[deadbabe]

People want to make it seem like you need to always use the latest and greatest frontier models to be taken seriously as a developer.

You really don’t need them. After a certain point, bigger models give diminishing returns. If you can get 80% of the productivity gain with a free local model, use the local model. It will still be way faster than doing everything by hand, but you also don’t have to pay for tokens to a cloud provider and the tools won’t be ripped away from you on a whim.

This is the new attitude enlightened people should adopt. Reject the arms race.

[organsnyder]

The biggest appeal of the frontier models is for those trying to get autonomous agentic systems running that do real work with minimal human input. I went down a rabbit hole trying that with frontier models, and after a lot of initial promise it ended up actually slowing me down.

[pimeys]

We've all been through that no? In the beginning you can do a ton of stuff without reading code. But the LLMs miss all the good abstractions, they just push and push unmaintainable code until at some point you start having more bugs and then you NEED that LLM to fix the codebase you don't understand anymore.

There are guardrails you can and must add to protect your team if you take the vibe approach: a good type system, a good database with clearly written business model and a good data model to drive your business. Make it loud and clear when something breaks with your tooling.

But... I'd definitely not vibe everything after a certain point. Reading and fixing code is also a lot of fun.

[apitman]

They're insanely good for prototypes though. To be able to actually see something working before deciding whether it's worth investing the time to build it for real is invaluable.

[dmbrThnU]

Made an account to semi-disagree with you, haha!

I have to advocate for the vibe-coded mess-colony.

There are applications where it either works or it doesn't, and it's simultaneously obvious whether it does. Think stock price prediction software. I've killed time in the evenings verbally chatting with agents about that specifically, and what emerged worked! It didn't work well, but it clearly outperformed randomness, and I was able to verify that myself easily.

I didn't look at a line of code, but I had an absolute blast.

[austinthetaco]

You couldn't have possibly verified that. Stock prediction based on what? What's your sample size over what period of time? Using what indicators? how far is your lookback?

[dmbrThnU]

This was a toy that I made.

Are you familiar with the concept of a Markov Chain? (If not, it's a simple tool that technically works better than randomly guessing for predicting stock movement.) I designed a very intense neural network meta-architecture, applied it, and the results were the same as if I'd used a basic Bayes model or Markov Chain. Which is a little humorous; I very much used a bulldozer to sweep the garage.

I used close minus open to determine up vs. down movment. Can't remember the lookback, but was predicting the immediate next day. Over the entire US market, a basic Markov-based model can predict the next day 52.5% of the time or something like that. (Given 1000+ stocks, you guess which direction all will go, 52.5% will be correct guesses.)

For what it's worth, I don't really know the details of the statistical tools. I do have a good grasp of train/test/validate sets, so I know what my results meant.

[naveen99]

Just keep a good grip on the data. Code is easy to replace then.

[ravenstine]

> People want to make it seem like you need to always use the latest and greatest frontier models to be taken seriously as a developer.

Except you kinda do. Try getting a job today without mentioning Claude experience. In another year it'll probably be something else. Saying you like to use Copilot today makes one seem elderly.

Not saying you need frontier models on a technical basis, but for career PR you probably do.