 The problem isn't just how you manage diffs on your laptop; it's how the central server, the "source of truth," stores and handles these new kinds of digital assets.12:21 ERSC is betting that the future requires a fundamentally new kind of remote storage layer for version control, one built from the ground up for a world of AI-generated code and massive model files.12:34 This is foundational work.12:35 It's like someone in 1950 realizing that as buildings get taller, you can't just keep using bricks; you need to invent steel-frame construction.12:44 And when you put these two stories together — the fat ChatGPT app and the quest for a post-Git version control — you see the same pattern.12:53 The explosion in AI capabilities at the top layer is forcing a radical rethinking of the hidden infrastructure underneath.13:01 The app on your desktop and the commands you type in your terminal are being quietly re-engineered for this new reality.13:08 So you step back from the week, and the picture that emerges is one of growing sophistication.13:14 We're moving past the initial shock and awe of "wow, a chatbot can write a poem." We're now in the messy, practical, engineering phase.13:23 It's a phase defined by a struggle for control, playing out in the choice between a local Mac Mini and a cloud API.13:30 It's a phase of architectural trade-offs, where an app intentionally becomes "bloated" to deliver privacy.13:37 And it’s a phase of foundational change, where the very tools we use to build software are being questioned and rebuilt.13:44 And through it all, we have the stark reminder from the FBI and the Nexus breach.13:50 153 million driver's licenses for sale on the dark web.13:53 It's a sobering anchor to all this technological progress.13:57 While we are busy optimizing our AI workflows and debating the merits of local versus cloud, the fundamental challenges of digital security and identity are not only unsolved, they're escalating.14:09 The same interconnectedness that allows you to access your private AI from your phone also creates vectors for attack.14:17 The same data-processing power that lets an AI analyze a document can be used by criminals to exploit a database of stolen identities at an unprecedented scale.14:27 This week shows that the future of AI isn't just about building more powerful models.14:32 It's about the painstaking work of building the ecosystem around them — the tools, the platforms, the security, the privacy.14:40 The most important shifts aren't always the ones that get the big headlines.14:45 Sometimes, the future arrives in a 1.7 gigabyte cache folder.
