This was done because we asked for a minor change to be done (talking maybe 5 lines of code) and tested. So now not only are we dealing with new debt, we are dealing with code that no one can explain why it was completely changed (and some of the changes were changes for the sake of change), and we are dealing with those of us that manage this code now looking at completely foreign code.
I keep seeing this with people that are using these tools and they are not higher level engineers. We finally got to the point of denying these PR's and saying to go back and do it again. Loosing any of the time that was theoretically gained from doing it in the first place.
Not saying these tools don't have a place. But people are using it without understanding what it is putting out and not understanding the long term effects it will have on a code base.
It is worse than that. We're all maintaining in our heads the mental sand castle that is the system the code base represents. The abuse of the autocoder erodes that sand castle because the intentions of the changes, which are crucial for mentally updating the sand castle, are not communicated (because they are unknowable). This is same thing with poor commit messages, or poor documentation around requirements/business processes. With enough erosion, plus expected turn over in staff, the sand castle is actually gone.
Ultimately not adopting them will religate you to the same fate as assembly programmers. Sure there are place for it, but you won't be able to get near as much functionally done in the same amount of time and there won't be as much demand for it.
Thank you, this says what I have been struggling to describe.
The day I lost part of my soul was when I asked a dev if I could give them feedback on a DB schema, they said yes, and then cut me off a few minutes in with, “yeah, I don’t really care [about X].” You don’t care? I’m telling you as the SME for this exactly what can be improved, how to do it, and why you should do so, but you don’t care. Cool.
Cloud was a mistake; it’s inculcated people with the idea that chasing efficiency and optimization doesn’t matter, because you can always scale up or out. I’m not even talking about doing micro-benchmarks (though you should…), I’m talking about dead-simple stuff like “maybe use this data structure instead of that one.”
A lot of people who entered the field in the past 6 or so years are here for the money, obviously.
Nothing wrong with that at all, but as someone with a long time programming and technology passion, it’s sad to see that change.
In the olden days, we used to throw it over to "Ops" and say, "your problem now."
And Junior developers have always been overwhelmed with the details and under pressure to deliver enough to keep their job. None of this is new! I'm a graybeard now, but I remember seniors having the same complaints back then. "Kids these days" never gets old.
I’m more of a solo full stack dev and don’t really have first hand experience building software at scale and the process it takes to manage a codebase the size of the Windows OS, but these are the kinds of issues I see regularly these days and wouldn’t in the past. I also use macOS daily for almost as long and the Apple software has really tanked in terms of quality, I hit bugs and unexpected errors regularly. I generally don’t use their software (Safari, Mail, etc) when I can avoid it. Also have to admit lack of features is a big issue for me on their software.
With a few more years under my belt I realized theres nothing wrong with doing good work and providing yourself/your family a decent living. Not everyone needs the passion in their field to become among the best or become a "10x"er to contribute. We all have different passions, but we all need to pay the bills.
Off-the-cuff, three groups:
1. There are people who are motivated by having a solid income, yet they take the professionalism seriously, and do skilled rock-solid work, 9-5. I'd be happy to work with these people.
2. There are other people who are motivated by having a solid or more-than-solid income, and (regardless of skill level), it's non-stop sprint performance art, gaming promotion metrics, resume-driven development, practicing Leetcode, and hopping at the next opportunity regardless of where that leaves the project and team.
3. Then there's those weirdos who are motivated by something about the work itself, and would be doing it even if it didn't pay well. Over the years, these people spend so much time and energy on the something, that they tend to develop more and stronger skills than the others. I'd be happy to work with these people, so long as they can also be professional (including rolling up sleeves for the non-fun parts), or amenable to learning to be professional.
Half-joke: The potential of group #3 is threatening to sharp-elbowed group #2, so group #2 neutralizes them via frat gatekeeping tactics (yeah-but-what-school-did-you-go-to snobbery, Leetcode shibboleth for nothing but whether you rehearsed Leetcode rituals, cliques, culture fit, etc.).
Startups might do well to have a mix of #3 and #1, and to stay far away from #2. But startups -- especially the last decade-plus of too many growth investment scams -- are often run by affluent people who grew up being taught #2 skills (for how you game your way into prestigious school, aggressively self-interested networking and promoting yourself, etc.).
