philwelch

As I recall, “No Silver Bullet” fundamentally rested on the assumption that the subroutine was the last word in abstractions to make programming more efficient, which probably wasn’t even defensible at the time because Lisp had already been invented, and is even less defensible after the past several decades of programming language research. Brooks was still onto something when it came to irreducible complexity, but offloading complexity an LLM can tackle to the LLM still saves time.

One of the lesser discussed Brooks essays is actually the best description of AI-first development: the “surgical team”. It just turns out the surgeon is the only human, and like many modern surgeries, the surgeon is controlling a robot instead of operating by hand.

It would be interesting to reread The Mythical Man-Month and see how each essay applies to AI-first development.

> Normal, emotionally stable people do sometimes make decisions about what businesses to patronize based on the political leanings of the business owners.

Maybe with issues like abortion or racial discrimination, but not tariffs.

Why would someone make up such a banal rumor? I’m not saying it’s true, I’m saying who cares?

Normal, emotionally stable people don’t care if the creators of a programming language disagree with them about tariffs.

Also, if Zig itself doesn’t accept AI contributions, it’s probably NGMI unless somebody is willing to maintain that fork.

If the penalty for hiring too many people for the wrong job role is being barred from hiring people ever again even for a different job role, then companies are just not going to hire people in the first place.

> Generally lawyers need to be involved to make sure any rejections are compliant. There's a whole cottage industry around this.

That's why most PERM job postings are advertised in the classified ads of the print edition of a newspaper, instead of anywhere that anybody actually looks for jobs in the 21st century. If nobody applies, nobody needs to be rejected.

Use/mention distinction

If it’s a natural language prompt, it’s not a hook.

It's a bit more complicated than I implied because many or most Taiwanese prior to the beginning of KMT rule were still ethnically Chinese; they just hadn't been part of "China" for 50 years (a period when there wasn't a stable, unified "China" anyway). "Occupation" is a controversial term for the period of Japanese rule and the Japanese weren't "hated" in Taiwan to the same degree they were in other occupied territories. The period of Japanese rule from 1895-1945 was a colonial government, but it was probably better than what was going on on the mainland at the time--domination by Western powers, the warlord era, the civil war, and a much more brutal Japanese occupation. The difference between Japan's treatment of Taiwan and mainland China is a big part of the difference in perspective towards the Japanese between the mainlanders and the Taiwanese.

Some of the main proponents of the "Japanese occupation" narrative are the KMT, who committed plenty of atrocities of their own after taking over Taiwan and, among many Taiwanese, ended up more hated than the Japanese. The KMT was also serious about their lost cause of retaking the mainland, at which point they expected Taiwan and China to remain unified under their rule, with the famous "One China Principle" representing not just the CCP's desire to control Taiwan, but the principle shared by the KMT that Taiwan is part of China and should be under the same government. In recent years, the KMT has pivoted towards cooperation with the CCP with an aim towards peaceful reunification, while the DPP favors explicit Taiwanese independence (Taiwan's official constitutional stance still being that it is the legitimate Republic of China).

To be fair to the KMT, they also ushered in Taiwanese democracy. When Chiang Kai-shek died, his son and successor Chiang Ching-kuo ended martial law, promised to be the last Chiang to rule Taiwan, and began the transition to democracy. His successor, Lee Teng-hui, was Taiwanese-born and finished the transition to democracy, winning the first democratic Taiwanese presidential election in 1996 before stepping down at the end of his term limit in 2000, at which point power transitioned to the DPP. Lee was also controversial with the hardliners in his own party for, among other things, his more sympathetic attitude towards Japan.

Interesting point. There are a lot of sports (football, basketball) where the cumulative rules end up requiring any player to have a humanoid form (references to elbows and knees and hands and feet, etc.), but even in ping pong it kind of seems like cheating to have a non-humanoid form factor.

The majority of Taiwanese are the descendants of the people who lived there before 1949, not the descendants of the Chinese Nationalists who fled there at the end of the civil war. In fact, the Taiwanese were, uniquely among East Asian nationalities, relatively happy being part of the Japanese Empire and have maintained good relations with Japan ever since.

The Marines did it with a rocket launcher. https://www.twz.com/marines-test-fire-robot-dog-armed-with-r...

There is a very broad consensus that labor theory of value is nonsense however.

And when it comes to a theory being “bad at guiding policy changes and decisions”, attempts to implement Marxism killed more people than the world wars, many through famine.

Used to be more true than it is now. I’ve been in a couple really shitty Marriotts.

They already didn’t get money or recognition.

The problem is much older than that. Academia didn’t start overproducing PhD’s and exploiting grad students and adjuncts in 2025.

