Comment by xantronix | Hacker News Reader

Comment by xantronix | Hacker News Reader

Parent Full thread

I know there are good uses of LLMs out there. I do. But.

The current fever pitch mandates from above seem to want it applied liberally, and pushing back against that is so discouraging and often career-limiting as to wear the fabric of one's psyche threadbare. With all the obvious problems being pointed out to people, there are just as many workarounds; and these workarounds, as is often revealed shortly thereafter, have their own problems, which beget new solutions, ad infinitum.

At some point it genuinely seems like all this work is for the sake of the machine itself. I suppose that is true: The real goal has become obscured at so many firms today, that all that remains is the LLM. Are the people betting the farm and helping implement the visions of those who have done so guaranteed a soft exit to cushion them from the consequences, or is rationality really being discarded altogether?

Sure, sound engineering principles can help work around these problems, but what efficiency is truly gained, in terms of cognitive load, developer time, money, or finite resources? Or were those ever an earnest concern?

The dirty secret if you work inside BigCorp and look around at the projects they're showcasing:

1. They're low stakes to get wrong.

2. The most common is MCPs or similar ai-tooling.

3. Making them look good takes time and effort still. It's a multiplier, not a replacement.

4. Quality and maintainability require investment. I had to restart an agentic project several times because it painted itself into a corner.

In my opinion you are just wrong.

It’s an absolute game changer, and it can now multiply your productivity fivefold if it’s a solo greenfield project.

Maybe half a year ago it was as you said. You had to wait for the agent to finish, you had to review carefully, and often the result was not that great. You did not save a lot of time.

Now I can spin up 3+ parallel conversations in Codex, each in a git worktree. My work is mainly QA testing the features, refining the behavior, and sometimes making architectural decisions.

The results are now undeniable. In the past I could not have developed a product of that scope in my free time.

That is what is possible today. I suspect many engineers have not yet tried things that became feasible over the last months. Like parallel agents, resolving merge conflicts, separating out functionality from a large branch into proper PRs.

"many engineers have not yet tried things that became feasible over the last months"

I have heard this statement every single day for 2 years and yet we still have no companies compressing 10 years into 1 year thus exploding past all the incumbents who don't "get it".

Well, the GP mentioned

> if it’s a solo greenfield project

which is a pretty large caveat. Anecdotally, I've found my side projects (which are solo greenfield projects, and don't need to be supported to the same standards as enterprise software) have gained the boost the GP was talking about.

At work, it's different, since design, review, and maintenance is much more onerous.

If you want an example of a project that condensed 5 years into 6 months and exploded past the competition I suggest looking at OpenClaw.

The first line of code was written on November 25th. It achieved adoption in the "personal agents" space that far exceeded the other companies that had tried the same thing.

(Whether or not you trust the quality of the software you can't deny the impact it had in such a short time. It defined a new category of software.)

Ideally, the given example would be something not ajacent to the presently white-hot category of "AI agents".

Like, look at e.g. YC minus the AI and AI ajacent companies. Are those startups meaningfully more impressive or feature-rich as compared to a couple years ago?

Not yet, no. I think that's because coding agents got good in November, most people didn't notice until January and it still takes 3-4 months to go from idea to releasing something.

I expect we will start seeing the impact of the new coding agent enhanced development processes over the next few months.

Its trash vibecoded markdown files around pi. This exemplifies well what op’is saying. We are at the ICO stage of llms. Hopefully there wont be an nft one

As much as I love to hate on AI: even the bad apples still produce something that one can reasonably work with.

Cryptocurrencies? Barely any other use than money laundering, buying drugs and betting on the outcome of battles in war. And NFTs? No use at all other than money laundering and setting money ablaze.

Privacy and security from government overreach is not enough?

What privacy? Enough drug dealers have already been busted with solid evidence from trailing the paths on public blockchains.

crypto is a few 100 of billions less big as an industry than GenAI is. I guarantee you that AI is a far better money laundering scheme. mb the two better money laundering schemes would be construction business and the global warming business. doesnt mean that some of the stuff produced is good.