The hardest part of any job I’ve had is doing the not so fun parts (meetings, keeping up with emails, solidly finishing work before moving on to something new)
0 autonomy, the only people who got recognition were those who self promoted on internal blogs, etc.
Similarly Docker is an amazing technology, yet it enabled the dependency tower of babels that we have today. It enabled developers that don't care about cleaning up their depencies.
Kubernetes is amazing technology, yet it enabled the developers that don't care to ship applications that constantly crash, but who cares, kubernetes will automatically restart everything.
Cloud and now AI are similar enabler technologies. They could be used for good, but there are too many people that just don't care.
How many developers do we imagine even know the difference between SIMD and SISD operators, much less whether their software stack knows how to take advantage of SIMD? How many developers do we imagine even know how RAM chips store bits or how a semiconductor works?
We're just watching the bar of "Don't need to care because a reliable system exists" move through something we know and care about in our lifetimes. Progress is great to watch in action.
hopefully some of those that did a computer science degree?
(all of what you've said was part of mine)
At the root of it, there's a profound arrogance in putting someone else in a position where they are compelled to tell you you're wrong[1]. Curious, careful people don't do this very often because they are aware of the limits of their knowledge and when they don't know something they go find it out. Unfortunately this is surprisingly rare.
[1] to be clear, I'm speaking here as someone who has been guilty of this before, now regrets it, and hopes to never do it again.
in my defense, though, it's possible the option didn't exist when i first read the objdump manual
They are/will be the management's darling because they too are all about delivering without any interest in technology either.
Well designed technology isn't seen as foundation anymore; it is merely a tool to just keep the machine running. If parts of the machine are being damaged by the lack of judgement in the process, that shouldn't come in the way of this year's bonus; it'll be something to worry about in the next financial year. Nobody knows whats going to happen in the long-term anyway, make hay while the sun shines.
The age of short-term is upon us.
I found this too. But I also found the opposite, including here on HN; people who are interested in technology have almost an aversion against using AI. I personally love tech and I would and do write software for fun, but even that is objectively more fun for me with AI. It makes me far more productive (very much more than what the article states) and, more importantly, it removes the procrastination; whenever I am stuck or procrastinating getting to work or start, I start talking with Aider and before I know it, another task was done that I probably wouldn't have done that day without.
That way I now launch bi weekly open and closed source projects while before that would take months to years. And the cost of having this team of fast experienced devs sitting with me is max a few $ per day.
Like - I presume almost everyone - somewhere in the middle?
That was a helluva dichotomy to offer me...
> how do you keep your AI generated content from turning the whole thing into an incomprehensible superfund site of tech debt?
By reading it, thinking about it and testing it?
Did I somehow give the impression I'm cutting and pasting huge globs of code straight from ChatGPT into a git commit?
There's a weird gulf of incomprehension between people that use AI to help them code and those that don't. I'm sure you're as confused by this exchange as I am.
I certainly didn't intend it that way, more like a continuum. O(1) --> O(10) --> O(100) --> ...
> Did I somehow give the impression I'm cutting and pasting huge globs of code straight from ChatGPT into a git commit?
Yes, a little. It seemed to me like you were advocating using the LLM to generate large amounts of tedious output.
I use AI either as an unblocker to get me started, or to write a handful of lines that are too complex to do from memory but not so complex that I can't immediately grok them.
I find both types of usage very satisfying and helpful.
But it is far more useful on verbose 'team written' corporate stuff than on the more reuse intensive tech: in CL or Haskell, the community is far more DRY than Go or JS/TS; you tend to create and reuse many things and your much of the end result is (basically) a DSL; current AI is not very good at that in my experience; it will recreate or hallucinate (when you pressure reuse of previously created things, if there are too many, even though it does fit in the context window) functions all over the place. But many people have the same issue; they don't know, cannot search or forget and will just redo things many times over; AI makes that far easier (as in, no work at all often), so that's the new reality.
In my own experience I've worked on repos with <10 other devs where I spent far more effort on consistency and mantainability than getting the thing to work.
I wonder if this is an age thing, for many people. I'm old enough to have started reading these discussions on Slashdot in the early 90s.
