{stc} retweeted
Great example of why we all don't have high IQs; selection operates on more than just individual intelligence.
Humans that survived moderate stresses were those that prioritized fairness & cohesiveness, not just profit maximization.
Feeding your neighbor was more important than money.
I once talked to someone at a party in college that made one brilliant point after another at the intersection of philosophy of mind and economics.
He was crystal clear in his thinking and he had an elaborate (but not overfit as far as I can tell) model of how society was not made of individuals, but rather of attention streams. Rather than carving the social ontology at the individual, he would break it down in terms of attention types and degrees of signal to noise ratio in verbal actions. He viewed economic activity in terms of attention streams recruiting each other for information processing and resource extraction. And so on. He said he was putting this into writing and emailing professors about it, and I wished him luck. He apparently got into a fight with someone at the group house hosting the party later that night but I never learned what exactly happened.
About a week later I found him at the cafeteria and asked him how his theory was going.
"Oh, that? I was going through a manic episode. I just came back from the hospital, and currently being stabilized with meds. Doctors suspect bipolar II."
Months later I talked to him again over dinner. Fully stabilized and "back to normal", he said he thought his theory was just bullshit coming from a manic mind. And now he was just recited full mainstream neuroscience ideas together with a Dennettian deflationary view of consciousness: "yeah, I guess it's all just my prefrontal cortex organizing sensations" or something to that effect.
I thought his manic self was brilliant; his normal self a complete normie without any interesting point to share.
Last time I saw him was at another party where he said he had stayed up all night on Vyvanse trying to induce a manic episode again. Didn't work out very well. Not that he he hurt himself (to my knowledge), but he just never was able to recreate the mental state of his episode, something which he dearly missed. I do remember he graduated and found a job, and I hope things work out for him.
Part of me is still hoping to see a book on "Economics as Attention Management" on the shelves one day.
{stc} retweeted
At 28, Dostoevsky was dragged before a firing squad, spared at the final second, and exiled to four years of hard labor in Siberia. Surrounded by brutal cold and severe illness, he lost his entire youth, but he never lost his spirit. Instead, those dark years forged his deepest wisdom and compassion for human nature. He went on to write Crime and Punishment at 45 and completed The Brothers Karamazov at 59.
No season of suffering is wasted. Yes, you can start all over again.
{stc} retweeted
Paper Mono!
sooo excited to finally release our free and beautiful mono font
it works incredibly well for design, coding, and even prose, and it’s packed with features: variable font weight, coding ligatures, duospacing (!) for long-form text, and more.
sound on:
Replying to @cradrrat
We have dumped JAX/SGLang and rewritten the entire RL and inference stack from scratch in C/C++, which took some extra time.
Training (aka pre-training) is done. Lot of cleanup was needed on the SFT data to get that mostly right. Now, 4.8 is in RL and the evals are starting to climb. We will do a soft rollout to Bot users later this week and see how it goes.
Grok 4.8 will bear no resemblance to 4.7 (thankfully). It is not yet incredible, but, this time, the foundations are built right, so I expect it to evolve to be something special.
The evidence from near-death experiences, deathbed visions, shared death experiences, children who remember previous lives, past-life regression, spiritual regression, and mediumship suggests we incarnate to learn—that life is a class for the soul.
{stc} retweeted
some alpha drop for doing inexpensive vuln research.
instead of writing an article on the prompts/process behind this bounty (and more), I am dropping some prompts and the thought process I used as part of this research.
ik you might say prompts don't matter much, frontier models' baseline intelligence can rawdog vuln discovery (and it is true to a good extent), but there is a high chance that you either do not have access to cyber models, or even if you do, they are expensive (and with heavy guardrails). The idea is, these prompts work well with open-weight/less expensive models. Not saying these will outperform the raw intelligence and capabilities of frontier models, but open-weight models perform much better with a bounded/goal-specific approach (where the goal is set, and the search space is massive) rather than open-ended exploration (where the goal is vague, and the search space is massive).
anyway, here are some prompts. I don't need to share much, because you can just create variations of these and apply them wherever you wish.
- find places where the app is supposed to enforce something, but doesn't actually enforce it.
- find security rules that are enforced in one part of the app, but just assumed to be true in another.
- find security assumptions the app relies on, but never actually checks or enforces.
- find places where the app depends on undocumented behavior, and what happens if that behavior changes.
- find operations that are secure on their own, but become vulnerable when you combine or chain them together.
- find cases where the same object can be accessed in different ways, but each way ends up having different permissions.
- find objects whose identity can be changed or rebound to another user, without going through the right checks.
- find requests that get rejected or fail, but still end up changing something or triggering a side effect.
- find security checks that only work if certain events happen in a specific order, and what happens if that order changes.
- find operations where the transaction fails or rolls back, but some of the side effects still happen.
- find things you can still do with an object even after it has been deleted, or access to it has been revoked.
- find state changes that can be reversed or undone by someone who shouldn't have permission to do so.
- find objects that can end up in two contradictory states at the same time, and what operations become possible because of that.
I take such prompts, put it in a mutator (which has threat model of the project I am doing sec engagement), the mutator will mutate the prompts based on the threat model, and spawn swarm of agents, with different starting prompt and a bounded goal, the process repeats till all paths are exhauted, or some new signal is generated, which then feeds back to mutator.
Doing this using a cheap model is very effective and sometimes it might yield you something even expensive models miss, we are just trying to maxximise the search space, from as many angles as possible, and once this process generates enough signals, just use a frontier model to confirm/triage issues, or for finding similar variants of such issues, that's basically it. There is v less moat is making a custom harness (unless you are an elite vuln research who can codify their approach in it or a 100x agentic engineer, I am neither).
The internet is now something agents gather from for you. It's no longer somewhere you spend time.
34
this account is gold.
//shower thoughts//
technological history has been dominated by the engineer, the logician and the programmer but now that they’ve been overthrown by their own creation, the future belongs to hot people who can translate existence into an aesthetic experience and imbue it with meaning tailored to your soul
31
Dream team v2:
@elliotarledge @somewheresy then maybe like alex zhang or ben aratame or somethin. or the ceo of exo labs
1
1
45
{stc} retweeted
tiny models will get so much stronger
consumer hardware will get more powerful
honestly, good timeline!























