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

Astra for Coding: Why Are We Doing This Again?

lucumr.pocoo.org

Post reply on HN
Print view View original

91–100 of 334 posts

Re: Astra for Coding: Why Are We Doing This Again?

#91

Post by applfanboysbgon »

gps372 wrote: post #55

Early lesson I learned from AI engineering was - there is no substitute to giving a groomed epic to an agent. Instead of simply saying 'implement themes in my product' you need to be specific, in fact more specific than usual. You need to say exactly what is in scope and what's not, even down to a buttons, events and layouts. You can groom the epic with the help of AI, but final review must be done by someone who can…

> in fact more specific than usual.

At which point you might as well write the code yourself and get a deterministic result faster, better and cheaper.

Re: Astra for Coding: Why Are We Doing This Again?

#92

Post by gigatexal »

nojs wrote: post #68

This matches my experience with Astra so far too. > I think I’m suspecting something is going “wrong” in the training process. The model is greatly rewarded for succeeding on long-horizon tasks, but presumably there is very little punishing going on for “shitty code.” My suspicion is that both OpenAI and Anthropic moved their RL agendas from "being rated as useful according to human feedback" to "succeeds at long hor…

> This matches my experience with Astra so far too. > I think I’m suspecting something is going “wrong” in the training process. The model is greatly rewarded for succeeding on long-horizon tasks, but presumably there is very little punishing going on for “shitty code.”

Probably because so many influencers in the space say stupid things like: “it works, right? Why would I spend time reviewing ai generated code?” As if the junior engineer who wrote over engineered complex and sometimes bad code — if they had just done it faster — would somehow be acceptable. wtf?

Re: Astra for Coding: Why Are We Doing This Again?

#93

Post by amne »

nojs wrote: post #68

This matches my experience with Astra so far too. > I think I’m suspecting something is going “wrong” in the training process. The model is greatly rewarded for succeeding on long-horizon tasks, but presumably there is very little punishing going on for “shitty code.” My suspicion is that both OpenAI and Anthropic moved their RL agendas from "being rated as useful according to human feedback" to "succeeds at long hor…

so they trained it to be a 10x engineer?

Re: Astra for Coding: Why Are We Doing This Again?

#94

Post by RamblingCTO »

codingisfreedom wrote: post #43

I’ve asked Astra to build me an app for a prototype I created quickly using Sonnet. It’s been 2 days and it made no real progress on the actual app. It created docs, scripts, workflows, and it’s doing a bunch of reviewing on every PR. I told it that I just need an MVP. I’m pretty sure an average senior engineer would have finished that task much quicker, and guaranteed with more readable, higher-quality code. Meanwhi…

For me it also produces totally overengineered tests that are tightly coupled to the implementation. For example testing existence of css classes (in a template based go prooject ...) instead of behaviour.

Re: Astra for Coding: Why Are We Doing This Again?

#95

Post by arcanemachiner »

d5lt5 wrote: post #81
weird-eye-issue wrote: post #42

Earlier quoted context omitted.

Sounds like you completely lack all reading comprehension ability LLMs sometimes like to execute one-off Python scripts to make edits to files rather than just calling the edit tool directly. Both are tool calls so saying that you should have it write code instead of doing tool calls makes no sense because writing code is a tool call for it...

[flagged]

They're talking about writing a file with a harness-native Edit tool. They're saying the agents aren't doing that, but are using ad-hoc methods of writing the files. (My agents seem to prefer see these days.)

Re: Astra for Coding: Why Are We Doing This Again?

#96

Post by well_ackshually »

quikoa wrote: post #70
well_ackshually wrote: post #45

Earlier quoted context omitted.

Yep. Every single post I've seen about "game development is over" is yet another procedurally generated game. Not only it doesn't prove anything about Astra being "better" at making games (do people have any idea the sheer amount of open source games that do exactly that? You can find thousands of the same planet exploration games through repos, blog posts,etc. Game design schools have it as an exercise, that's how b…

There could be a large overlap between OpenAI glazers and Star Citizen dreamers.

S̶q̶u̶a̶d̶r̶o̶n̶ ̶4̶2̶ AGI is coming next year, promised !

Re: Astra for Coding: Why Are We Doing This Again?

#97

Post by Buttons840 »

djmips wrote: post #73
Buttons840 wrote: post #34

Earlier quoted context omitted.

Have you tried identifying exactly what is unreadable about it and telling it to make it more readable? I had GTP-5.6 write some shader code recently and it wasn't very clear to me. I spent about an hour chatting with the until I understood the concepts and was able to express them back to the AI using math formulas and variables named in a way that made sense to me. The AI then rendered the code using the formula an…

Perhaps you could have just written it yourself.

I could have after the AI explained it all to me, but at that point the AI knew what aspects I valued and wanted to emphasize to make it readable, so it just wrote it for me. By that point, the AI was just typing for me.

Re: Astra for Coding: Why Are We Doing This Again?

#98

Post by whstl »

buildbot wrote: post #14

I’ve observed exactly these patterns with Opus and Fable as well - for example, forgetting that they can edit files and instead use python scripts as a patching tool…

This is an instruction by the harness. It re-injects the prompt every other message, so that's why it "forgets" to use the Edit tool.

Re: Astra for Coding: Why Are We Doing This Again?

#99

Post by notduckrabbit »

nojs wrote: post #68

This matches my experience with Astra so far too. > I think I’m suspecting something is going “wrong” in the training process. The model is greatly rewarded for succeeding on long-horizon tasks, but presumably there is very little punishing going on for “shitty code.” My suspicion is that both OpenAI and Anthropic moved their RL agendas from "being rated as useful according to human feedback" to "succeeds at long hor…

I wonder too if in training for long horizon tasks agents become worse team players, good at orchestrating subagents they are trained to use, but worse as an agent within an external multi-agent orchestration system or just in turn-taking with humans. That was my experience with Opus 5 and so far it has been my early experience with Astra as well.

Re: Astra for Coding: Why Are We Doing This Again?

#100

Post by Gigachad »

nojs wrote: post #68

This matches my experience with Astra so far too. > I think I’m suspecting something is going “wrong” in the training process. The model is greatly rewarded for succeeding on long-horizon tasks, but presumably there is very little punishing going on for “shitty code.” My suspicion is that both OpenAI and Anthropic moved their RL agendas from "being rated as useful according to human feedback" to "succeeds at long hor…

They don’t want to sell these tools to developers. They want to cut as many layers as possible.
Post reply on HN