however because everything in claude code is implemented in 5 million different ways, those protections are a completely orthogonal set of codepaths from how CLAUDE.md files are read. conversely, the file read tool seems to be completely naive to symlinks while the CLAUDE.md reader is not

See, if a person had written the system and had some kind of reasonable design or whatever then this kind of vulnerability may have turned into a full skeleton key to crack the entire system wide open. Instead, the superior machine intelligence ensures that all components will break in slightly different ways under slightly different conditions, thus ensuring no single fault can allow an adversary to completely compromise the system, provided of course that they can’t just inject a prompt somewhere because as we all know that’s a structural vulnerability that can completely crack the system open.

The goal is for this process to take no more than 60 seconds.

You go into the water, you come out of the water, and you’re done.

When you step into the water, you’re standing on top of a platform. The platform is connected to rails and begins to descend into the water - an elevator gently lowering you at around 2 inches, or 5 centimeters, per second.

Assuming 30s down and 30s up, this platform will lower you a total of 60 inches, or 5 feet. So even in the magical fantasy land where this works at all you’re going to have a great tool to detect what’s happening in the body as long as it’s below the average person’s heart, an organ that is famously unnecessary and irrelevant.

Gay people I respect:

I can pretend they’re straight

Gay people I don’t:

I can’t pretend they aren’t gay and I’ve decided this is a ‘them problem’

This feels like yet another case of “what no postmodernism does to a mf”. Because I can kind of agree they ASD is a bit of a weird diagnosis in some respects, but that’s entirely because of an ongoing discussion between different social models of mental diversity. On one hand, much of the autistic community has embraced a model rooted in neurodiversity. The associated behaviors aren’t “abnormal” and happen to nearly everyone to some degree, but some people experience them in different ways that impact how they interact with other people and the world. In this model, we need to emphasize empathy and support to help everyone find a place in the world where they can be happy and fulfilled.

But Scott as a practicing psychiatrist is strongly invested in the older model of mental illness or mental disability, where some people have mental and behavioral problems that make life more difficult and dangerous for them and those around them, and those people need help to mitigate those problems so they can exist in society. This is obviously a less kind and more authoritarian model than neurodiversity, but it’s easy enough to understand the appeal in circumstances where the challenge of “how do I help people be happy and successful” necessitates asking “how do I reduce or eliminate the risk of physical violence from this person.” Hell, compared to the other model that gets invoked to deal with that question, criminality, it still at least acknowledges that people with high support needs are people whose well-being deserves consideration. A patient is a human being at least to the same degree that a child is, effectively.

But Scott, being afraid of postmodernism, can’t really allow himself to recognize these as separate models that are valuable in different circumstances, not the least of which is because he’s a fashy little bastard who would have to give up the authority of being a capital-D Doctor in favor of just being an expert on certain elements of the human mind and body. So instead he and his friends are totally neurotypical, yes. Don’t listen to them describe their own experiences, listen to the Doctor Man.

As I recall the impetus for this was that the team working on grok’s code generation ability was preferring to use cursor. Imagine making a model so shitty that even the kind of ai-brained buffoons who want to work on fucking grok don’t want to use it. Now consider that apparently that model is grok itself.

3. Shit piles up at the bottom of the hill

4. Bottom of the hill elevates until it reaches the enshittifiers

5. Dinosaur eats man

6. Woman inherits the earth

I’m sorry are you telling me that there people’s social circle includes the same like 50 oligarchs and apparatchiks and they’re making a dating site for themselves? I know they lost Epstein but this is just pathetic.

How is it that whenever I learn something new about this family it’s always a new way for them to be awful to each other or their kids?

Well yeah. Just because they call their crypto scheme an Oracle doesn’t mean they’ve actually solved the oracle problem.

Yeah. I’m not gonna spring for the paid subscription to read the whole thing, but if we’re gonna be angry at reasonable general-purpose life advice being melded to a toxic and hateful and then put forward as a revolution in human thought and relationships, then I have bad news about the whole rationalist community.

Maybe I’ve reached a unique and terrifying level of terminally online goblinhood, but I wonder if part of the issue is also that people are already using basically 100% of the apps they care to use. Like, I don’t think there’s much in terms of entertainment, productivity, social engagement, or whatever that can be replaced or augmented by new apps, so depending on how we’re measuring total usage we shouldn’t expect to see dramatic increases just because there are more apps. This doesn’t change the fact that vibe coding isn’t actually making things people want to use, and in fact I would say that this dramatically increases the chance that any slopware that does do something legitimately useful is going to quickly find itself competing with more well-designed and well-constructed versions of itself that was created by an actual mind.