OpenClaw is definitely not a "5 years" project pre-AI though. That was more like a month of greenfield work compressed into a weekend -- which is still really impressive, don't get me wrong! -- but I think the point is we're not seeing mature, legacy codebases get outcompeted by new, agile, AI-driven codebases; we're seeing greenfield projects get spun up faster. Which, again, is still impressive and valuable.

If agents could really compress 10 years of development into 1 year, you'd see people making e.g. HFT platforms and becoming obscenely rich, not making a fun open-source project and getting hired by OpenAI as an employee.

You're framing it like the only barrier to writing wildly successful money printing software is software development skills.

If that were true, all of these anti-AI greybeards who have been in the game for 30 years would all own their own jets.

41,964 commits is a lot more than "a month of greenfield work".

https://tools.simonwillison.net/github-repo-stats?repo=OpenC...

Seriously? Commit count is right up there with lines of code as a classically dumb measurement of productivity.

Sure, but it's still a good counter to "a month of work".

No it isn't. There's basically no upper bound on the number of commits an LLM can generate. If the LLM takes 10,000 commits to do what a human would do in 10, then the comparison is meaningless.

I don't know anything about the code quality of OpenClaw, but telling me the number of commits tells me precisely nothing of use.

OK, now do that for 369,293 stars, 76,193 forks, 138 releases and 2,133 contributors.

I expect there is no number I could bring up here that won't be instantly shot down as telling "precisely nothing". My mistake for bringing up any numbers at all.

OpenClaw is a good example of a completely new project written using coding agents that made a significant impression on the world and would not have been built without them.

I'm surprised this is a hill I have to die on, but there we are.

(I'm not even a user of OpenClaw! I don't think it's secure or safe enough to use in my own life.)

> OK, now do that for 369,293 stars, 76,193 forks, 138 releases and 2,133 contributors.

You're counting forks and stars as code metrics now? Oy.

Look, those aren't nothing -- they're a decent enough proxy for popularity -- but they aren't a rebuttal to the original comment. (The other day some LLM dudebro got a bajillion stars on GH for his vibe-coded hot mess of a repo that sets three environment variables. I should go check the number of commits on that...)

> OpenClaw is a good example of a completely new project written using coding agents that made a significant impression on the world and would not have been built without them. I'm surprised this is a hill I have to die on, but there we are.

The fundamental problem here is that you were asked to provide an example of some software where LLMs have made a revolutionary difference, and OpenClaw is what you chose. That just says a lot, right there.

I don't even really care about that debate, since OpenClaw probably meets the literal requirements of the original question (if not the spirit), and sure, it's had a big splash. But the point of the OP is well-taken: everyone is so "productive", but if the only thing we're seeing from it is Moltbook and 10,001 half-broken pokemon games, then eventually the bloom is going to fall off the rose.

The fact that you felt you had to rebut the "I could do that in a weekend" guy with commit counts is both poetic and oddly fitting for where we are with these things.

I stand by what I said. OpenClaw proved that "personal digital agents" are a category with a huge amount of demand, to the point that people will jump through major hoops and completely ignore the colossal security risks involved in adopting that software.

It's spawned dozens of imitations, some of which are looking quite credible.

Anthropic themselves have been cloning OpenClaw features.

I get that it's not cool to say "OpenClaw is significant and influential" but I truly believe that it is.

> OpenClaw is a good example of a completely new project written using coding agents that made a significant impression on the world and would not have been built without them.

Nobody is denying that OpenClaw is popular, and nobody (in this thread, at least) is denying that AI rapidly speeds up the ability to make an initial release or prototype for greenfield projects. But the comment that spawned this discussion was:

> we still have no companies compressing 10 years into 1 year thus exploding past all the incumbents who don't "get it".