But between 2000 and 2010, Slashdot changed and became much less open to new ideas.
The same may be happening to HN right now.
It feels like a lot of people are stuck in the tech of 10 years ago.
Personally, I don't use LLMs. But I don't mind people using them as interactive search engines or code/text manipulations as long as they're aware of the hallucination risks and took care of what they're copying into the project. My reasons for it is mostly that I'm a journey guy, not a destination guy. And I love reading books and manuals as they give me an extensive knowledge map. Using LLMs feels like taking guidance from someone who has not ventured 1km outside their village, but heard descriptions from passersby. Too much vigilance required for the occasional good stuff.
And the truth is, there are a lot of great books and manuals out there. And while they teach you how to do stuff, they often teach you why you should not do it. I strongly doubt Copilot imparting architectural and technical reminders alongside the code.
For my never finishing side projects I am too; I enjoy my weekends tinkering on the 'final database' system I have been building in CL for over a decade and will probably never really 'finish'. But to make money, I launch things fast and promote them; AI makes that far easier.
Especially for parts like fronted that I despise; I find 0 pleasure in working with css magic that even seasoned frontenders have to try/fail in a loop to create. I let Sonnet just struggle until it's good enough instead of me having to do that annoying chore; then I ask Aider to attach it to the backend and done.
So, like before AI then? I haven't seen AI deliver illogical nonsense that I couldn't even decipher like I have seen some outsourcing companies deliver.
This is one of the challenges of being a tech lead. Sometimes code is hard to comprehend.
In my experience, AI delivered code is no worse than entry level developer code.
AI doesn't do stuff like because it could not, which to me, is a good thing. When it gets better, it might start to do it, I don't know.
People here live in a bubble where they think the world is full of people who read 'beautiful code', make tests, use git or something instead of zip$date and know how DeMorgan works; by far, most don't, not juniors, not seniors.
"Back to that two page function. Yes, I know, it’s just a simple function to display a window, but it has grown little hairs and stuff on it and nobody knows why. Well, I’ll tell you why: those are bug fixes. One of them fixes that bug that Nancy had when she tried to install the thing on a computer that didn’t have Internet Explorer. Another one fixes that bug that occurs in low memory conditions. Another one fixes that bug that occurred when the file is on a floppy disk and the user yanks out the disk in the middle. "
https://www.joelonsoftware.com/2000/04/06/things-you-should-...
Of course, you can try to split the 15000 line function in logical blocks or something, but don't assume the ifs in it are useless.
I have. If you're doing niche-er stuff it doesn't have enough data and hallucinates. The worst is when it spits two screens of code instead of 'this cannot be done at the level you want it'.
> that I couldn't even decipher
That's unrelated to code quality. Especially with C++ which has become as write only as perl.
We use it internally and the technical debt is an enormous threat that IMO hasn't been properly gauged.
It's very very useful to carpet bomb code with APIs and patterns you're not familiar with, but it also leads to insane amounts of code duplication and unwieldy boilerplate if you're not careful, because:
1. One of the two big bias of the models is the fact that the training data is StackOverflow-type training data, which are examples and don't take context and constraints into account.
2. The other is the existing codebase, and it tends to copy/repeat things instead of suggesting you to refactor.
The first is mitigated by, well, doing your job and reviewing/editing what the LLM spat out.
The second can only be mitigated once diffs/commit history become part of the training data, and that's a much harder dataset to handle and tag, as some changes are good (refactorings) but other might be not (bugs that get corrected in subsequent commits) and no clear distinction as commit messages are effectively lies (nobody ever writes: bug introduced).
Not only that, merges/rebases/squashes alter/remove/add spurious meanings to the history, making everything blurrier.
Either you will go bust, OR you will be able to hire enough people to pay those debts, once you get traction in the market.
it's usually written "feature: ...
Is this a bad thing? Maybe I'm misunderstanding it, but even when I'm working on my own projects, I'm usually trying to solve a problem, and the technology is a means to an end to solving that problem (delivering). I care that it works, and is maintainable, I don't care that much about the technology.