How does that compare to adjunct pay?

> You can’t merge partially merge a large “review commit by commit” PR so you are forced to wait until it is all ready to merge.

These are two different use cases. I thought we were talking about the one where a set of changes is more readable commit by commit but you still want to merge the whole set of changes, not the one where the change is too big to review and merge at once so you have to break it up into multiple reviews. The latter use case is more rare—frankly, it’s a bit of a red flag otherwise—and wasn’t difficult anyway.

Microsoft didn’t need to build anything because it was already built into Git. The only problem is, if people knew how to use Git, Microsoft couldn’t lock them into a proprietary version control platform.

We may not have AGI yet (“artificial general intelligence”, AI equivalent to a human), but we sure have AGI (“artificial Guardian intelligence”, AI equivalent to The Guardian).

Corollary: since self-driving isn’t a solved problem yet, this proves that cab drivers are already smarter than the Guardian.

You can review PR’s commit by commit, and you can ask people to review PR’s commit by commit. Not a big deal.

Not at all. One of the tricks of using AI is context management and managing the Git history yourself can be a big part of that. If the AI has a stupid idea and implements it, even when you tell it, “that was a stupid idea, don’t do that, change it back”, the history could persist and the stupid idea will poison the context window every time an agent reads the commit history. It’s even worse if you had the stupid idea!

Also, my current workflow actually has hooks to block agents from creating or changing commits. I know at some point this will be a limit to scaling, but I think that will result in me spending more rather than less time in git.

Squashing is fine if you’re just making a mess of temporary commits as you work and you don’t want to keep any of those changes separate in master, but that’s not a useful review workflow. A lot of times I’ve built a feature in a way that decomposed naturally into e.g. two commits: one to do a preparatory refactor (which might have a lot of noisy and repetitive changes, like changing a function signature) and another to actually change the behavior. You want those changes to be separate because it makes the changes easier to review; the reviewer quickly skims the first commit, observes that it’s a mechanical refactor, and the change in behavior has its own, smaller commit without all the noise.

“What if there’s feedback and you need to make changes after the code review?” Then I do the same thing I did before I posted the code review: make separate “fixup” commits and do an interactive rebase to squash them into my commits. (And yes, I do validate that the intermediate commits build cleanly.)

There’s nothing you get from stacked PR’s that you don’t also get from saying “please review my feature branch commit by commit”.

No but they might be able to fix authentication problems, which is what this is.

Jobs was a perfectionist and a minimalist. Part of minimalism is that sometimes you delete marginal features (arrow keys) that you still end up wanting back.

If you never delete too many features, you aren’t deleting enough features.

You’re right, it’s much easier to replace the entire federal bureaucracy.

You’re confusing a few of the issues here. This isn’t an interstate commerce case, it’s a tax power case, which does reflect on the NFA but not on Wickard v Filburn or civil rights. The important aspects of federal civil rights law aren’t legally justified by the commerce clause either.

Though I would gladly see Wickard v Filburn overturned. Commercial regulations already vary by state, and the US would still be more cohesive than the EU is today, but the amount of water that flows through my showerhead doesn’t have to be a concern of the federal government. In fact, we don’t even need Wickard v Filburn to be a more cohesive federation than Canada, which doesn’t even have free trade between provinces.

There isn’t really a shortage of money for those things, just rampant levels of fraud, corruption, and incompetence in the government to make those things artificially expensive. California spends so much money on high speed rail and gets 0 feet of track because they’re not paying for track; the whole thing is a scam where the politicians give taxpayer money to their political supporters in exchange for political support. Defense isn’t immune to this either; Boeing, which builds a shitty heavy lift rocket out of Space Shuttle spare parts and delivers it late and over budget, pulls the exact same bullshit with their defense contracts, and there’s always some shitty Senator siding with them against the American people whenever anyone gets upset.

Coding is one thing that is genuinely more enjoyable with AI than without it. It’s a different (but overlapping) skill set, but my median AI sessions remind me of the most exhilarating design discussions I’ve had with colleagues, and I get a lot more done more quickly than I used to.

Customer support is kind of something you can use AI for; most companies will foist you off to some system of exchanging written messages, which is annoying, but then you can use an AI to write your side of the conversation. It’s ill-mannered to do this when you’re interacting with actual people, but customer support is another story.

> Look at Suno. Fantastic tool, but where was the capital need to make music generation so cheap that no musician could ever compete with it? Did the world really wake up one day and concluded that, "wait, we're spending too much on musicians"?

People didn’t know what LLMs would be capable of until after they were invented. Cheap music generation turned out to be easy once we had cheap text generation, and cheap text generation turned out to be a tractable problem.

What a load of pointless handwringing.