softwaredoug

Yeah they may not be interested. But pretending you have that moat a dumb strategy that's not working.

Well its not law. Its more terms of service.

For example, if OpenAI / Anthropic were actually open, other US labs could be building near-frontier open weights models by distilling off OpenAI / Anthropic. But because US companies don't want to be sued, US labs who obey terms of service, will be at a disadvantage to Chinese peers.

Maybe US labs need to just not care and distill from OpenAI / Anthropic anyways?

OpenAI and Anthropic should enter into distillation agreements with other US labs. Turn a threat into a profit center.

Other US labs cannot directly distill from OpenAI/Anthropic as it’s a violation of the terms of service. It holds other US labs back. Leading them to build second tier models And in the end OpenAI/Anthropic may be unable to prevent distillation.

Why fight it when there’s clear money to make here?

If they did this in the US they would almost certainly be sued.

Meta, for examples, doesn’t want employees to use Claude Code due to distillation risk.

I think it’s important to have some kind of domain knowledge that goes with software ability. Not just be a software engineer.

The closest comparison to what’s happening today is writing. In olden times a few in the professional class could write. It was not a common skill. Now everyone can write. There are technically more “jobs that involve writing” and writing skills CAN help you succeed. But very few people have a job just writing.

What people do have are domain specific jobs where a writing skill is more useful. Journalism, Industry analyst, etc.

The future won’t be some small set of software engineers focused only on functional code, it’ll be domain-knowledgeable people that plan out domain solutions, write, code, all in one role.

To be fair the lag varies tremendously. When Deepseek first came out it was competitive with frontier lab models at the time.

I think GPT 5.6 Luna works great honestly for most boring coding tests. And saves money.

Is anyone using these models as their daily chat driver? I am skeptical.

The vast majority of Westerners will interact with ChatGPT/Claude/Google in a browser. They'll use these models to try to save some money coding.

Something I did not realize. The White House instructed intelligence agencies to help in preventing distillation of US models.

Whether or not distillation matters a small amount or a big amount, still interesting:

https://www.whitehouse.gov/presidential-actions/2026/06/nati...

Haven’t we been in this “China is 3-6 months behind” for a while now (maybe up to a year? Longer?)

The actual difference is how much scrutiny and time was put into the Mythos / Fable and GPT 5.6 release. Making it feel like “these are a big deal”. Spring and summer THAT was the AI story

Then Chinese labs release models that approach Fable performance. We’re shocked they just seemed to appear out of nowhere.

It’s less about the gap closing. It’s more about the weight we put into Fable-capable models.

By the same token, don’t expect China to do anything about distillation attacks on the frontier labs. I think it is mistaken to attribute all of the success of Chinese labs to distillation, but it’s just as much of a mistake to pretend like distillation doesn’t give Chinese labs a big advantage.

I think we see this with Meta being paranoid about internal Claude usage, to avoid inadvertently distilling[1].

If distillation is a driver, then smaller American labs could be distilling, but are not for legal reasons.

But that's a big if we just don't know for sure.

1 - https://cryptobriefing.com/meta-restricts-claude-code-codex-...

It's hard to ignore the overall environment

US has made itself visibly unaffordable, anti-science, and hostile to immigration.

For top scientists at these companies, there should be clear upside for the immigration to the US. That just doesn't exist anymore. Especially as quality of life increases in China

Also the courts that let him get away with it.

That said Congress can and does sometimes make the funding less discretionary. They can mandate expenditures. And make withholding legal funds personally painful for OMB. They also haven’t don’t this yet.

When you see Chinese emerging AI dominance, you can’t help but connect it to the administrations hostility to immigration and research.

We should be attracting and retaining the world’s top researchers, not driving them away with dysfunction and xenophobia.

In the past it’d be hard for Chinese labs to retain researchers. Now the upside to the US isn’t there.

It feels like an inflection point of lost US leadership in technology? A year plus ago you would say while China led in green energy and manufacturing, at least the US was ahead in software - as demonstrated by the state of US AI models.

We could point at a lot of factors on the US side. From political paralysis / head-in-the-sand attitudes towards emerging tech like green energy. To something of disdain for workers that will be impacted by AI (creating a backlash). To education that continues to lag. Add to this so many other self-inflicted economic wounds from the current administration.

I don't know if its nearly as terminal, as say the UK after WW2. The US is still large, wealthy, and resource rich. Yet at a minimum the triumphalism about US leadership after Trump was elected by the tech elite feels silly in retrospect.

