LLMs really do still just reassemble things in their training data. There’s just a lot of it now, people anthropomorphise and struggle visualising large things. Some people say it’s truly reasoning but hit a topic that is under represented in the data of any LLM and it’ll transport you very quickly back a couple of years and ruin the illusion quickly.

Similarly I've run into several "you just gotta know" type problems where LLMs still just fail. My favourite was a quirk in how async relationships work inside emberjs. Three different models gave a variation of the same incorrect answer. When I searched myself, I initially came up with nothing and eventually found what I think might be the only example of the same issue on the internet, a single stack overflow question wth two responses. The first is what Claude, gemini and chatGPT said, the second response was the OP saying it was wrong. I asked in the Ember discord and a core team member responded instantly with the answer.

There's a video I love on Youtube, where a guy uses ML to assemble a blank jigsaw. It performs amazingly, the jigsaw being blank is of no consequence and if it did have an image, it'd perform worse. That's all LLMs do. Just because the jigsaw pieces are smaller, they're still just getting assembled in whatever way fits, there's no mechanic for interpreting the image on the front.

"If one teenager works hard and saves for a few years to buy themselves a car and another has theirs bought for them on a whim, who would appreciate their car more?"

Odds are it's the former, not always. But mostly.

Worst of all I've seen good engineers lean in and begin thinking uncritically and magically.

But overall yes, time spent thinking is the thing that matters.

It's less "there should be more friction" and more "LLMs remove too much". Instant gratification is great if you're a consumer, awful if you're a creator.

See, table saws are dangerous. Famously so. One of, if not the most dangerous tools available to the general public. They spin quickly with lots of torque and pull things in faster than you can react. Pressure can also send loose pieces of wood backwards at high speed. Fast enough to pass through a person sometimes. It's like being hit with an arrow.

Tablesaw accidents can remove fingers and hands instantly, puncture organs.

They can be used safely but they're circumstantial, the worst thing you can do with a table saw is experiment. Once you realise there's a 12-inch razor sharp blade spinning at 3000rpm with up to 5HP you begin to respect how dangerous it could be and want to warn others.

It's intentionally hyperbolic. But you see what I'm saying here?

But for sure a systems language is going to be far faster on paper and Rails is far from perfect and does have some performance foot guns you need to avoid. And yeah, architecture is everything.

The elephant in the room is so many projects already operate like this without formally announcing it.

If you look at Blender, one of the biggest and most successful OSS projects out there, it's effectively run as source available. Some PRs make it through, but for the most part there have been heavy barriers to entry to get your work into the product. In this example, it's been key to such a large and complex project with millions of users staying afloat. It's an inconvenient truth.

It's one of those unspoken things in open source - the bigger the project the less you can accept or vet contributions. The less able you are to respond to users because there are too many. The amount of code you need to own balloons. The signal to noise to too much. LLMs have massively exacerbated this issue.

"We just ported Vite to ActionScript in 11 minutes, we swear for legit technical reasons"