This x 10 . I don’t understand how people are saying you can’t use LLMs to get crazy productivity gains. If you can’t write quality code with LLMs at ludicrous speed, you’re holding it wrong. You will have occasional bad days and regressions. But overall you’re still going to be able to 4x your progress.
That’s kinda what I mean. Maybe it only works well in some languages, but with the harness I built for C and C++ does a fantastic job of adhering to very strict architecture and style guides. Way cleaner, more readable, better factored, and more interpretable than human generated code, except maybe one or two devs I have worked with. YMMV I guess?
TBF I do burn 200k tokens just preloading the context with onboarding, not including any code, just document trees of development policy documents, style and architectural standards, code and documentation review processes, company ethos and culture, etc. it’s a token fire, but it really works for us.
Also, documentation driven development all the way down.
If you're an enterprise (including startups), you worry about customers, not code quality. There are famously many startups that gained traction despite shit code and then eventually got around to fixing it, to whatever extent was possible, like Facebook HHVM, Stripe's Sorbet, etc.
> Startups failed because they cound not untangle own code after 4 months.
That's rare, though. If they could not untangle their own code after 4 months, it's because they were not making enough money to pay a team to untangle it - that's not a code problem, it's a revenue problem.
IOW, the startup failed because their revenue was too low.
There are orders of magnitude that failed because they did not solve the right customer problem. Code quality is merely incidental the vast majority of the time.
Ok, and? You can live with that if there are more important things to deal with.
I've stared at ugly LLM code, that I had just had generated, and worked well enough for my purposes. (generally, some quick recursion into a nested python dictionary in order to dig out some property -- especially for linting or quick data analysis).
And I wanted something better, sure, something a bit more readable ...but I just needed it to work well enough to recurse through a yaml file for config file linting, not be battle-hardened against every test case.
So to deal with the mess, I shoved it in a pure function, threw a few basic sanity unit tests around it, put a comment with a disclaimer of "#this is LLM generated code, it is lightly tested, do not use it for anything truly load-bearing without a lot more tests" and I moved on to something else.
You're on Hacker News. This is a site full of developers who are convinced that "proper software engineering" is 100% of what makes a business successful, and everything and everyone else is useless. You can't just waltz in here and point out that code in business is a means to an end and expect not to get downvoted.
> As a technical product manager, this 1000%. It's just irrelevant how bad code is unless it impacts the business.
If you are, in fact, "a technical product manager", I would hope you understand that "bad code" is identified as such specifically because it "impacts the business."
The engineers I have worked with most definitely define "bad code" as having intrinsic limitations and/or latent defects which impact successful system functionality/operation. Indicators provided to stakeholders such as yourself which support this assessment are, but not limited to:
- the system doesn't work that way
- the system lacks test coverage, so changes take longer
- adding feature "X" is not feasible
- there is no repeatable way to onboard team members
- the backlog grows exponentially
- that "one point task" is going to take a couple weeks
All of the above impacts a business.
It is up to you, the "technical product manager", to understand what your team is trying to tell you.
Please stop being rude to me. I'm a human being, I'm a very experienced product manager and engineer (you can google my name, I'm the only one), and the way you are behaving sucks.
Everything you're saying is true, sometimes. Assume I'm still right, and that you might be able to learn something from someone else.
I do not see how I was being rude, unless it was my use of quotations around the title you claim.
> I'm a human being ...
I did not doubt this.
> ... I'm a very experienced product manager and engineer ...
Again, if it was my use of quotations which you found to be rude, then I do not know what to say about that.
> ... and the way you are behaving sucks.
I respect your perspective and support your right to express yourself. And no, I do not think you are being rude by doing so.
> Assume I'm still right ...
Why would I? You responded to:
>> This is a site full of developers who are convinced that "proper software engineering" is 100% of what makes a business successful, and everything and everyone else is useless.
With:
> As a technical product manager, this 1000%.
Finally, you write:
> ... you might be able to learn something from someone else.
Maybe you can learn something from someone else as well.
They weren’t rude enough. Your complete apathy towards the many antisocial effects of badly engineered software, caring only about increasing shareholder value, is the reason why modern software not only sucks but actively makes our lives worse to use it.
So you've got this low code quality that somehow doesn't impact product quality but at the same time makes someone's life worse? Can you give me a concrete example? Help me understand. I'm coming at this from having been in test for a long time before product, I care quite a bit about cquality, so it's really fascinating to see you respond this way.
> So you've got this low code quality that somehow doesn't impact product quality ...
Low quality code always impacts product quality in a negative manner. That is why it is "low quality code."
"High quality code" also impacts product quality, but in the predictably positive manner; fewer defects, ability to be enhanced, understandable behavior, etc.
