setgree

And yet, their models are often behind and the team is bleeding talent. A moat is not a compendium of intimidating features, it’s a matter of whether a company deters competitors and maintains a competitive advantage. So far, Google obviously does not have one here.

And what should we do about apps' getting hacked, sending out malicious updates that get auto-updated and thereby infecting tens of millions or billions of phones simultaneously?

I'm not saying we have the perfect system but anything that slants the system towards "easier downloads" or "less gatekeeping" brings large, obvious risks. I don't see how regulation would address them.

> In some cases, though, its prose is denser than Claude Fable 5's: sentences run longer and there are fewer paragraph breaks.”

This sentence reads like Claude wrote it. Perhaps it did, or perhaps Claude has learned to write like the folks who work at Anthropic?

(Had I edited this, I would have said that a colon is not the right separator here. The second clause does not _explain_ the first, per se, bur instead expands upon it. Consider instead: "In some cases, however, its prose is denser than Claude Fable 5's, with longer sentences and fewer paragraph breaks.")

It's fun to see that even an extremely large company can find unexpected product market fit [0]. Per this article, "The company reportedly did not possess an engineering team dedicated to business customers or staff focused on developer relations, and lacked an enterprise AI strategy." That sounds insane in retrospect, but I think there's just inherent uncertainty in what people actually need and will use things for.

[0]https://pmarchive.com/guide_to_startups_part4.html: "In a great market—a market with lots of real potential customers—the market pulls product out of the startup... The product doesn’t need to be great; it just has to basically work."

Jeremy Irons's CEO character in _Margin Call_ explains his job like this:

> Do you care to know why I'm in this chair with you all? I mean, why I earn the big bucks?... I'm here for one reason and one reason alone. I'm here to guess what the music might do a week, a month, a year from now. That's it. Nothing more. And standing here tonight, I'm afraid that I don't hear - a - thing. Just... silence.

I'm sure many CEOs are bozos who can be replaced by a statistical averaging and guessing machine. But this skill of saying "it's time for us to dramatically change course because the music has stopped" seems to me very contrary to the current LLM paradigm. RL makes the chatbot agreeable ('aligned'). But changing course is inherently disagreeable. It makes enemies and people are going to try to talk you out of it at every turn. I think we'll need a new idea of what LLMs are for, and how to train them, before they can do that. (Much easier to mimic and therefore replace a Ballmer than a Nadella, in other words.)

This thing is going to cost >$10K, probably, which I think means it's not in the consumer segment

I enjoyed "up to an 80-core GPU, and a staggering 512GB of unified memory"

Staggering?? I can count on no hands the number of times that a fact or figure has caused me to stagger.

~salute-emoji~

I do not hate you :) I think you will probably have more positive interactions on the internet if you choose a different register to communicate in but I promise I do not actually care

What I wrote was neither passive nor aggressive. My telling you now that I think your response was ill-considered is also not passive aggressive, though it's arguably aggressive.

I get questions about academic papers I've written from non-academics sometimes. It's fun, I like that people are reading. However, some of them are clearly written by AI. I typically answer anyway, but on my most recent, I appended at the end:

> P.S. I intuit, and Pangram agrees, that this last email was AI-generated (and apparently so was the first one, now that I checked). As general career advice, I wish to observe that learning to write emails unaided is an extremely valuable skill in the dual senses that 1) writing is thinking [0] and 2) many people who receive an email they believe to be AI-generated will not be interested in corresponding, even if you're asking good questions, which you are. It would be a shame for that to get in the way of your research goals.

A polite, generic response like this is reasonably low-effort -- you must draft it once -- and might really help someone. Who knows though.

[0] hyperlinked to https://www.nature.com/articles/s44222-025-00323-4

First, I think it’s likely that you hear more about one than the other because you read websites like Hacker News and are not (I assume) an FBI agent or someone else who works on this stuff. A public defender I know would say that child abuse of all sorts is ever-present and relevant to their work, unfortunately.

Second, detecting CSAM leads to its producers who are by definition abusers. Here’s a nice article in Wired about the digital forensics of cracking down on a CSAM ring with some interesting details about the role played by crypto, and abusers’ misconceptions about it: https://www.wired.com/story/tracers-in-the-dark-welcome-to-v...

With that said, I agree that a lot of political concern for this is a smokescreen for a creating more surveillance. “think of the children!” has the flavor of a rhetorical trump card.

The goal of a modeling exercise like this, which you don’t have to buy, is to generate a simple set of initial conditions that can explain things we already know. Then, we can manipulate some initial parameter value to make predictions about things we don’t see, but might.

Likewise, it is obvious that gravity exists, but a simple model that explains where it comes from (in quantum terms) would be a big breakthrough iff it came with plausibly testable implications that could be tested via experiment.

