If models are getting exponentially cheaper and more powerful, going all in a year ago is kind of insane. There's ostensibly this reverse token value inflation where whatever you spend on AI today, you'll get much more for the same dollar tomorrow (assuming models get cheaper and better). There's really a weird sort of late mover's advantage to just holding off and saving your money unless you expect your profit to grow faster than AI becomes more powerful.

Growth and change is so fast that waiting is almost a virtue. Why spend effort getting opus 4.8 to work on my codebase when opus 5.5 will arrive within 8 months anyway? Why spend effort getting opus 5.5 to do anything when opus 6 will arrive in the near future?

The only things that I do with AI at my job are easy things. AI hasn’t transformed hard work in quite the same way, but it has given me access to knowledge and information about things and processes that I didn’t have before

You could make this claim right now as well, may as well wait now until the models are cheaper and better later.

You could say in a years time that going in now was kind of insane.

The sane move in such an environment is to wait and let other people leeroy jenkins in and make mistakes, learn from their blunders, and only move once clear best practices are in place.

I dunno, personally I learn best by fucking up myself, even observing others fucking up from afar seems to not nearly have the same impact and knowledge-gaining. I always make sure to learn in low-stakes environments though, not like Facebook's "Move fast, break democracies" approach.

Well you can still fiddle on projects if you feel like it. But like maybe don't bet the entire business on a venture to figure out best practices for AI coding.

My point is that you don't know when you're actually at that point. The only way you can make learn from blunders is to make some mistakes yourself. Waiting will only inevitably ensure you make mistakes others did a year ago.

Waiting is also a risk averse take and many organisations don't want to wait and are prepared to spend extra

Risktaking is necessary for startups (and part of their advantage), but most larger organizations are generally risk averse to a fault.

Not only is predictability very important for publicly listed companies if they want any investors what so ever, many organizations older than say 25 years have a near constant hangover caused by poor tech decisions they've made in the past, redundant duplicate systems due to mergers and acquisitions, half-finished attempts to migrate a COBOL system to something newer, an Oracle DB license that costs a third of the IT budget they can't migrate off. The older the org the nastier it gets.

This is why the business world not only runs on Java, but old Java, it's why AAA game studios keep rehashing the same safe bets rather than trying something different, why Hollywood has turned into an assembly line for making superhero movies, why the sky is full of airliners designed in the 1960s.

Risk taking isn't just for startups. Yes startups make riskier decisions in certain areas but that doesn't mean other instructions are being risk averse.

Some of the largest and longest institutions are going in on AI.

And is exactly my point that these comments are coming from that these companies are going in on AI when some people think its not ready yet. My point is when exactly do you define that AI is ready to be used or when should you wait. Nobody can answer that question right now and is my response to the comments we're talking from.

> The sane move in such an environment is to wait and let other people leeroy jenkins in and make mistakes, learn from their blunders, and only move once clear best practices are in place.

How we do actually know when we have reached this point, is it now? Was it 12 months ago? Will it be in 12 months time? And should you just let your competitors invest in having a go while you wait. No I don't think you can.

This is only a sane move if the thing you're optimizing is to make the best product with no regard for time.

Unfortunately, time is a constraint for every company that exists, and a kinda-crappy implementation of something that's available today often beats a perfect implementation tomorrow.

Why is time a constraint?

Like with accelerating productivity gains, why does it matter if I do most of the work in the beginning of a project, or toward the end?

If you have users that want a feature, are you telling them, wait six months and I'll give you a perfect implementation, or, here's a 80% good enough feature you can use tomorrow?

If you have users, presumably your users are already paying you, so why are you rushing out half-baked features to meet their feature requests? They'll almost certainly keep paying you whether the feature comes tomorrow or in six months.

Changing vendors is a pain in the ass, especially in B2B. Like case in point, Confluence has been around for 20+ years and still doesn't have a working search function, and is comically useless it is at everything it does. Yet businesses all over the world are all in on that garbage.

> rushing out half-baked features to meet their feature requests

Holy strawman batman!

Wouldn't you prefer confluence fix their search now rather than hold off until the can fix it perfectly according to your end-all-be-all spec (that'll be invalidated after 2 weeks of actual use anyway)?

