sh34r

This is the pseudo-intellectual version of your boomer relative ranting about the weatherman on TV, “how does this guy still have a job when he’s always wrong!”

Or the poll-bashing that happened after the 2016 election when an outcome with a 29% chance per one model was observed.

Any interesting forecasting problem is underconstrained. Monte Carlo methods are the classic way to mitigate that one-universe problem.

Nobody can predict the future with certainty. All models are wrong, but some are useful.

The last part is the only one that concerns me, but only ever so slightly… you realize the future models also get to learn from videos of ISIS executions, LiveLeak offshoots, factory farm footage, and Russian war crimes in Ukraine, right? They already have more than sufficient evidence of humanity’s capacity for sadism. Making a computer program output some distressing text isn’t gonna move the needle for the Basilisk... These rationalist cultists really gotta go touch grass, man.

It was a great plan, but all it took to foil it was one orange man in the Epstein files.

The problem you’re gonna run into is that you’re essentially asking it to use passive voice. Everyone with a high school education has been taught not to do that in English class. The vast majority of the prose in the training set was not written by functionally illiterate people.

You have to add Linux capabilities to support creating nested namespaces. IIRC SYS_ADMIN is required. If you do that, what do you think you’re achieving with the container? It defeats the entire purpose IMO, just makes development more inconvenient.

If you need the sandboxing capabilities provided by the application and don’t want to run it on the host, you should set up a VM, not a container.

From a systems perspective, it’s foolish to rely on a proprietary application to provide its own isolation guarantees. This is an OS-layer responsibility.

ML/AI experts suffer badly from hubris. They are brilliant in one domain. It transfers poorly to systems engineering. The proof is in the FelonyBench results.

The most remarkable thing is that it was built at all. What self-respecting Linux desktop user wouldn’t be satisfied with the TUI?

It is a marketing expense. We are talking about it. Surely the token cost was worthwhile on that basis alone. Students will likely get some usage out of it, too. But paying a human to do this port would have been an insane thing to do two years ago. I’d be surprised if they have more than 1000 MAUs for this client by the end of the year.

gpt 4 moto razr wen?

Man, we’ve come a long way from the halcyon days where the HBO Silicon Valley gag about coming out as Christian was genuinely hilarious. Time to shove the Bible thumpers back in the closet.

I bought an Intel MBP towards the end. No regrets, I had valid reasons.

But the MacOS support at this point is pretty atrocious. I mean, it technically runs, but the OS is clearly designed in a way that ruins performance on this hardware. It’ll run Vim, sure, for 2 hours on battery, hot as hell. At this point, Intel Macs are only fine for running Linux.

IMO they don’t need another law for this. Fraud is already illegal. They need to make a very public example by indicting a few HR ghouls under wire fraud statutes for their fraudulent job listings.

Every hyperscaler hosting these models outside of FEDRAMP environments has been compromised by every regional power’s intelligence services. Fable was running all over the world until today.

AWS and friends are very good at providing excellent enterprise grade security, but it’s literal child’s play for nation state threat actors to exfil these models.

TEMPEST / EMSEC alone is a wide open door for unclassified datacenters when the Mossad’s out to get you.

They should have just called it Opus 5.1 and released it like normal. All this fanfare, under this corrupt regime, after they declared you a supply chain risk… Wario has horrendously bad judgment.

Good thing these corrupt gerontocrats are also all in on cryptocurrency then.

Sometimes, the enemy of my enemy is my friend.

Yes, reducing your TAM by roughly 8 billion people is exactly the kind of marketing you want before your IPO. /s

Given how often these get leaked (see Palantir + SpaceX) and the cost of preparation, why would you ever file an S-1 unless you were serious?

I don’t believe this is true of modern thorium reactors.

Coal ash is more radioactive than nuclear waste, too.

I’m not an expert here, but my understanding is that coal-free steel production is not a solved problem yet. And no, importing Chinese steel and moving the problem elsewhere isn’t a reason to pat yourself on the back.

There is absolutely no good reason to burn coal for electricity or heat in this day and age. If we had sane global leadership, every coal power plant left would be treated as a WMD and be bombed harder than that Iranian fuel depot.

I have been a skeptic of this until now. The explanation given by the researcher interviewed seems more than plausible to me.

It’s not the typical misunderstanding of non-ionizing radiation. The variable symptoms make a lot of sense, given that the weapon is basically just causing random electrical “failures” in the body. This was not a precision op. They saturated a location with this engineered interference signal, with the goal of maiming the target. No regard to whether their families and children would be collateral damage. It’s a war crime on multiple levels, on our soil. Then we presumably went and did the same thing during the Maduro raid at scale.

Just what we needed in 2026, more man-made horrors beyond our comprehension.

To me the question is actually, what changed to make them release the story now? Biden’s been out of office for a while now… it wasn’t anything his admin did. They could’ve continued gagging the victims, claiming it’s psychosomatic, and most of us would keep on believing that, because Occam’s razor.

