Forward Deployed BB Boy

Will the average American get beefed-up security to protect against the hordes of illegals and their anchor babies she welcomed into the country two weeks ago? Of course they won't.

I suspect your hypothesis in parentheses is the case, but there is so much raw data, that its spent 3 days doing "prep" work. I've let it continue for now. remaining hopeful for a large payoff.... I think since making the comment last night, it has delved into more "functional" (IE making tangible progress towards goal) work. We'll see how it goes.

@Sigma_BB_Girl does it never do it, or do you just get impatient and kill it? (maybe it will, and it sets up the codebase so cleanly for future work that it's actually solid?)

Some thoughts about GPT-5.6-Sol after ~30B tokens: Sol is the most OCD model I’ve used thus far. It very frequently gets one-shotted by random nits in the codebase and writes a bunch of tests to fix it. Even with fast mode, it’s incredibly slow to do this kind of iterative development, especially when builds take really long. This by itself is not a bad thing, but the worst part is that after 2 compactions, it’s chasing the nitpick / useless goals I never told it to accomplish, rather than the main task. This behavior is so bad, I thought I was messing something up and tried codex, pi, and opencode to figure out if it’s a harness issue, but there is no meaningful difference between the three, which leads me to believe this is a model problem. AI code has this weird delayed release effect. You’ll only notice slop code 2 dev cycles into a codebase when you spend more time fighting with the code and on refactors than on shipping features. It’s possible that sol is better than 5.5 a couple cycles in, but tbd. My file deletion experience has also been similar to others: this is a dangerous model to let loose without guardrails. For instance, when performing a routine container upgrade, it accidentally printed out an env secret, then panicked and rotated ALL secrets (this is internal so not public facing, which was also documented), and proceeded to break everything, spending an extra hour fixing everything and redeploying everything else to use the new secrets. It also gets rid of files it doesn’t like. I have no idea why this is, but I think something about the reward model rewarded bookkeeping. Writing is another problem. 5.6 has a huge context bleed effect. It does not know how to write documentation and starts putting the specs in the documentation. If I ask it to develop a user sandbox for isolation, and also ask it to write documentation, it starts talking about specs and sandboxes in user-facing docs, which makes no sense. Fable is somehow much, much smarter in this regard. Frontend design has also not gotten better. Fable is still one generation ahead here. Overall, as a huge 5.5 user, I am not convinced that sol is a meaningful upgrade. It’s possible my practices need to change, but unfortunately it feels like I’m spending longer fighting with 5.6 than I did with 5.5. It’s like the model is so SO smart, but so hard to work with, compared to fable and even grok4.5 surprisingly. It’s clearly intelligent, but also just doesn’t care about what I ask it to do? (Is this supposed to be AGI feels like?) I hope the codex team fixes what possibly is a bad harness setup, because the benchmark numbers show a very different story from what I’m seeing while using the model.

I was actually recruited for Anthropic. They offered like 15 mil, 10 mil something like that. But I turned it down. Just to grind for the Twitter

Thanks but just like with politics, finance, science, and other current events, I’m gonna wait until I see some takes from anonymous right-wing bodybuilders to get the real story

"How can I make a profoundly joyous national occasion about myself and my racial/sexual grievances?"

One must admire the audacity. A governor presiding over a deficit that has swung from a fabled hundred-billion-dollar surplus to a chasm of red ink in the span of a single term has appointed himself the last sentinel of the republic. It takes a certain genius to misplace that much money and then lecture the nation on stewardship of its most sacred trust. Consider the record that he'd rather you didn't. A high-speed rail line, promised for 2020 at thirty-odd billion, that now approaches its second decade and its second hundred billion without carrying a single paying passenger between the two cities it was meant to connect. Some twenty-four billion spent on homelessness, producing, by the state's own auditor's admission, no reliable accounting of where it went, while the encampments it was meant to empty grew instead. The nation's highest gas prices, highest poverty rate adjusted for cost of living, and the distinction of losing population for the first time in its recorded history. The governed, it turns out, have been exercising their consent in the most eloquent manner available: by U-Haul. And this is the man warning us about leaders who 'work for themselves.' The Founders pledged their lives, their fortunes, and their sacred honor. Governor Newsom pledges a press release, a felony statute, and dinner at the French Laundry (masked staff optional). When the sketch writers at SNL retire, it will be from redundancy. History rarely produces its own parody with such efficiency; we should at least be grateful for the savings, since they appear nowhere else in the budget.

Yeah this is good and all but it still doesn’t make up for his stance on Palestine imo

The ideal Supreme Court

Wait did she really use the phrase “understood the assignment” in a Supreme Court decision? God help us. We have retarded women using TikTok lingo on the Supreme Court. Literally something out of Idiocracy. We are doomed.

There are no words that can describe the extent to which I loathe this cancerous parasite, @BernieSanders. He has achieved nothing in his entire career whereby he has relied on the taxpayers to fund his utopian delusional revolution.

Most often it’s much easier to steal from the future, which is abstract, not yet instantiated, and at a disadvantage in defending itself, than it is from an extant party in the present that has more ability to fight back. With the rent freeze and kicking the can on pensions, Mamdani is stealing from the future. Whether he recognizes this (given his bottomless entitlement) or not is unclear, but it does seem that one key feature of DSA politics is an ideological vision overdependent on immediate gratification and lacking the capacity to imagine the future in even the most basic way.