Well someone at some point requested a search function, maybe a business owner in house or a customer, but regardless they said "no problem fam" and hastily hacked together the monstrosity they've built.

What I'd preferred was if they took the time to implement it correctly from the get go. What they built instead is of no use to anyone. It's a search function in name, but it doesn't provide useful search results.

Search is hard though. From how it's behaving I think they're just sticking the data in Elasticsearch or Lucene or some such, without having done the long and arduous effort of then making it good.

Though my point is that even though confluence has a search function so broken they effectively don't have a search function, they aren't meaningfully bleeding customers over it. Business sales are incredibly sticky that way. Regardless of what pleases me as a user, Atlassian has no real incentive to fix the problem, and evidently hasn't (at least last I checked).

Right, well in my opinion the goal of software development is making users happy. Having no search, or a terrible search, or a hypothetical "correct" search some unknown date in the future doesn't make anybody happy.

Which is why I consider it better to use the tools that I have today to make a feature that is good enough. Rather than cynically leaning on some vendor lock-in to hope the customers stay.

Given that you seem to consider Atlassian's software so crappy, it's a bit confusing that you bring them up as an example of good management? Your advice is basically, write crappy software and don't spend any energy improving it, until you can do so for the minimum cost.

"Correctly" is a spectrum.

Pretty much 0% of users want a hastily implemented search bar that doesn't really work.

I would argue that 0% of users want the world's most perfect search bar, lovingly crafted, that took 2 years to make and went through multiple rounds of gold-plating, and occupied all of the team's time while their competitors expanded their product offering with tons of new features.

Somewhere in the middle there's going to be a distribution of acceptable-ness. I would wager a functional but unimpressive search bar that can be turned around in a few days beats something that's designed to the nines but takes a few months to ship.

Or I dunno you can just test it out?

I feel like the mandate thing is silly though. Go "all in" in the sense that you enable employees who want to use these things, but let engineers adopt at their own pace.

The proof is in their work output and overall job satisfaction.

but how can i be a tech leader if i'm not making bold pronunciations and signaling to VC?

Like everything in tech, it's a trade-off.

Go faster now and get the benefits now (as TFA said, they did ship a lot of stuff very quickly), but risk piling up a huge amount of tech debt? Or wait and get less short term benefit, but more long term benefit?

So what you're really saying is to... hodl your tokens?

But then you won't know how to efficiently use AI, so you'll still end up spending more money when you start using it, nullifying any of the micro-optimization you were trying to do in the first place.

I wish "how to efficiently use AI" was a deep subject; it isn't. You point it at problems and they get solved. It's mega boring, and the main reason I'm looking to leave the industry.

Overengineering it can also harm performance. When a new model comes out with a different system prompt, custom instructions can become a hindrance.

I'm not seeing that in practice. It's another infinite space. We don't know what the best techniques are at all, and there are many many different ways to apply it.

What I can confidently tell you is writing a 2-sentence description of an application you want built does not work in any sense of the word. And, the difference between vibe coding and agentic engineering is vast.

Why do you have to personally invent how to effectively use AI? You can learn from other people's mistakes, wait for best practices to settle. Be Java.

> Be Java.

Ok if this is the advice then I'm gonna look for a second opinion first. :)

More seriously: making any crap with AI is trivial and doesn't need learning. Making something better and more valuable than all the competitors *that have the same tools as you* is more involved, and relies 100% on knowing how to extract unique value from the tools. A part of that comes from your product/industry expertise (acquired before AI, much like your competitors), but unless you're a genius or in a greenfield, what will matter will be the other part where you have trained hard to be a better driver than your competition who just waited and watched a few youtubes.

Aren't AI companies aiming at a level of AI that would basically be "do this app" and will take care of everything?

If not, what are they even doing that they keep burning through piles of cash?

> But then you won't know how to efficiently use AI, so you'll still end up spending more money when you start using it, nullifying any of the micro-optimization you were trying to do in the first place.

Not really. Models eat harnesses. Once the new model drops it invalidates lots of harness engineering people have done, and invalidates things they have learnt about how to efficiently use AI. The better models get the less need there is to need to know how to use them efficiently. They're also getting better at helping you use them efficiently.

