Hacker News Reader

Comment by cracell | Hacker News Reader

This a huge feature for coding.

I don’t want creative code. I want boring code following generally agreed upon best practices.

If you need more creativity it can come from the user or the harness.

In your code, this makes the model subtly devolve into certain patterns and stereotypes that aren't best practices, it's just a randomly frozen trajectory that is extremely hard to change zero-shot, without running a second pass. In other words, it's the same problem as in writing. Moreover, this is harmful for software design, where the ability to recognize the uncertainty matters. This is the reason everybody does the same thing. Frontend design also suffers from that.

>If you need more creativity it can come from the user or the harness.

Not fully possible. You can randomize the prompt all you want but you'll always get repetitive patterns and same choices in the same generation. This failure mode makes the model learn a very rigid mapping between this input idea and that single output idea (at best! at worst it's multiple input ideas -> single output idea).

> In your code, this makes the model subtly devolve into certain patterns and stereotypes that aren't best practices, it's just a randomly frozen trajectory that is extremely hard to change zero-shot, without running a second pass.

But in practice, who uses models “zero-shot”? It is standard for coding agents to review their own code even before returning to the user — so a zero-shot agent invocation isn’t a zero-shot model invocation. And then many users will get a different run of the agent to review the change a(frequently even a different model being invoked by a different harness-e.g. I personally mix Claude, Copilot and Cursor)

> This is the reason everybody does the same thing.

If you have a human-in-the-loop reviewing the agent output, different humans will have different opinions on what the agent should have done instead, and they’ll tell the agent to redo its output in different ways. And then they tire of repeating themselves, and it becomes guidelines in AGENTS.md.

Additionally, a coding agent asked to add a feature to a large code base will frequently check for the code base’s existing design and coding patterns, and so will end up producing very different code than it would have given a blank slate.

Yes, obviously external sources like human input, random generators, tools will bring variation across a single generation, but there's not remotely enough semantic entropy in them, unless you make meaningful decisions each little step. Longer reasoning and unattended agentic loops and self-reviews won't introduce variation, the intuition is wrong in this case. Anything isn't explicitly conditioned by external sources will be sampled from model's knowledge space, and if it's constrained by this effect compared to the post-training set distribution you get reduced capabilities. You can observe it in the form of everything looking the same in practice. It's a bit of a butterfly effect, some training quirk in a model somewhere is essentially shaping the entire world.

If you fix mode collapse, the effect on software engineering use case would be neutral at worst, or an improvement at best. It's not a verifiable task and has plenty of room for subjective judgement which benefits from semantic variety. For the same reason greedy decoding (temp 0) isn't used even for coding, it leads to degraded and pathologic outputs.

> Anything isn't explicitly conditioned by external sources will be sampled from model's knowledge space,

If you are using multiple models – especially models from multiple vendors – then you aren't dealing with a single "model's knowledge space".

Sure, however spaces of these models overlap to the point where we have articles about pelicans on bikes always facing right.

Right, and I definitively don't want to guess at what I might want. A "round trip to the user" is much better than getting bogged down implementing something I didn't ask for and don't want.

Which is good is you're developing a tool to turn natural language into functional code. But it's bad if you need a trillion dollars because you promised everyone a magic wishing genie.

Reply on news.ycombinator.com