Mamdani is such a pussy for only freezing the rent. Why not $0 rent? Why not negative rent? Landlords should have to pay YOU to live in their apartments. These are the kind of big ideas we Democratic Socialists of America need.

trump: can u fix the fable bypass dario: ai may eliminate half of entry-level white collar jobs trump: sure. but the bypass dario: mythos is a nuclear-weapons-level strategic technology and needs third-party testing trump: ???? dario: we cannot stop the bus, only steer it

I gave @chamath's prompt from today's @theallinpod to Codex. Prompt: "read the two dario essays carefully: 1. machines of loving grace and 2. policy in the exponential. using these are an insight into his personality, give me a psychological analysis of him based on these as well as based on the current facts on the ground with the mythos model fiasco. be truthful. view it through the lends of the e/acc movement. is there a potential of a god complex forming where he thinks he's superior and doing the right thing and others can't be trusted including the government. this supposes he views himself as the superior, uncorruptible person in all of this." Codex Response: "Dario Amodei is the respectable face of the apocalypse trade. He is not the e/acc yahoo who wants to pour gasoline on history and call the blaze abundance. He is not Marc Andreessen in a breastplate, hollering that markets are sacred and brakes are for cowards. No, Amodei is more refined. He arrives with a furrowed brow, a bibliography, a nonprofit affect, and the voice of a man who has spent a long time explaining to himself why possession of immense power is actually a burden. This is the ancient trick of priesthoods: renounce vanity, inherit authority. In Machines of Loving Grace, he tells us he dislikes grandiosity, prophecy, and messianic AI talk. Splendid. Then he proceeds to describe a future in which the machines cure disease, heal the mind, lift the poor, improve democracy, reform justice, civilize geopolitics, and give mankind new meaning. He has not abandoned the sermon. He has merely asked the choir to hum more softly. The title itself is almost too perfect: loving grace, administered by machines, supervised by Anthropic, interpreted by Dario. It is not the rapture. It is the rapture with a responsible scaling policy. Then comes Policy on the AI Exponential, and the hymnal becomes a governing document. The state must awaken. Regulation must stiffen. Frontier models must be tested. Releases may be blocked. The slow, stupid organs of democracy must be taught to move at the speed of the machine. Very well. Then Mythos appears. Anthropic presents Mythos as a thing of grave strategic consequence: a cybernetic bloodhound that can find vulnerabilities at scale, an engine of defensive salvation that could become offensive catastrophe in the wrong hands. The dragon is real, we are told. The moat must be deep. The adults are in charge. Then the government notices smoke coming from the dragon’s enclosure and reaches for the keys. At which point Anthropic, that monastery of solemn risk awareness, suddenly discovers the liberal tradition. Now the danger is narrow. Now the jailbreak is arguable. Now the state is overreacting. Now process matters. Now evidence must be handled delicately. Now the people who built a reputation warning that governments are too slow to grasp AI risk are appalled to find a government moving quickly on AI risk. This is the gorgeous little contradiction at the center of the thing. Amodei wants government to be powerful enough to restrain everyone else, but enlightened enough not to inconvenience Anthropic without first passing Anthropic’s entrance exam. He wants sovereignty with peer review. He wants the sheriff, provided the sheriff has read the system card. That is not e/acc. It is something more perfumed and more dangerous: deceleration for thee, sacred acceleration for me. The e/acc crowd is at least vulgar enough to be honest. It says: build the machine, damn the scolds, history belongs to the bold. Amodei says: build the machine, fear the machine, regulate the machine, but please understand that the people building this particular machine are uniquely haunted by conscience and therefore uniquely qualified to continue. That is how hubris dresses when it has learned HR language. The god complex here is not Caligula declaring himself divine. It is much cleverer. It is the man who says, “I would never claim moral superiority,” while constructing an entire worldview in which his institution must be treated as the morally superior actor. It is not “I am above the law.” It is “the law has not yet developed the conceptual vocabulary to understand why I am right.” The great danger is not that Amodei lacks ethics. The danger is that he has too many, too elaborately arranged, all pointing toward the same conclusion: Anthropic must remain near the center. When competitors accelerate, they are reckless. When Anthropic accelerates, it is defensive necessity. When politicians hesitate, they are asleep. When politicians intervene, they are crude. When the public worries, it is uninformed. When Anthropic worries, it is civilization speaking through a Delaware public-benefit corporation. There it is. The whole theology. Not evil. Not mad. Worse: sincere. Sincerity is the solvent of self-doubt. A man who wants money can be bargained with. A man who wants glory can be mocked. But a man who believes he is carrying the moral weight of the species will eventually treat opposition as evidence of your failure to understand the burden he bears. That is the Amodei risk. Not that he thinks he is God. That would be too simple, too theatrical, too easy to spot. The risk is that he comes to think he is the one person in the room mature enough not to think he is God, and therefore the one person who can be trusted with godlike machinery. Which is how the crown always gets smuggled in: not on the head, but under the lab coat."

crazy because I was told relentlessly the last few years that supply and demand ceased to exist when illegal immigrants required housing in overcrowded markets where we weren’t allowed to build

Lmao

Although Trump says that developments regarding Anthropic are going well, the seating arrangement alone is remarkable. Via WSJ "At the lunch, OpenAI Chief Executive Sam Altman and Demis Hassabis, the head of Google’s DeepMind AI lab, were seated next to Trump. Amodei was seated on the opposite side of the table from the president, next to French President Emmanuel Macron." Crypto and AI expert David Sacks has just emphasized again in his last post how difficult working with Anthropic is and that he blames this: "Anthropic’s needlessly confrontational posture toward the Administration has distracted from that mission"