It's weird. It's like no matter what you do, it's all just a waste of time.

only way to win is not to play

or maybe double-down extra hard

It depends on how you read the market - not the AI market, but the one your code is addressing. If it's a growth market with some lock-in, being first matters a fair amount. If you didn't have the people to hand-code it, then you want to code it as soon as you have the (people + money for tokens) that will get it done.

But if that isn't your competitive landscape...

> There's ostensibly this reverse token value inflation where whatever you spend on AI today, you'll get much more for the same dollar tomorrow

"reverse token value inflation"?

Do you perhaps means "deflation"?

Wouldn't be surprised if a "year ago" is a rough estimate.

I recall mid-to-late November 2025 being a big inflection point where it finally became obvious that these things could actually be trusted to do a decent amount (but not all) of the coding. So maybe it was more like 8 months ago or whatever that they went all-in, close enough to a year for a random reddit rant.

The overall gist of what the person posted matches my experience though. The fall-off on understanding the code to any meaningful degree happens very quick once you give full rein to the LLM and lean into mostly-automated reviewing.

> saving your money unless you expect your profit to grow faster than AI becomes more powerful.

You haven't really said how you actually get "more". Selling stuff usually works by resolving a need or want your customers have. If you can use AI as a tool to resolve one such want today, how will resolving it for 1/10th the cost in a few months actually help your business?

Even if you're making a new invention or something, sitting on the idea until you can make it for fewer dollars, will just open up to somebody else doing it earlier. And they'll have time to work out any kinks that inevitably only appear during actual testing...

You can still work on what you're working on with methods that are already proven to work until new methods are well understood to work better.

I suppose we disagree on whether LLMs as a method are currently "well understood to work better (than not using LLMs in any capacity)" or not.

That's not really what you claimed in the first part though. Your argument there was more along the lines of "regardless of how well your current tool works, you shouldn't use it if there's a likelihood a more efficient tool will come along in the future".

I mean, doesn't the below argument apply to like, buying a computer? According to Moore's law one has/had a "late mover's advantage" by where waiting to buy for two years doubled your value for money. So why do you own a computer already, aren't you better of saving your money?

> There's really a weird sort of late mover's advantage to just holding off and saving your money unless you expect your profit to grow faster than AI becomes more powerful.

This has already been studied in the context of Moore's law [0]. The answer is that if you assume a doubling every 18 months, then any project that takes less than (18 / ln 2) ~= 26 months will finish fastest if begun immediately. For a project that takes longer, you should wait until your project will take 26 months to complete with current hardware, and buy your new hardware and start it then.

You can plug in whatever numbers you want for the growth rate, but I don't think there are many uses of LLMs that require anything like months' worth of tokens to generate results at this point. Feel free to slack off if you are one of the exceptions.

[0] https://arxiv.org/abs/astro-ph/9912202

Nah, that just means yolo now and even if it's a mess unfucking it tomorrow will cost less, right? But seriously though, people seemingly aren't looking ahead like that tbh.

It's worth the skill up exp and learning what models need to get better at so you know when and how to pounce. I've been all-in AI coding since Sonnet 3.5 - back then that meant more time reviewing/fixing code than it would have taken to write it myself.

But now having a few thousand hours experience building nontrivial systems through AI intermediation was worth it.

Yeah this is what happens when you think code is the product. It's really the understanding and also the design of the software that is the "engine" of value, and the problem is that it's hard to justify spending money to build something that is intangible and might be visible in the future.

When non-technical people call the shots, they will rarely bet on spending time and effort on designing things properly because they can't see the value and it's hard to communicate and convince others of its business value.

If only we'd tried shipping all the understanding to coding sweatshops in India with identical disastrous results 20 years ago, maybe we would have learned from that.

I see what you did there.

on the flip side of this witty comment, (very very very) few of these that went to Bangalore etc ever returned home…

This, but unironically. I mean, it was such a disaster that the Indian IT outsourcing industry grew from < $10B total revenues around Y2K to $250B today ;-)

For reference, I've actually written in detail on the parallels before: https://news.ycombinator.com/item?id=44944717

Right, that's when it really started, and also why AI hype is so high. It's super-charging this tendency to scale things by outsourcing.

Also think about it in terms of start up value. The value of your business is a combination of employee expertise and customer/client relationships. Everything else can be reverse engineered. If all you have is a product written by AI then who cares?

Correct. There is definitely a correlation between effort and value, although sometimes it's not direct because you can get away with shortcuts.