It reminds me of a talk Raymond Hettinger put on a while ago about rearranging the flowers in the garden. There is a tendency from new developers to rearrange for no good reason, AI makes it even easier now. This comes down to a culture problem to me, AI is simply the tool but the driver is the human (at least for now).
Bingo, this, so much this. Every dev i know who loves AI stuff was a dev that I had very little technical respect for pre AI. They got some stuff done but there was no craft or quality to it.
In terms of directly generating technical content, I think he mostly used gen AI for more mechanical stuff such as drafting data schemas or class structures, or for converting this or that to JSON, and perhaps not so much for generating actual program code. Maybe there's a difference to someone who likes to have lots of program logic generated.
I do think there is a difference between a skilled engineer using it for the mechanical things, and an engineer that OFFLOADS thinking/designing to it.
Theres nuance everywhere, but my original comment was definitely implying the people that attempt to lean on it very hard for their core work.
However, it does look like LLM's are racing to make these junior devs unnecessary.
The main utility of "junior devs" (regardless of age) is that they can serve as an interface to non-technical business "users". Give them the right tools, and their value will be similar to good business controllers or similar in the org.
A salary of $100-$150k is really low for someone who is really a competent developer. It's kept down by those "junior devs" (of all ages) that apply for the same jobs.
Both kinds of developers will be required until companies use AI in most of those roles, including the controllers, the developers and the business side.
.. the exact same content for a screensaver with Todd Rundgren in 1987 on the then-new color Apple Macintosh II in Sausalito, California. A different screensaver called "Flow Fazer" was more popular and sold many copies. The rival at the time was "After Dark" .. whose founder had a PhD in physics from UC Berkeley but also turned out to be independently wealthy, and then one of the wealthiest men in the Bay Area after the dot-com boom.
Thanks for reading my post. It’s nice to know I’m not writing to a complete void!
I wrote a kotlin idea plugin in a day; I’ve never used kotlin before and the jetbrains ui framework is a dogs breakfast of obscure edge cases.
I had no skills in this area, so I could happily lean into the assistance provided to get the job done. And it got done.
…but, I don’t use coding assistants day to day in languages I’m very familiar with: because they’re flat out bad compared to what I can do by hand, myself.
Even using Python, generated code is often subtly wrong and it takes more time to make sure it is correct than to do it by hand.
…now, I would assume that a professional kotlin developer would look at my plugin and go: that’s a heap of garbage, you won’t be able to upgrade that when a new version comes out (turns out, they’re right).
So, despite being a (I hope) competent programmer I have three observations:
1) the code I built worked, but was an unmaintainable mess.
2) it only took a day, so it doesn’t matter if I throw it away and build the next one from scratch.
3) There are extremely limited domains where that’s true, and I personally find myself leaning away from LLM anything where maintenance is a long term goal.
So, the point here is not that developers are good/bad:
It’s the LLM generated code is bad.
It is bad.
It is the sort of quick, rubbish prototyping code that often ends up in production…
…and then gets an expensive rewrite later, if it does the job.
The point is that if you’re in the latter phase of working on a project that is not throw away…
You know the saying.
Betty had a bit of bitter butter, so she mixed the bitter butter with the better butter.
But it made the better butter bitter.
More stuff will get done, the barrier of entry will be lower etc.
The craft of programming took a significant quality/care hit when it transitioned from "only people who care enough to learn the ins and outs of memory management can feasibly do this" to "now anyone with a technical brain and a business use case can do it". Which makes sense, the code was no longer the point.
The C++ devs rightly felt superior to the new java devs in the narrow niche of "ability to craft code." But that feeling doesn't move the needle business wise in the vast majority of circumstances. Which is always the schism between large technology leaps.
Basically, the argument of "its worse" is not WRONG. Just, the same as it did not really matter in the mid 90s. Does not matter as much now, compared to the ability to "just get something that kinda works."
In the scheme of things however, that status hardly matters compared to the "ability to get something shipped quickly" which is what the vast majority of people are paid to do.
So while I might judge those people for not meeting my personal standards or bar. In many cases that does not actually matter. They got something out there, thats all that matters.
There are two interpretations of this:
1) Those people are imposters
2) It's about you, not them
I've been personally interested in AI since the early 80's, neural nets since the 90's, and vigilant about "AI" since Alexnet.
