rglover

That's exactly what the focus should be. It may be naive in a modern context where everyone is addicted to "number go up," but it's the only way to get an authentic, not manufactured culture.

> So the Chinese labs have thrown a lifeline to the Western loss-making labs, and I just have no clue as to why.

"Thus the expert in battle moves the enemy, and is not moved by him."

They figured out a clever method for avoiding excessive training costs via distillation. That forces the hand of frontier labs to move faster, produce better models, etc. (to avoid embarrassment and 'falling behind'—all the while shouldering most of the cost), which they can just keep distilling—or applying other techniques against—much to the dismay of said frontier labs.

Checkmate.

> When does the psyop end?

> I can only help with official U.S. government services and information, not with claims about “psyops” or political commentary.

It doesn't work.

People aren't mortgaging their existence to get a piece of the almond market.

> But it seems to me an impossible task to find a solution while our thinking is guided by geopolitical interests.

...guided by money.

Forget the compute, what about the energy (and other resource) requirements?

IMO, when they have to hide, that's the beginning of the end. Too many people are aware of the what/why now. And above all else, once the tables turn on elected officials and they realize they're under the same surveillance, I bet their tune will change (likely due to a public egg-on-face moment involving illegal or suspicious activity). If people keep the pressure on, this will be a "remember when" rather quickly.

I'd be open to smart home stuff if it wasn't sending data out to third-parties. Are there any good homebrew setups anyone is using that are local-only (or at least only sending data to a "cloud" I control)?

> I honestly believe they have robots solved and that's why AI CEOs are shitting their pants now, and everybody is wondering what's going on.

Or it's the oodles of money they're on the hook for and subsequent reputation destruction haunting them like the grim reaper.

> Don't you think it's weird that they have a production line for humanoid robots targeting 20,000 units a week, and meanwhile say "the robots cannot generalize yet".

This wouldn't be the first time in history that happened. Lots of companies overshoot capacity being overly-optimistic [1].

[1] https://en.wikipedia.org/wiki/DeLorean_Motor_Company

Tell us more about the chainsaw tool. Fighting with chains and tensioners put hair on my chest.

> I'm thinking now the reason why we don't see robots is because they want us (the general public) at all costs to not freak out.

Not quite [1]. Your reaction is what's being sought after: to believe it's more powerful and capable than it is to (presumably) keep the cash flowing.

[1] https://electrek.co/2026/09/25/tesla-optimus-production-ramp...

Wasn't familiar with Johannes but what a body of work. RIP. You can tell he was one of the real ones.

Likewise. I get drained quick when I let it just spit out who knows what code.

I find a lot of value in using plan mode to get the LLM out of dead ends. If it's in Act and starting to spiral (evidence of looking for something or trying to better understand something), I'll kick into Plan and tell it to think a bit harder—also avoids it wasting tokens retrying the same techniques—which results in a much more considered, and more often than not, correct approach.

Just like rushing made messes in the before times so too does rushing via LLM. The exact same outcome will happen, but at a far greater velocity and scale than anything we've seen in this industry before. Old Testament, Mr. Mayor, real wrath-of-God type stuff!

Enjoyed your post too. Hadn't heard of Math Academy until today, appreciate the in-depth review.

Dude, hell yeah! I was just shopping for this exact thing not that long ago [1]. I've never been great with math and wanted something I could practice with away from my laptop. Love that it works in the browser, too. Well done, this is really nice.

[1] https://graybearding.bearblog.dev/how-im-teaching-myself-mat...

From the DHH talk discussed here:

> "Lean in to the max possible. Because the utopia is almost here."

Man. I use AI daily (in a more nuanced, HITL way) but this is turning into some Jim Jones shit. It's a tool, dude. Relax. I'm all for encouraging people to build and create things but it's just downright fucking spooky to see how many people are kneeling at the altar with their respective sacrifices.

Herein lies the problem: assuming that all people in a location/area are like that. Black and white thinking is the antithesis to a humanistic approach. Even worse is thinking your perspective is the "best" one and that whatever decisions you make are, of course, best for everyone.

We'd do a lot better if we accepted that everybody is different and homogeneity and the drive to force it on others is a disease of the mind. Yeah, there are plenty of assholes, but consider why: because someone taught that person as they were growing up (either directly or indirectly) that being an asshole protects you. I'm frequently an asshole because I got picked on when I was younger. But guess what? For the right situations, it works. For others a more gentle touch is fitting. That's the dynamic to encode into tech: leveraging a combination of deterministic and non-deterministic systems to create adaptable systems that relish in a mess as opposed to fall apart because they're too rigidly shaped to accommodate the most and only favor the perspective of the few. That's how you get inhumane systems: they're designed and maintained by people living completely different lives from the people they're building for.

You can distill it down to: don't be a piece of shit, think carefully about how what you're building is used, talk to real humans and listen to them, and stop driving your product development process mimetically. Instead, solve real problems that real people have (not people in SF; I'm talking deep reality like a 60 y/o woman who had to get a job at a bible belt grocery store just to make ends meet but really wants to help you find the powdered peanut butter for your recipe but can't use the app the store gave her to find things).

This doesn't happen because there's a "lack of humanities in tech." It's because there's a lack of care and concern for others in tech. Most are motivated by the usual stuff: status and wealth. No amount of prescription in the form of humanities can fix that or will. You don't learn how to take care of real people by burying yourself in study. And you can't negotiate with sociopaths [1].

You just have to actually be a decent person. You can't game it. Thats part of tech's problem: it's always trying to solve problems from angles that the majority of people can't see, don't share, or can't even afford access to.

[1] https://youtu.be/nVp0Dvh1XQM?is=Tn76Ed3AlIv4nqjt (first part in the car)

What does building those have to do with my point?

All of this to avoid the reality that experience and expertise are a thing that can't be easily commoditized.

Reminds me of a thought I had about galaxies being instance runs while writing Github actions to build database binaries.

Each galaxy and any planets/life it contains are just artifacts of an "existence run." That's why life feels like a simulation so often (and human history repeats itself): we're inside of an organic (human word) "program" that's trying to perfect itself by running similar, but consistently tweaked "logic" over and over. Things like evolution are just an algorithm embedded in the system.

If the theory discussed in the article is correct, then aging is (in the scope of the above thought) just one program among several that we interpret as existence/life.

Just a daydream but fun to think about.

MCP is a solid, obvious idea. The problem is the implementation and roll out. It was shotgunned from the start and made an otherwise simple thing far too complicated. IME, a shoddy foundation always leads to shoddy construction on top.

When done properly, MCP can be really useful as an API bridge. MCP is (or can be), after all, just a JSON-RPC endpoint that returns a standardized response format. "Tools" are just function calls w/ definitions that are LLM-consumable.

But, imo, it's foolish to dismiss MCP outright.

It also helps to develop better judgment and awareness so you never get into bed with bad people in the first place. "Lay down with dogs, you'll wake up with fleas" applies to business (even getting a job is "doing business") as much as it does other things.

Had to screenshot this as it's easily the best articulation of the housing problem I've ever read.

Beautiful. Can't believe how mini and colorful.

Is there a way to get in touch and chat about use cases? I have some low-stakes stuff I could test this against without causing a meltdown.

Assuming this isn't astroturf ("sources say"), this is a prime example of why you don't wholesale delegate your thinking and strategy—in a military context or otherwise—to an LLM. I really hope people are paying attention and don't just turn this into a joke. If this is real, this is a big, big, big fuck up.