I'm hesitant to use 'always' language but the innovator's dilemma is indeed another application of the same principle. It's remarkable when a company like Google can pivot in a way that threatens its existing core revenue streams.

I like the clarity, tone, and readability of your webpage. Also your FAQ is refreshing

> When Should I talk to sales? > Talk to sales if you need high-volume pricing beyond 2M credits/month, custom rate limits, SSO / SAML, SCIM provisioning, an uptime SLA, annual invoicing, an MSA / DPA, or a dedicated support channel. Reach us at hello@context.dev or through the contact page.

Would that this were the norm everywhere, rather than (say) a sales rep from Datadog scraping my phone number from who knows where to ask about my company's needs after I sign up for a free account on a whim :)

I'm not an expert but I think this is an old lesson in warfare, that guerillas can triumph over larger adversaries by being more exploratory/iterative and less rules-bound. Tolstoy tells this story in the second half of War and Peace. Likewise with Iraqi militants wreaking havoc with IEDs. People repelling an invader have every incentive to move fast.

Ah, I got confused by the name and acronyms.

My name is Seth Green, which I share with a more famous actor [0]. I go by Seth Ariel Green for disambiguation.

GPT-5.5 tells me about the actor, but Claude Opus 4.8 and, weirdly, Grok 4.2 know who I am [1]. I wonder if that's because I use Claude more? Grok I have no clue why.

[0] https://www.imdb.com/name/nm0001293/

[1] https://med.stanford.edu/profiles/seth-green

The part about this that's amazing to me is that they still are doing nothing after he noted another GDPR violation [0]. He's obviously both competent and litigious. What does the company expect to happen next??

[0] "Under Article 77(2) of the GDPR a supervisory authority is under a binding legal obligation to keep a complainant informed of the progress and the outcome of their complaint. It is not a courtesy and it is not discretionary - it is written into the law. I filed my complaint with IMY, IMY passed it on, the case ended in a multi-million euro enforcement action, and not one of the authorities involved thought to tell the person who started it."

Yep, Sansa climbed the ladder better than he did. As she says to him at the end, "Thank you for all your many lessons, Lord Baelish. I will never forget them."

I work at a research lab that was previously supported by an R01 grant that did not get renewed last year. It’s been tough and the staff (including me) have been moved to part-time employment.

However, it also made us put ourselves out there and fundraise, which led to new connections and new opportunities.

So yes, it’s been chaotic, but like Petyr Baelish says, chaos is a ladder.

> Hardware of that caliber requires a highly optimized operating system to function properly.

But unfortunately, you get Windows

`<!-- Pelican Eye / Sunglasses (Cool Retro Aviators) -->`

wtf

`<!-- Gold Rim -->`

WTF??

You know the old adage about if you meet one jerk in a day, it's them, but if everyone you meet is a jerk...anyway I'm glad you like CrossFit, if climbing weren't my thing, I'd probably do something in that category instead :)

Good for you, OP! Climbing gyms are especially good for making friends because you are working on problems with people. My gym has a weekly meet up for people looking for belay partners as well as classes where folks talk. Crossfit might also do the trick, as might a running club. Good luck!

Pardon, you’re right

It’s also not correct; et al. is conventionally applied to three or more authors (it means “and others,” plural)

"Thank you so much for your thoughtful, candid feedback. You are absolutely right to be annoyed. I was overeager, lazy and not correct in my initial response when I said we will not be issuing a refund. However we will not be issuing a refund."

Every so often the Trump administration seems like they might actually care about getting my vote. A recent executive order making it easier to do research on psychedelic therapy is another example [0]. A policy shift to reform IRB review for social and behavioral science [1] would be really targeted at me.

I know politics is hard to talk about, but I generally think that we underappreciate the importance of being agentic in politics. Obviously I prefer that our government follow the law and uphold the constitution. But the many ways in which the current administration got things done by being quick, by "flooding the zone" [2], and by using tactics that apparently no one noticed before [3-4] are worthy of study and emulation.

I know the obvious response to this is to note that a lot of what they're doing is illegal, and again, I think that's bad. But they really make the current Democratic leadership seem out of touch and old [5] by comparison. Combined with policy positions that are far from the median voter's [6], it doesn't make for a winning look/platform.

[0] https://www.whitehouse.gov/presidential-actions/2026/04/acce...

[1] https://www.cspicenter.com/p/its-time-to-review-the-institut...

[2] https://marginalrevolution.com/marginalrevolution/2025/02/tr...

[3] https://www.newyorker.com/magazine/2025/10/27/russell-vought...

[4] https://www.newyorker.com/magazine/2026/03/16/the-unmaking-o...

[5] https://www.nationalreview.com/the-morning-jolt/the-democrat...

[6] https://www.slowboring.com/p/the-median-voter-is-a-50-someth...

Did you do a coding bootcamp or such?