I've also been in a tech lead role for the past ~25 years. If someone is talking about newer "AI" models in a nonsensical way, I cringe.
You get what you measure. Nobody measure software quality.
So, pretty similar to how it was before. Except that motivated junior developers will improve incredibly fast. But that's also kind of always been the case in software development these past two decades?
https://news.ycombinator.com/item?id=41465827
(the two comments might have moved apart by the time you read this).
Edit: yep, they just did.
Those kinds of safeguards should instead be part of the framework you're using. If you need to prevent SQL injection, you need to make sure that all access to the SQL type database pass through a layer that prevents that. If you are worried about the security of your point of access (like an API facing the public), you need to apply safeguards as close to the point of entry as possible, and so on.
I'm a big believer in AI generated code (over a long horizon), but I'm not sure the edge case robustness is the main selling point.
> I wonder if the study includes the technical debt that more experienced developers had to tackle after the less experienced devs have contributed their AI-driven efforts.
You also may find this post from the other day more illuminating[0], as I believe the actual result strongly hints at what you're guessing. The study is high schoolers doing math. While GPT only has an 8% error rate for the final answer, it gets the steps wrong half the time. And with coding (like math), the steps are the important bits.
But I think people evaluate very poorly when there's ill defined metrics but some metric exists. They over inflate it's value since it's concrete. Like completing a ticket doesn't mean you made progress. Introducing technical debt would mean taking a step back. A step forward in a very specific direction but away from the actual end goal. You're just outsourcing work to a future person and I think we like to pretend this doesn't exist because it's hard to measure.
For them programming is a means to an end, and I think it is fine, in a way. But you cannot just ask an AI to write you tiktok clone and expect to get the finished product. Writing software is an iterative process, and LLMs currently used are not good enough for that, because they need not only to answer the questions, but at the very minimum to start asking questions: "why do want to do that ?" "do you prefer this or that", etc., so that they can actually extract all the specification details that the user happily didn't even know he needed before producing an appropriate output. (It's not too different from how some independent developers has to handle their clients, isn't it ?). Probably we will get there, but not too soon.
I also doubt that current tools can keep a project architecturally sound long-term, but that is just an hunch.
I admit though that I may be biased because I don't like much tools like copilot: when I write software, I have in my mind a model of the software that I am writing/I want to write, the AI has another model "in mind" and I need to spend mental energy understanding what it is "thinking". Even if 99/100 it is what I wanted, the remaining 1% is enough to hold me back from trusting it. Maybe I am using it the wrong way, who knows.
The AI tool that work for me would be a "voice controller AI powered pair programmer": I write my code, then from time to time I ask him questions on how to do something, and I can get either an contextual answer depending on the code I am working on, or generate the actual code if I wish so". Are there already plugins working that way for vscode/idea/etc ?
It is very nice in that it gives you a handy diffing tool before you accept, and it very much feels like it puts me in control.
If I don't deliver my startup burns in a year. In my previous role if I didn't deliver the people who were my reports did not get their bonuses. The incentives are very clear, and have always been clear - deliver.
Here's a real example of delivering something now without worrying about being the best engineer I can. I have 2 CSRs, They are swamped with work and we're weeks away from bringing another CSR on board. I find a couple of time consuming tasks that are easy to automate and build those out separately as one-off jobs that work well enough. Instantly it's a solid time gain & stress reducer to CSRs. Are my one-off automation tasks a long term solution? No. Do I care? Not at the moment, and my ego can take a hit for the time being.
The few attempts I've made at using genAI to make large-scale changes to code have been failures, and left me in the dark about the changes that were made in ways that were not helpful. I needed suggestions to be in much smaller chunks. paragraph sized. Right now I limit myself to using the genAI line completion suggestions in Pycharm. It very often guesses my intentions and so actually is helpful, particularly when laboriously typing out lots of long literals, e.g. keys in a dictionary.
You are not a fucking priest in the temple of engineering, go to fucking CS dep at the local uni and be the one and preach it there. You are worker of the company with customers, which pays you a salary from customers money.
Also, I've personally seen more interest in AI in devs that have little interest in technology, but a big interest in delivering. PMs love them though.