Of course it impacts product quality, in fact that's really what matters, code quality is just a useful proxy for that since they're so highly correlated.
No, you made the point of saying that anything which doesn't affect the business is unimportant. Looking at the quality of software products by the largest and most successful big tech companies, it appears quality is not important. You can see how this is contradictory.
It’s hardly rude, let alone “extremely” rude. People who are transgender are straight up told to kill themselves daily on social media by bigots and you’re melting down over someone having the temerity to question some random account’s (ie, you) credentials? Grow some thicker skin.
I think having the audacity to call “extremely rude” something as benign as barely questioning an anonymous account’s credentials is the definition of melting down.
> It's definitely rude to suggest someone is lying.
Please consider the context of my quotation which I documented elsewhere in this thread.
A person wrote:
>> This is a site full of developers who are convinced that "proper software engineering" is 100% of what makes a business successful, and everything and everyone else is useless.
To which the person claiming said managerial title responded with:
This is something I wish I understood sooner. There is strong merit to "good enough".
Of all the "concise" and "beautiful" code I worked hard to produce, I was the only one to ever lay eyes on it. It didn't actually matter, and nobody cared but me. The people in charge of my raises could never perceive quality of code, because it wasn't their area of expertise. They only cared (rightly so) that it did what it was supposed to, and all the elegant abstractions didn't practically help that purpose. It was, literally, wasted life that I should have spent just getting off work early, like most of my colleagues.
Every single thing around you was pioneered by people who are dead and forgotten. From the materials science of the clothes you wear, to the very language you speak.
Get over yourself. We're all ephemeral, dead and recycled in the blink of an eye. Our species doesn't even clock on the geologic timespan.
If you think your code (or any of your artifacts or possessions) matter beyond their immediate utility, you're mistaken. Work will either fall into disuse or be replaced. It's scaffolding for what comes next along a well-traversed path.
I refuse to accept your existential nihilism. This mindset is not only toxic to the soul, but toxic to those who must suffer the effects of someone who only cares about “immediate utility”. What a depressing comment.
It's ironic because around 20 years ago here, people knew HN was (more) explicitly for startup founders and the comments reflected that, with much more discussion on getting customers than writing code.
I measured an ~8x increase in my project's commit count after AI, and I'm painstakingly reading, reviewing, understanding and editing everything the models write. It's gotten to the point I'm trying to slow down in order to let the new knowledge crystallize. I'm manually writing articles about what I'm doing as I go.
I can only imagine what people are doing at their jobs with unlimited token budgets.
Yes, but my field has not been hit by the AI frenzy yet. Outside the usual attempts to automate us, that is. I've used AI at work for research and corroboration but it hasn't led to 100x performance or anything of the sort.
In my case it was commits, not lines of code. I wasn't chasing after it, I just asked Claude to calculate some statistics after a month or so of AI usage.
It's not just statistics either. I know for a fact that I made major progress by using LLMs. Here's a summary from around a month ago:
I wonder if the people getting 10x productivity gains are spending less time on HN and more time tending to their agents. Personally I now spend so much time productively arguing with agents that it feels like an utter waste of effort arguing with humans, if people can't see the value in LLMs by now I'm not sure what I could say to change their minds.
So you are accomplishing a year’s worth of work in a month? If that’s been happening for a few months, you must have a few years of work to show people right?
I actually do, I completely and elegantly solved a core problem of my niche industry that had been plaguing it for over 40 years. I was already planning on doing it and had put aside 4-6 months to do it. I did it with the help of AI in 3 days. The AI couldn't do it on it's own I had to walk it through the process piece by piece, and it kept trying to avoid doing as I asked by lying to me, so I kept chastising it. It eventually relented and it worked. There was a deep theoretical underpinning which I believe was beyond AIs ability to originate but once it understood what I was doing it was great at the mechanics of it. I've tried to explain similar ideas to humans with much less success.
Now I've massively expanded my scope and I'm still worried I'll run out of work by the end of the year. It's so much fun. Previously I would have to try to convince people with money to fund such applied research and often the strings they'd attach would doom it to failure. I was worried I would never have the resources to test out my ideas in my lifetime and I would die not knowing if they were any good, and since one of the big ones has already worked I'm pretty stoked.
Definitely enjoying the lack of eye-rolling, being asked to explain obvious things multiple times, and stopping things being done for resume-stuffing reasons.
There's a small minority of people who are adamantly refusing to change, such as there are in every technological revolution. Ego prevents them from even wholeheartedly trying the tool, because it would be admission they were wrong.
The opportunities available for these people are rapidly, rapidly shrinking. I believe it's possible to be a developer today who's EXCEPTIONAL and never uses AI. Most opponents are not exceptional, though, and even these opportunities are shrinking.