According to a later post from them there’s a chance this is more of a list of attendees at a past or upcoming event, but tbh I don’t think there’s much of a difference between “member of the nu-money illuminati” and “accepted an invite to hang out with the nu-money illuminati at their clubhouse.” I can only assume it has the same statue of a hand clutching a globe as the Deus Ex opening cinematic.

Anthropic: Oh no! Our new model is too powerful! It’s dangerously good!

US Government: okay then you can’t export it or allow foreign nationals to access it

Anthropic: Wait not like that

We have reviewed a report that we believe is the basis of the government’s directive and validated that the level of capability displayed there is widely available from other models (including OpenAI’s GPT-5.5), and is used every day by the defenders who keep systems safe.

Of course this has less to do with the actual capabilities of any model and more to do with Anthropic openly telling the Trump administration “no” on exactly one occasion, but we can still roast some marshmallows over this dumpster fire, right?

I also question the degree to which “the other 99% of readers” actually don’t care about AI slop. Even outside the awful bubble here I see AI images get met with at best a weary sigh of “I guess this is the world now” rather than actual acceptance. And I know we’ve talked at some length about how gell-mann amnesia isn’t a very useful model, but I think most people are much less tolerant of slop in areas that the know more about. Maybe I can’t tell an slopware history paper from a real one, but historians certainly can, and they hate that shit. In that sense the “median reader” statistic is misleading because the median reader isn’t particularly invested in most of what gets written anyways, and it’s the lack of investment that makes slop seem plausible. Even the concerns about deep fakes seem like they’re not separate from this general problem, since the same disconnection that makes people uncritically accept deep fakes makes them uncritically accept fake news without an accompanying video, or with a context-stripped video from an unrelated event, or whatever.

This is fascinating to me and has shades of Iris Merideth’s The Problem is Culture. In other engineering fields, if you had a tool that cut costs but caused a threefold increase in failures you would be looking at a massive scandal, probably because if this was structural engineering rather than software engineering you’d be looking at a new Grenfell Tower or Hyatt Regency Walkway from every other project that used this shit. From what I’ve been following I don’t know that vibe coding has directly racked up quite so literal a body count yet, but if this pattern holds (and I see no reason to expect otherwise) then it’s only a matter of time before someone fucks up something important.

Also the fact that the framing here doesn’t seem to treat this as an existential risk to the project of AI coding is fascinating. If you’re not producing stable and secure applications in prod then what in the actual fuck are you writing all that code for?

I don’t know if I’m more impressed by how many occasions you find to link that or by how often it’s relevant. Either way, you’re doing the Lord’s work and congratulations on the 1/10000 of a Hugo.

My favorite part of that whole beautiful affair was towards the end with dgerard’s Chinese counterpart saying “no don’t ban them yet I need more material for an article I’m halfway through writing.”

In D’Souza’s interview with the Australian newspaper, he explained why: “It’s only the top 1 percent who matter. These are the people who are going to be the value creators” when, in his view, AI soon completely transforms just about every aspect of economic life.

D’Souza continued, “Ultimately, what’s the last job? It won’t be knowledge work. It won’t be physical work. It will be interfacing between the physical and the digital worlds, and right now that frontier is journalism.”

Taken together it becomes incredibly transparent that the actual goal here is to transform themselves into a kind of priest-king class, exercising absolute authority on behalf of the remote and unfathomable god that they built. Just please pay no attention to who built the AI, who runs the AI, or where all the money and power end up.

I have some modest proposals for handling private equity but they would all probably count as fedposting. We still haven’t found a decent replacement for the market niche JoAnn fabrics occupied.

I love the deep lack of specificity in “other sectors of the economy facing shortages”. Either you have to acknowledge that you’re talking about the cafe economy and gig economy and those sectors aren’t so much facing labor shortages as much as leveraging the worker’s chronic underemployment to keep costs down or you’re making shit up wholesale. Also please note that American companies are already finding that as the investor capital subsidies run out it’s often cheaper to hire a person than pay the token costs to do the job with AI.