Something I also think about is how much stronger The West overall would be if instead of antagonizing allies, there was a single ecosystem working closer together.

There are many people with unbridled enthusiasm and an ability to write fast

Knowing Simon - he’s very much both :)

Whatever the headaches of the internals, I haven’t heard any substantive outcomes in hacker news that have been impacted by this change.

Yes if we had a Zig->Rust transpiler maybe it would produce similar output. But I can’t find one. We’d have to use an agent to build one first :)

People shouldn’t overthink blog posts. Just write what’s on your mind and have fun.

Old and busted: benchmaxxing

New hotness: pelicanmaxxing

It's also a meta-lesson about Kaggle.

Kaggle winning solutions rarely make sustainable engineering solutions for teams. Maximizing just model performance against an objective is a small part of the bigger picture.

So Chinese labs are driving essentially towards commodotized intelligence. Even if its a few months behind the US.

Is this a classic 'commoditize my compliment' situation? They want to sell the hardware and infrastructure behind AI and make the software part not the value driver / moat?

I can see it. But also even two Chinese labs sinking 100s of millions USD into training isn't exactly commoditization. It's still a ton of effort with dubious payoff.

I just write code.

Not nearly all the code. Maybe 5% - the important code. To keep things sane, explore new areas, get a deeper understanding of the system, delete/simplify agent slop, improve test coverage, or make some other foundational change.

Feels like we’re bending over backwards to avoid just doing this. But it seems for me, at least, the way to have the highest leverage on an agent heavy project, it’s fulfilling, it helps the agent, and it keeps me in flow.

I still think I haven’t gotten much more productive than a year ago:

I wrote the barebones, scaffolding, and where I cared about the details. Agents stamped out the rest from those patterns.

Everyone speaks about this stuff in anecdotes. Or heroic attempts to have agent only projects.

But very little actual data on real gains in actual quality as experienced by users? Harder to measure IMO. And may even be subjective to each developer.

One day we'll get to the plateau of productivity of this hype cycle and wonder WTF were we thinking.

Not that we won't have most code written by agents. I truly beleive that to be true. But that we got so far down the "never look at code" rabbit hole that our abstractions will become so divorced from the reality that is the actual.. freaking... code. We will truly have spaghetti code, bloated towers of babble.

We'll tut-tut these times and think "if you only spent half a day thinking about the big picture, looking at code, dare I say write some code you could have saved the project a whole mess of trouble".

In the meantime we're going to be generate 10K in spaghetti what should probably be 500 reasonably understandable LoC.

Spending a day reading some code can save you $100s of tokens, and weeks of headaches.

But why aren't lower level "ideas about execution" more clearly expressed in code? Or at least pseudocode?

Many of us would struggle with our reading comprehension of an English description of an algorithm. But the pseudo / actual code? Much easier to reason about.

I'm not saying you should review even 90% of code. But it feels like an arbitrary line to say "never look at code". It's like never taking a wrench to a car to look at what's built to see whether you trust the car factory.

This article feels too black and white

How are you supposed to review 5k lines of code every day?

Maybe there’s grey areas of reviewing 10% of the code, ensuring tests actual are meaningful, the big ideas are correct etc. I personally find that more efficient to do by looking at some code than a proxied description of the code.

“how is exactly the design of that part? How does it work?”

Isn’t code and syntax often, but not always, a more precise way to see that? Won’t it depend on what you mean by “how does it work”? I’d prefer pseudo code in some cases and actual code in other cases. There are times line by line details matter. There are times when bigger ideas matter.

It’s all like saying you run a car factory and should only ever give out the big ideas, never take a wrench to a car and figure out of if the factory builds cars to your expectations.

I write 2% of code. And read 5-10%.

I feel like everyone is being too black and white. And everyone has different ways of staying in flow with a software task.

Probably because agents in typescript don't waste time reasoning about so many different problems that come up in a lower level language. They stay higher level, and don't waste context on thinking about borrow checking, like you would in Rust.

Exactly this. Spending a few hours crafting and debugging a test you trust save hours of pointless agent time.

There are ways humans have an advantage. Not in the line by line code, but in the experience of having the whole mental model in your head.

Writing some code with the whole context in your head is the human advantage.

I’m not precious about that code. Maybe it needs to be rewritten by an agent. Or we could build good abstractions so it’s trivial to see the system. We could also use systems that by their design force us + agents away from security, performance, etc