jeremyloy_wt

You could unironically vibe code this with a prompt shorter than your complaint.

I could not grasp the “why” here, not for the life of me. A Mac mini/studio would have been so much straightforward for the actual cited problem of “needing to develop on Apple hardware”.

Sometimes we do things just for “the love of the game” and learn something along the way but I couldn’t help scratching my head the entire time while reading this

I’m so happy Russ still contributes even though he isn’t lead anymore. I always enjoy reading his blog posts

Admittedly I’m pretty ignorant of the details, but I thought this was the mixture of experts architecture

They napkin math is fairly easy to do. One human works around 250 days per year, and if we assume Bay Area salaries we could assume ~300k/y conservatively for a fully loaded cost.

$1200 per day.

Your estimation is 50*11 days so $660,000. That’s 4x what Claude cost.

That’s assuming that you actually get those 50 people to work without blockers, stepping on each other, or other coordination issues. The coordination complexity alone is astounding.

I don’t like it necessarily, but Claude wins here, easily. It’s not close.

And if the engineer bills for $100/hr or more, the trade off is worth it

> the way music is recorded and mixed today takes all the soul out of music

> So much precision is required that session musicians are playing most of the things you hear, not the actual artists

I’m sure the session musicians don’t appreciate this statement. Just because they can play with high precision and reliability doesn’t mean they are playing without soul.

If the featured artists can’t do so on their own, that’s sort of a knock on them, isn’t it?

I think there will always be some sort of mastery element to communicating with AI. You can see that nowadays with juniors who blow through tokens and still don’t produce good results. There will always be some group of people using AI better than others. Maybe the results will be equal in the far future, but some other dimension like cost may be better used by some people

Emerald is well regarded as the best of Generation 3, which is the final of the traditional 2D games and can trade with Fire Red/Leaf Green (remakes of the classic)

So you have available all of the original Pokémon

I don’t understand your explanation on mitigating the confused deputy. If the attacker has access to the database, can’t they just read the IDs for the target row they are overriding first so they can generate the correct hash?

> we as humans can guide the LLM toward a rigorous test suite, rather than one that has a lot of "coverage" but doesn't actually provide sound guarantees about behavior.

I have a hard enough time getting humans to write tests like this…

Funnily enough, the Sleep Schedule settings screen on iOS (accessed through the Health app) looks very similar to this.

The official Typescript compiler is being rewritten in Go as we speak.

The second that you would benefit from using a DBMS specific feature, the ORM begins getting in the way. It is highly unlikely that an ORM provides support, much less a good abstraction, over features that only 1/N supported DBMS have.

Your code ends up using the driver raw in these cases, so why not just use the driver for everything? Your codebase would be consistent at that point