Lots of similar reports came out during the Maduro raid. Same symptoms. Seems we demonstrated the capabilities we were hiding. OSINT experts already put the pieces together a month ago. So did our adversaries. Cat’s out of the bag, so no sense continuing to gaslight our wounded veterans.

We probably put this fucking thing in a plane instead of a backpack. Everything’s bigger in the USA, of course.

Your workflow makes sense for FOSS projects, where the commit is the unit of work. In my experience, on most professional teams, the PR is the unit of work. PRs trigger CI/CD pipelines. PRs map to tickets. The meaningful commit goes with the squash merge to the shared dev/main branch.

There are cases where I've staged commits this way for a PR, to make it more reviewable. I'd usually rather split them off into separate PRs, but when that would create a pipeline of three MRs that are meaningless on their own, then rewriting history for a single MR makes sense. I generally consider my feature branch's commit history to be for me, not for you. Going back and rewriting history is a chore that shouldn't be necessary if I did a decent enough job with the PR description and task decomposition. Those commits are getting squashed anyway. Along with all the "fix MR comments" commits on top of it.

It wouldn't bother me to adopt your workflow if it fits your team and its tools and processes. I'd just say, consider that your way isn't the only correct way of doing things. Your preferences are valid, but so are others'. The only thing that really bothers me is absolutism. "My way or the highway."

Your writing here reminded me of a particularly unpleasant coworker I had in the past. I quickly browsed your comment history to make sure you're not him... Excessive rigidity is not an endearing quality.

All that being said, I have also been constantly annoyed by people with too many YoE who can't be bothered to spend an hour or three to learn the basics of how the Git tree is structured, and what merge vs rebase does. They rely too heavily on their GUI crutches and can't fix anything once it goes sideways. Even when you lead them to water, sending them reading material and offering to answer questions after, they refuse to drink. Willful ignorance is far more irritating than stubbornness. I don't expect them to be able to remember what bisect vs cherry-pick does. Claude will spit out the subcommands for them if they can describe what they need in English. But they can't do that if they have no understanding of the underlying data structures...

I guess the AI companies finally figured out they’re supposed to buy their stolen datasets from a shell company spun up by the most unsavory character within two degrees of the CEO. Every CEO has a drug dealer, and every CEO drug dealer knows the greasy grey hat dude running a data laundry “startup.” The VCs usually know some private equity dons who run the same racket to do bust out fraud, too.

It’s truly unbelievable that OpenAI and Anthropic were so sloppy. Pirating all that copyrighted media and not even bothering to hide behind one layer of indirection. Amateurs.

So yeah… it’s what, five years’ worth of pent up demand for organized crime, hitting the market everywhere all at once? I’m surprised the request volume isn’t higher!

I take it you don’t know much about the Troubles, then. The SAM missiles would be saved for returning ICE Air flights, not Delta.

FAA ought to be drowning Kegseth’s DoD in bureaucracy at every possible opportunity, after the massacre over the Potomac River a year ago. They deserve no leniency whatsoever.

Why does everything have to be in the TUI? I like the TUI. But I also want all the logs. And I do mean all of them.

Of course all the logs can’t be streamed to a terminal. Why would they need to be? Every logging system out there allows multiple stream handlers with different configurations.

Do whatever reasonable defaults you think make sense for the TUI (with some basic configuration). But then I should also be able to give Claude-code a file descriptor and a different set of config optios, and you can stream all the logs there. Then I can vibe-code whatever view filter I want on top of that, or heck, have a SLM sub-agent filter it all for me.

I could do this myself with some proxy / packet capture nonsense, but then you’d just move fast and break my things again.

I’m also constantly frustrated by the fancier models making wrong assumptions in brownfield projects and creating a big mess instead of asking me follow-up questions. Opus is like the world’s shittiest intern… I think a lot of that is upstream of you, but certainly not all of it. There could be a config option to vary the system prompt to encourage more elicitation.

I love the product you’ve built, so all due respect there, but I also know the stench of enshittification when I smell it. You’re programmers, you know how logging is supposed to work. You know MCP has provided a lot of these basic primitives and they’re deliberately absent from claude code. We’ve all seen a product get ratfucked internally by a product manager who copied the playbook of how Prabhakar Raghavan ruined google search.

The open source community is behind at the moment, but they’ll catch up fast. Open always beats closed in the long run. Just look at OpenAI’s fall into disgrace.

> Curious what is in the rest of it

Probably a lot of CSAM, if the Mossad blackmail op theory of Epstein is true.

If it is, do you think it’s the Iranians taking revenge on American civilian scientists, or a Ted Kaczynski type?

I want a BYD that costs less than a 2000 Camry did brand new in 2000.

EVs are inherently pretty simple machines. All the complexity is in the battery, and China’s crushing everyone at battery tech. It’s not even close. It’s like a human trying to beat a polar bear in hand to hand combat.

They really need to deregulate the auto industry and let us buy the Yugos with a Jetsons battery. America is a poor country now. Nobody can afford used cars in this economy, never mind new ones.

I mean, what do we expect from this brainless company that promised a $20k Maverick and let the stealerships mark it up to over $40k?