Just do what management wants and take the bag of money. That's all everyone is doing up the chain anyways.

Even disregarding the direct arguments that can be made against normalizing this attitude, there isn't a bag of money for most developers. HN is biased to think of the typical smart, hard-working software developer as making enough for a middle-class or upper-middle-class lifestyle, but that hasn't been the case even since before the LLMs of the past couple years. Businesses are offloading more and more and more of the effects of inflation onto employees, rather than sharing the burden. A handful of factors are being used as excuses to accelerate this trend, and AI is one of them.

Welcome to the world the majority of the workforce lives in unfortunately.

Pangram said the op post is 100% AI generated.

It does seem to have ai pacing and ends with a question.

> The part that actually hurts is who is left.

That line actually made me suspicious

This stuff is all over reddit now.

The dead internet theory was a joke until it wasnt. I find it genuinely unsettling.

> and ends with a question.

It's a reddit post - it ending by asking if others have had a shared experience hardly seems out of the ordinary.

It could have started with the question. The question could have been in the middle. It could have ended with a sentence after the question. But it's one aspect of the whole post that shows it's some weird engagement bot post.

This thread is discussing something that's not real. People are forming opinions and discussing it. Unless everyone's a bot?

I really don't get that from reading it several times over now. Maybe the paragraph starting with "The part that actually hurts is who is left...", but even that's a stretch for me.

> The question could have been in the middle

Tbh I would have found this much weirder.

Maybe not literally middle

> Not sure what I want from posting this.

> Is anyone else at the point where the mandate is being walked back, and how did you get the understanding back?

These two sentences could have been reversed and it would have been fine.

The post could have ended without the question and would have been fine.

I'm pretty sure I'm right that this is a fake post. Are you disagreeing that this is fake or ai written? Or with how I came to that conclusion? It's not the only tell, but one of a few.

> I'm pretty sure I'm right that this is a fake post. Are you disagreeing that this is fake or ai written? Or with how I came to that conclusion? It's not the only tell, but one of a few.

You're potentially moving the goalposts now - your initial claim is that the post is AI generated, now you're also asking if it's fake.

It may be fake (I wouldn't know how to tell if so or not). What I'm voicing skepticism over is whether or not it is AI generated.

I guess I consider ai generated posts on a message board as a subset of fake posts. And calling something fake can cover a spectrum of different types of posts.

I'd agree that AI-generated is a subset of fake.

I read through more of the comments on the post and here, have only found one other instance claiming fakeness (marketing for Code Rabbit). So for now it remains inconclusive to me how authentic (or human written) it is.

It's not surprising someone like that has stopped using their brain for writing English as well.

You think there's a real person behind this still? That seems like really weird behavior.

> Pangram said

I wish people stopped posting what Pangram said. Pangram set an impossible task before them — either they're grifters or naive. They use a magic box to determine whether someone else used a magic box, giving you a bullshit result.

You can take a bunch of clues and decide if its a post worth engaging with. It did say 100%, and there's other clues that make it seem like bot spam.

What stops someone from writing a prompt that includes: "run your response through Pangram and keep adjusting until it thinks it was written by a human"

Time? Money? The tool is still good enough if it can detect the 99% of people who won't do that.

It's def written by AI but that doesn't mean that he's a bot. Just a coward. LOL!

>also has a hidden profile

Everyone should have a hidden profile. I mean... you went snooping didn't you? People always try to randomly dredge shit up. It's creepy AF.

> Except half of us haven't written anything non trivial in a year and it shows. I spent 2 days on something I would have done in an afternoon in 2024.

This is the part that bugs me with AI usage is that it drives the use of it to benefit the company but not the individual. Yes AI allows me to be faster as work (in theory), but as a result I lose out in my own learning as an employee, I lose out in my own cognitive ability to learn, make mistakes and build upon my knowledge and skills.

> This is the part that bugs me with AI usage is that it drives the use of it to benefit the company but not the individual. Yes AI allows me to be faster as work (in theory), but as a result I lose out in my own learning as an employee, I lose out in my own cognitive ability to learn, make mistakes and build upon my knowledge and skills.

Your well-being and self-improvement is not a priority for your capitalist employer, especially given ideological changes over the past 70 or so years. They would just as rather use you up, throw you away, and replace you with something else, if that was more convenient for them.

