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

Eight more months of agents

crawshaw.io

Post reply on HN
Print view View original

101–110 of 250 posts

Re: Eight more months of agents

#101

Post by girvo »

estimator7292 wrote: post #51
bdangubic wrote: post #41

Earlier quoted context omitted.

the number of experienced, senior programmers though, who are in “anti-LLM” camp, is still fairly staggering.

I mean when the tag line is "this will replace senior engineers and you, the senior engineer, must be forced to use it" Then yeah, it makes sense.

Yeah I’m baffled why people are surprised that senior+ engineers who are being told in one breath they will be replaced by this tool and also they MUST use this tool to make it better to replace them aren’t happy about it or want to use it willingly.

I also find it wild how we’re sleepwalking into this, but I’m also part of the problem and using these things too.

Re: Eight more months of agents

#102

Post by gip »

loveparade wrote: post #86
sarchertech wrote: post #84

Earlier quoted context omitted.

I’m not saying this is definitely a bot. However, this is the 7th time I’ve read a post and thought it might be an OpenAI promotion bot, clicked on the username, and noticed that the account was created in 2011. I have yet to do this and see any other year. Was there someone who bought a ton of accounts in 2011 to farm them out? A data breach? Was 2011 just a very big year for new users? (My own account is from 2011)

It's definitely a bot, just like probably around 10% of comments on HN at this point, and the majority of upvotes. And it's only increasing. Calling it bot is a bit dismissive though. It's an agent!

Care to have a phone call with who you call a bot tonight?

If so, send a DM on twitter to @edfixyz with your phone number and I will call you immediately. Or give me your twitter handle.

I'm tired of that BS - when people don't like what you write they call you a bot.

Re: Eight more months of agents

#103

Post by girvo »

giancarlostoro wrote: post #61
blibble wrote: post #46

Earlier quoted context omitted.

> We all write insecure code in our first pass through no, we don't

Yes we do, you don't just start a brand new web project and spit out CORS rules, authentication schemes, roles, etc in one sitting do you? Are you an AI?

Yes I really do, because this has been a solved problem for a while. Also it’s necessary to get right because retro fitting it later is a pain.

Re: Eight more months of agents

#104

Post by sarchertech »

gip wrote: post #92
sarchertech wrote: post #84

Earlier quoted context omitted.

I’m not saying this is definitely a bot. However, this is the 7th time I’ve read a post and thought it might be an OpenAI promotion bot, clicked on the username, and noticed that the account was created in 2011. I have yet to do this and see any other year. Was there someone who bought a ton of accounts in 2011 to farm them out? A data breach? Was 2011 just a very big year for new users? (My own account is from 2011)

I'm not a bot. You are saying that because for some reason you resent people who have a good experience with Codex / OpenAI. Curious what that is - people hate the CEO or what? I like Claude Code too btw. The crazy thing here is that I wrote the initial comment myself!

> It's a data point but this weekend (e.g. in 2 days) I build a desktop + web agent that is able to help me reason on system design and code. Built with Codex powered by the Codex SDK. It is high quality. I've been a software engineer and director of engineering for 10 years. I'm blown away.

Assuming you’re not a bot. It’s nothing to do with you having a good experience, it’s the way you wrote about that experience that sounds like a product placement.

I asked OpenAIs very own ChatGPT 5.2 powered by OpenAI to tell you why it sounds like a product placement:

“ Because it hits a bunch of “native ad / testimonial” tells at once: • Brand-name density in a tiny space. “Built with Codex powered by the Codex SDK” repeats the same brand in two adjacent phrases, like copy that’s trying to lodge a name in your head rather than naturally describe a build. • Overly polished value signals. “High quality” is a generic superlative with no concrete evidence (features, metrics, constraints, tradeoffs). Ads often lean on verdict words instead of specifics. • Credential + astonishment combo. “I’ve been a software engineer and director of engineering for 10 years” is classic authority framing, immediately followed by “I’m blown away.” That’s a common testimonial structure: I’m hard to impress → I’m impressed. • Time-compressed “miracle build” narrative. “This weekend (in 2 days) I build a desktop + web agent…” reads like the “you can do it fast/easily now” story arc you see in promos. Not impossible—just a familiar marketing shape. • “It’s a data point” language. That phrase feels like social-proof seeding: “don’t treat this as hype, just one datapoint,” which paradoxically makes it feel more like deliberate persuasion. • No friction or downsides. Real engineer excitement usually includes at least one caveat (bugs, rough edges, limitations, cost, setup pain). The total absence makes it sound curated. • Benefit phrased like positioning. “Able to help me reason on system design and code” is basically a product pitch line (target user + problem + outcome) rather than a personal anecdote (“it helped me untangle X design and refactor Y”).”