The issue is that you're extrapolating OpenClaw, which upon release was a month of pre-AI development work compressed into a few days, to cover the "10 years into 1 year" scenario. However, this isn't appropriate because software development is non-linear. As anyone who has worked on a greenfield project pre-AI should know, those first weeks and months have much faster development cycles. There's no tech debt to worry about; there's no urgent bug tickets or feature requests from customers; there's no thinking about whether it's okay to ship a breaking change.

It isn’t man. Anyone can easily split a single good commit into 10 just to inflate the numbers. C’mon, this is 101 git

> 41,964 commits is a lot more than "a month of greenfield work".

I meant a month for the initial release, not current state.

Regardless, much like lines of code, number of commits is not a good metric, not even as a proxy, for how much "work" was actually done. Quickly browsing there are plenty[0] of[1] really[2] small[3] commits[4]. Agentic coding naturally optimizes for small commits because that's what the process is meant to do, but it doesn't mean that more work is being done, or that the work is effective. If anything, looking at the changelog[5] OpenClaw feels like a directionless dumpster fire right now. I would expect a lot more from a project if it had multiple people working on it for 5 years, pre-AI.

[0] https://github.com/openclaw/openclaw/commit/e43ae8e8cd1ffc07...

[1] https://github.com/openclaw/openclaw/commit/377c69773f0a1b8e...

[2] https://github.com/openclaw/openclaw/commit/ffafa9008da249a0...

[3] https://github.com/openclaw/openclaw/commit/506b0bbaad312454...

[4] https://github.com/openclaw/openclaw/commit/512f777099eb19df...

[5] https://github.com/openclaw/openclaw/blob/main/CHANGELOG.md

That's why my original comment said:

> (Whether or not you trust the quality of the software you can't deny the impact it had in such a short time. It defined a new category of software.)

I brought up OpenClaw here because the challenge was:

> we still have no companies compressing 10 years into 1 year thus exploding past all the incumbents who don't "get it".

Didn’t we learn anything from the past? Using loc or number of commits or github stars to measure success or productivity is so backwards. It seems everyone on the AI wagon is either young (and so they don’t know our history) or simply forgot about all the good practices in software engineering

That latter bit is my experience. As soon as AI enters the equation, we have to immediately ignore everything we ever learned and just type text into the prompt box, or you're not doing it right.

My bashscript can do that in some hours. The git repo contains no working software after that, but when that is what you want to meassure...

I don’t see OpenClaw making much of an impact. Maybe in your bubble?

There are credible reports of regular people in China attending dedicated events for help getting started with OpenClaw. They're not in my bubble!

https://www.reuters.com/technology/openclaw-enthusiasm-grips...

> It defined a new category of software

Which is exactly why you can't use it as an example, there is no control. This is basic stuff.

The condensation argument is totally true.... Strikes me though the other metric Id look at is how long code survives before being re-written. Feels like for that one a bit early to tell...

> if it’s a solo greenfield project

That's a big if. I don't have numbers but most professional engineers are not working on such projects

The thing is, I don't care any longer. I sincerely believe velocity without direction is not a good strategy for delivering quality in the long term. And that's the thing about it: How sustainable is this velocity, in terms of socioeconomic concerns, product strategy, and mental health?

Velocity without direction?

I‘m personally directing and QA testing every feature.

I don’t know how socioeconomic concerns, product strategy, and mental health are a concern for me here.

I‘m having a great time with my project and it’s been the most fun I‘ve had in many years of building.

> and it can now multiply your productivity fivefold if it’s a solo greenfield project.

Why do I not see 5x as many interesting greenfield projects than before?

All of the "solo green field projects" I let LLMs mostly write, despite supplying the scaffolding, structure and specific implementation details as code, prompts or context, I can't tell you much about 6+ months later, except for the parts I did write.

It's like I never wrote them, because I didn't. I've got the gist of them, but it's the same way I get the gist of something like Numpy: I know how it works theoretically, but certainly not specifically enough to jump in and write some working Fortran that fixes bugs or adds features.

