Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why does Opus 5 feel worse to work with?

mun-logadan.github.io

Post reply on HN
Print view View original

501–510 of 915 posts

Re: Why does Opus 5 feel worse to work with?

#501

Post by craigmcnamara »

zucked wrote: post #434
mlsu wrote: post #428

Earlier quoted context omitted.

Everything that claude writes fits into the same aesthetic structure. The aesthetic is that of an expert slowly revealing an insight to the user. The actual content doesn't matter. - "Introduction that rephrases your prompt." - "3 paragraphs, with one section of bullet points" - "The Twist" - "The Bottom Line" It's really obvious once you see it. Every single prompt, from a quantum physics question to a mundane obser…

The second bullet point, down to the comma in the middle of the sentence, is what has been driving me absolutely batty of late. It's a surefire tell that I cannot seem to beat out of my outputs. It CONSTANTLY does it, even when you say not to. Between that and the insistence on "this, not that" structure makes me want to install the caveman skill and use it even for non-code workflows.

This load bearing concern belt and braces.

Re: Why does Opus 5 feel worse to work with?

#502

Post by Anoian »

TLDR: We need agents that can read our minds.

I think we want two opposing things:

1. An agent that acts autonomously 2. An agent that acts like we would

The problem is that an agent can only act like we would if it would know our mind and all the bits and pieces we did not define but are obvious or clear to us.

The only real solution to get an agent to act like we would is to make it ask clarifying questions, breaking the first requirement we have. Until we have agents that can literally read our minds, we cannot have both.

Optimizing the harness/context is the best way to make it act like we would, but this of course isn't working perfectly.

Re: Why does Opus 5 feel worse to work with?

#503

Post by fidotron »

D13Fd wrote: post #120

I’ve been doing some heavy work on a personal project lately. I burned through the limits on Claude, the plus a few hundred dollars in credits, and ultimately decided to move to an OpenAI account just so I can keep going. I was surprised to find that OpenAI Sol is much much nicer to work with than Opus 5 or Fable at the moment. Especially on Opus 5, the way it communicates is just exhausting. It keeps “being honest”…

It's so obnoxious it has to be deliberate, like a colleague from hell.

There's a constant strand from the AI safety brigade that "people get used to sycophantic LLMs which give them unrealistic expectations of human interaction" so Anthropic are overcompensating by making their models verging on antagonistic to deal with, so that we stay appreciative of our human brethren or something.

They seem to have forgotten they remain in a highly competitive market and they were merely top dog for a while. The enormous questions here are will people actually switch providers, and can Anthropic get back on track.

Re: Why does Opus 5 feel worse to work with?

#504

Post by jorl17 »

superdisk wrote: post #338
jbs789 wrote: post #244

Earlier quoted context omitted.

I found myself swearing at it more recently. It was going off today about having “shipped” something and I was like no… nothing has even been committed. And then it produced an incredibly verbose comment about hypothetical future changes. And all I could think was sure, let’s keep it short, or add a simple test that will break if that hypothetical becomes true. Or maybe I’m just more easily annoyed recently…

So true. The breaking point for me was when it was constantly saying to wrap up because we'd done enough work for the day, but we had barely even done anything. Or constantly estimating that the next steps would take X number of weeks, and then knock it out in a single prompt. I demanded in no uncertain terms to stop giving pointless bogus estimates or telling me to stop working, and it just wouldn't. I also had a pr…

This!!

I like to work weird hours of the night and Opus consistently likes to "wrap up" and say "it's been a long night" or "it's late" and "we've made great progress"

It's infuriating, just do the work!

Re: Why does Opus 5 feel worse to work with?

#505

Post by fastball »

barrkel wrote: post #12

The single biggest annoyance with Opus 5 is that it writes too elliptically. Sentences that orbit a point, then jump to it like it's a revealed insight. Unnecessarily abstract phraseology. Constantly using inanimate nouns as the subjects in sentences in order to unlock variety in verb choice, especially when it helps construct a sentence where the real action can 'land' like a surprise at the end. It is definitely mo…

In my experience your criticism of the writing is valid in all Claude 5 models, so I wonder if it is somehow related to the new tokenizer introduced in gen 5.