Re: Eight more months of agents

#105

Post by baq »

happytoexplain wrote: post #55
derefr wrote: post #25

Earlier quoted context omitted.

The negative impacts of generative AI are most sharply being felt by "creatives" (artists, writers, musicians, etc), and the consumers in those markets. If the OP here is 1. a programmer 2. who works solely with other programmers and 3. who is "on the grind", mostly just consuming non-fiction blog-post content related to software development these days, rather than paying much attention to what's currently happening…

>"Anti-LLM sentiment" within software development is nearly non-existent This is certainly untrue. I want to say "obviously", which means that maybe I am misunderstanding you. Below are some examples of negative sentiments programmers have - can you explain why you are not counting these? NOTE: I am not presenting these as an "LLMs are bad" argument. My own feelings go both ways. There is a lot that's great about LLM…

> 4. Expertise loss

> A lot of positive outcomes with LLMs come from being already experienced. Some argue this will be eroded - both for new devs and existing experienced devs.

This is true, but the pace of progress is so mind blowing the experts we have now might just be enough until the whole industry becomes obsolete (10-20 years assuming the lower bound of the trend line holds?)

Re: Eight more months of agents

#106

Post by xyzsparetimexyz »

> Pay through the nose for Opus or GPT-7.9-xhigh-with-cheese. Don't worry, it's only for a few years.

> You have to turn off the sandbox, which means you have to provide your own sandbox. I have tried just about everything and I highly recommend: use a fresh VM.

> I am extremely out of touch with anti-LLM arguments

'Just pay out the arse and run models without a sandbox or in some annoying VM just to see them fail. Wait, some people are against this?'

Re: Eight more months of agents

#107

Post by dmos62 »

gip wrote: post #80

> In 2026, I don't use an IDE any more. I don't think it is the best way to look at it. I think that now every team has the power to build and maintain an internal agent (tool + UX) to manager software products. I don't necessarily think that chat-only is enough except for small projects, so teams will build agent that gives them access to the level of abstraction that works best. It's a data point but this weekend (…

Curious what kind of agent did you build? I'm building a programming agent myself, it's intentionally archaic in that you run it by constantly copy-pasting from-to fresh ChatGPT sessions (: I'm finding it challenging to have it do good context management: I'm trying to solve this by declaring parts of code or spec as "collections" with an overview md file attached that acts like a map of why/where/what, but that can't scale indefinitely.

Re: Eight more months of agents

#108

Post by baq »

bitwize wrote: post #68
derefr wrote: post #25

Earlier quoted context omitted.

The negative impacts of generative AI are most sharply being felt by "creatives" (artists, writers, musicians, etc), and the consumers in those markets. If the OP here is 1. a programmer 2. who works solely with other programmers and 3. who is "on the grind", mostly just consuming non-fiction blog-post content related to software development these days, rather than paying much attention to what's currently happening…

Facebook's algorithm has picked up on the idea that "I like art". It has subsequently given me more examples of (human-created) art in my feed. Art from comic book artists, art from manga-style creators, "weird art", even a make-up artist who painted a scene from Where the Wild Things Are on her face. I like this. What's more, while AI-generated art has a characteristic sameyness to it, the human-produced art stands…

I don’t code anymore basically, but I have no issues accepting that there are huge gaps in the training sets in e.g. finance code bases. There is a long tail of niches to still code (semi) manually in.

Re: Eight more months of agents

#110

Post by gip »

dmos62 wrote: post #107
gip wrote: post #80

> In 2026, I don't use an IDE any more. I don't think it is the best way to look at it. I think that now every team has the power to build and maintain an internal agent (tool + UX) to manager software products. I don't necessarily think that chat-only is enough except for small projects, so teams will build agent that gives them access to the level of abstraction that works best. It's a data point but this weekend (…

Curious what kind of agent did you build? I'm building a programming agent myself, it's intentionally archaic in that you run it by constantly copy-pasting from-to fresh ChatGPT sessions (: I'm finding it challenging to have it do good context management: I'm trying to solve this by declaring parts of code or spec as "collections" with an overview md file attached that acts like a map of why/where/what, but that can'…

Send an DM on twitter to @edfixyz (one of my account) and I'll reply with a link to the website tomorrow to give you a sense. Can't share a link here, it will kill my backend.
Post reply on HN