I now have a bunch of stalled projects I'm not very familiar with. I no longer do solo green field projects that way.

There's two sides to the AI mandates.

The degenerate side is clueless upper management and fad-driven engineering. We have talked extensively about this.

There is a more rational side to it that I've seen in my org: some engineers absolutely refuse to use AI and as a consequence they are now, clearly and objectively, much less productive than other engineers. The thing is, you still need to learn how to use the tool, so a nontrivial percentage of obstinate engineers need to be driven to use this in the same way that some developers have refused to use Docker or k8s or whatever.

Ah yes, we must force these obstinate engineers to the right path! Only after getting everyone to see the light will they understand and thank us for boundless productivity!! /s

Perhaps these “obstinate” engineers have good reason in their decision. And it should be their decision!

To be so confident in what is “the right way (TM)” and try to force it onto others is... revealing.

Engineers that didn't move past src.v35.final.zip version control don't really have jobs today, either.

You would be absolutely shocked how many software projects are still run, to this day, without source control at all. Or automated (or manual) testing. And how many hand crafted artisanal servers are running on AWS, never to be recovered if their EC2 instance is killed for some reason.

Sure, but that’s a small and shrinking market. Not a source of economic security or growth for its employees, nor for most of its companies (though some have defended niches).

I've seen growing companies running multiple million ARR through systems like that. It's way more common than you'd think if you're a professional software developer.

I seriously don't see how version control and LLMs are comparable. A deterministic way to track code changes over time, versus an essentially non-deterministic statistical code generator that might get you what you want, and might do it in a reasonable time frame, and that might not land you in a minefield of short-term-good/long-term-bad design points.

> an essentially non-deterministic statistical code generator that might get you what you want, and might do it in a reasonable time frame, and that might not land you in a minefield of short-term-good/long-term-bad design points.

Sounds like a human? The ‘statistical’ part is arguable, I suppose.

There is an absolute embarrassment of modern tooling in other categories I have no problem whatsoever embracing. I'm not a holdout for being stuck in my ways. Maybe I value things other than expediency at massive cost. Maybe I speak just as well to computers as I do to humans.

I'm sure I will have no problem whatsoever remaining in the employ of a firm that trusts me to make products and tooling that still push the envelope of what's possible without having to resort to the sheer brute force of trillion parameter-scale models.

There is no massive cost. For 80% of the brute work that needs to be done day in and day out LLMs provide code as good as a senior engineer provided you have sufficient competency in steering the model, but done at breakneck pace.

I ran the statistics myself and my company is spending 40% less time doing feature development since AI agents began to be used en masse and pushing 50% more tickets without any noticeable increase in regressions.

After 18 months the hard evidence is in place. And much like replacing bare-metal servers for many use cases where evidence shows that the burden of k8s or the substitution of shell scripts for Terraform, it's time to move on.

I don't really see a place for no AI usage in line-of-business software apps anymore.

What did you use to fill the time you aren't doing feature development in with? Or are you all now working 20 hour work weeks?

Faster feature development, more strategic thinking in how to keep the dev pipeline full, doing braindead mechanical improvements that pay off tech debt that would have otherwise not have management sign-off to justify, writing GUI-based tools for support teams that previously had to scour reams of shell scripts, spending more time on refining specifications and estimations, writing throwaway concepts of different design ideas in order to have better architetuce discussions based on real code instead of pseudocode, clearing out the backlog of bugs that used to be terribly annoying to reproduce and that now I can just throw brute compute for resolving.

Sounds awful. Just filling the time with worthless stuff. You are basically a liability. Wouldn’t like to have you in my team. Less is more (nowadays more than ever)

Sounds like you don't read and you don't understand what adds value in a engineering team.

Around the turn of the century there were the same exact arguments being made about automated testing (not just TDD, but any automated tests at all!)

Reply on news.ycombinator.com