Most exceptional developers in my org adopted AI in their workflows and went from 10x developers to 20x developers.
If you refuse to adapt, you're going to be out of a job complaining about the kids and their newfangled technology REAL quick. You have a few years remaining, maybe less.
I can’t turn 10x work into 20x work because I have to ensure the two juniors in my team who are now creating 50x work won’t merge complete garbage, reviewed by another engineer that has already given up on caring.
I can’t turn 10x work into 20x work because my Product Manager thinks changing fundamental premises of tasks I already spent two weeks on (mostly removing human blockers) is very simple. After all, when he asked Claude to update his prototype, it only took it 10 minutes.
I can’t turn 10x work into 20x work because the company dedicated entire teams to write company-wide skills for everything. They suck, but if I don’t use them, I’m not following the new “golden path for engineering”, and I lose points in my performance review.
I can, however, turn 10x work into 20x work, or even much more than that, if AI actually did what it’s promising and eliminated most of my team, the product manager, and the middle managers. Or me. I could use a break.
What about the 6x developers? Was there just a doubling multiplier across the board, resulting in them becoming 12x developers, or did they too become 20x developers?
>> What concrete business advantage are you getting from LLMs?
> Speed.
Speed of what?
Speed of understanding what needs to be done? I highly doubt it.
Speed of LoC checked into git? Sure, I'll give you that.
But one can use any number of tools to generate hundreds of thousands of lines of code. See any build tools which support specifications such as RAML, OpenAPI, CORBA, etc.
fixing minor bugs takes one slack message for us now. bugs go down, goodness go up.
fixing more serious regression also easier. connect honeycomb mcp, ask agent to debug while i walk to coffee and get some pistachio rose dates. by time im back with my oat latte ive got a full report on what happened and can send the next slack message to fix.
I needed to deeply understand a code base I had no experience with in a language I don't normally use with what I would describe as haphazard documentation at best. You can't argue with the speed at which I gained the required understanding of the project.
In the time it took you to type that, your hourly market comp went down another basis point.
I am appalled none of this is clicking with you anti-AI folks. This is all so exciting -- alarming even! --, and software careers are never going to be the same.
I don't know how you just metaphorically stand there and act like nothing at all is happening. We've never seen anything like this in our entire lives.
Some of you are standing right in front of the steam roller, yelling to all of us that steam rollers aren't real.
A: The sky is blue!
B: No it's not.
A: Yes, it is, please look up.
B: No, you must prove it to me through reason.
A: But, if you would just pretty please look up.
B: No.
I run a company, I've been running it for 10 years, we do alright. I'm a shitty manager. Every time I've hired developers, the business freezes. The business isn't anything super important, the main consequence of bugs is that my family loses money. Everything has always rested on my shoulders. In theory there is some path for me to become a good manager, but I never landed on it. But now, with Claude, it's great. So far Claude has paid itself off in real profits at least 20x over, and that's with significant API usage on top of the monthly sub. I can prototype new features in an afternoon that before were on my giant list of "maybe somedays if I ever get to breathe" list. Our user experience has improved in so many ways that I knew were probably worth it, if I could just find the time. Now I can.
There are situations where yeah, it probably isn't ready yet. But, there are so many where it's amazing. Seriously, it's worth looking up.
I think there are many valid reasons to be against them - I think a lot of them are more right than wrong. It’s the “It can’t really do much” that I think must be from people that haven’t really tried it.
This is a great case for the benefits of using GenAI, in that you already possess an understanding of what you want to achieve. You know what it is you want to prototype, what is on your "giant list of 'maybe somedays if I ever get to breathe' list", what you want to end up delivering.
My point is and remains:
A) GenAI did not give you this understanding.
B) GenAI can only assist in your expressing this
preexisting understanding.
C) GenAI is a statistical token (text) generator and
cannot, by definition, "make" a person understand
what they want/need to do.
Did you use an LLM to write this for you? How odd.
For all of you people who think these LLM models are “earth shattering” how the hell do you reconcile that it’s a net positive for anyone but those who want to consolidate knowledge and power.
I guess I'll chime in as someone who thinks LLMs will be earth shattering, and specifically don't think it's a net positive for anyone but those whose power will be consolidated.
From my brief window of Fable usage, speed wasn't its strong point at all.
For actually building software, I'm starting to suspect a human with a dumber (but faster) model is going to get the job done quicker than Fable (and possibly even cheaper). Bug-finding and vulnerability detection is a different story.
My conclusion was the exact opposite. Maybe each individual response was slower, but it took so many fewer round trips to get what I wanted wanted. I had a project fable was progressing steadily and correctly on. Opus on the same project keeps handing me garbage it insists is working and meets the stated requirements, but isn’t and doesn’t.