Re: Why does Opus 5 feel worse to work with?

#506

Post by fidotron »

argee wrote: post #455
40four wrote: post #355

Earlier quoted context omitted.

I hear you guys, but it sounds like we’re taking about the default settings or “personalities” baked into the models by their creators. Either of them will act exactly the way you want if you explicitly tell them too. Add the instructions to your own system prompt. If you don’t want a companion, say so. If you want shorter answers in a different style, tell them. They will obey :)

Read this. https://openai.com/index/where-the-goblins-came-from/ > We retired the “Nerdy” personality in March after launching GPT‑5.4. In training, we removed the goblin-affine reward signal and filtered training data containing creature-words, making goblins less likely to over-appear or show up in inappropriate contexts. Unfortunately, GPT‑5.5 started training before we found the root cause of the goblins. When we…

Did they ever discuss what the root cause of the goblins turned out to be?

Re: Why does Opus 5 feel worse to work with?

#507

Post by zamalek »

> Try as you might, it's nearly impossible to get the entirety of the context, intentions, business implications, budget constraints, and what-have-you written down and accessible to a coding agent. There will invariably be ambiguity and choices to be made, and it is nice to know that an agent will stop and ask when needed.

In general, I find that the grill-me prompt[1] helps with this - but I am definitely not hand-waving the complaints here. I feel like Anthropic peaked at around 4.5, and I have personal reservations about how far transformers can get us - but grill-me does a lot of legwork.

[1]: https://github.com/mattpocock/skills/blob/main/skills/produc...

Re: Why does Opus 5 feel worse to work with?

#508

Post by mrandish »

ValentineC wrote: post #283
stefan_ wrote: post #222

Earlier quoted context omitted.

CLAUDE.md is mostly powerless against the reinforcement learned crap. I'm up to three separate instructions telling it to cut out the hyper verbose, retelling history comments and it still writes them every time.

Claude Code has an "output styles" setting that supposedly directly modifies the system prompt: https://code.claude.com/docs/en/output-styles

I suspect the root problem is these issues aren't at the system prompt level, they're in the RHLF/fine-tune. And due to safety/jailbreaking fears, all prompt content and user-instructions are nerfed in priority.

Re: Why does Opus 5 feel worse to work with?

#509

Post by fidotron »

empath75 wrote: post #299
MyFirstSass wrote: post #24

I've gone back to 4.8. 5 would constantly veer of in random directions if not working from 100% strict and narrow instructions. I find it weird there's not more discussion here on HN on how the most used model now has clearly degraded in quality and it seems we've hit a peak and are on a downslope - because the model is clearly smaller or more economical for Anthropic no doubt about it, and the benchmaxxing they do i…

Opus 5 has a habit of taking what I asked for, doing something tangentially related to it, and then lying to me and saying it did exactly what I asked.

It even adds comments to say you asked for [thing] then leaves snarky comments about when you correct it.

At that point I decided it's just not worth the babysitting that's required, and you are better off working entirely with other models.

If the harness itself was open source then maybe we'd be able to wrap it up in a reasonable layer of sanity.

Re: Why does Opus 5 feel worse to work with?

#510

Post by mlsu »

Panoramix wrote: post #500
mlsu wrote: post #428

Earlier quoted context omitted.

Everything that claude writes fits into the same aesthetic structure. The aesthetic is that of an expert slowly revealing an insight to the user. The actual content doesn't matter. - "Introduction that rephrases your prompt." - "3 paragraphs, with one section of bullet points" - "The Twist" - "The Bottom Line" It's really obvious once you see it. Every single prompt, from a quantum physics question to a mundane obser…

You're right, and the load-bearing part of the argument is not what you think it is. Two ambiguities worth resolving before moving on: whether what you wrote also applies to ChatGPT, and whether you have custom instructions set up. Failure mode worth flagging explicitly: I didn't read TFA. (I'm becoming allergic to how these things write).

You are diabolical.
Post reply on HN