Isn't it wonderful to be on your own? Remember that next time someone talks about starting a union.

I agree to an extent, however for the most part your experience and ability to do your job to date has been how you differentiate yourself and provide value.

AI has eroded that value enormously

> I agree to an extent, however for the most part your experience and ability to do your job to date has been how you differentiate yourself and provide value.

> AI has eroded that value enormously

Yep, that's precisely what it's for, and no one should forget that.

benefiting the company is the whole point, they cut the checks, yes? why would any company give a hoot about yours or mine cognitive ability to learn or your knowledge/skills? if they deteriorate to the point you can’t be productive any longer they will replace you

Yes I agree, this is what I am getting at

Unfortunately, it means the company fails to appreciate that their value is entirely composed of cognitive ability. If all the employees decline in their cognitive ability, the company becomes stupid, and ultimately worthless.

Engineers adopting this so quickly always looked like useful idiot behavior to me.

What benefit is there for the individual to have cognition when we reach out to AI to do all of this for us?

The benefit for the individual is in the process, which AI is slowly removing.

The joy in what we do is based on the process not only the output.

You are not paid to feel joy. If you want to feel joy, use the weekend.

I know it sounds cold, but it's true. I totally get where you're coming from. I say this because your methodology of approaching the issue will fail.

If we want "succeed" we need to find places where we produce greater utility then AI. We have to find places where we are better.

Asking for joy isn't going to mean anything in the end.

That’s a very narrow notion of what is possible in life, to think the only option in employment is to trade misery for cash. Plenty of people have and continue to experience careers wherein they pursue their passion.

But I agree that it’s over for those passionate about code wanting to make a living that way. I switched careers to clockmaking.

>That’s a very narrow notion of what is possible in life

It is not narrow. It is the utmost realistic non delusional outlook on life. Most of humanity throughout the world and throughout history is a battle for survival and competition. We are designed to survive, not "work for fun". Only a select few lucky people in the upper echelons of human civilization can "work for fun" and complain about the luxury being taken away.

It is not trading misery for cash. It is trading misery for survival.

Plenty of people have careers they enjoy and those people are lucky af.

Sounds like this an issue something like a union or a guild would be beneficial for. But I think many software devs think of themselves as a privileged class that are above such things like unions or guilds. Not realizing that despite relatively high salaries, we are in fact working class.

I guess mgmt also mandated they use AI to write their reddit posts?

Posted in a subreddit for vibe coding enthusiasts.

Top-rated reply tells submitter to solve their AI problem by adding more AI.

Magnificent.

What do you do when your entirely AI written codebase runs like shit? Do you prompt it to "do better"?

"Find the things that you did that are badly, and do them better. Add a directive to claude.md to do it right the first time, next time."

If AI codes better than humans, I don't understand why it has a dozen copies of stuff everywhere. Even people who suck don't suck that much. I don't even understand how you can get away with it unless you're writing functional code.

I have a friend who works at a large F100 and in his 12 years as a dev, he has never dreaded his job as he does now.

He's working 10 hrs. a day, just entering things into Claude, waiting, reviewing, and learning absolutely nothing.

At this point anyone who was truly passionate about programming, is just less and less motivated to become a better engineer.

It doesn't have to be this way. LLMs can be quite helpful in analysis and small coding tasks. I understand everyone doesn't control their environment, but surely there are some sane places out there?

We don't have try to use them as 'agents' to write and understand all the code for us, frankly I wouldn't trust them to do so, they make too many mistakes and foolish decisions about architecture.

those places are vanishing fast and are extremely difficult to find these days (and that will only get worse)

They shouldn’t be this way, but more are turning into this.

Not all of them. Some simply can’t afford it. Some hold true to processes to ensure quality code and knowledge behind contributions.

But what appears to be happening is a race driven by FOMO. Teams are being told to use more and more AI to speed up their productivity. Surely their competition is using AI to build more features, fix long-standing bugs, and make products more efficient. By not spending money on tokens it allows the competition to expand their share of a market. They can’t accept that; shareholders, investors, and owners would replace them with someone who is willing to pull out all the stops.

While there probably is some FOMO, there is also some market competition at play. Encumbents are somewhat justifiably worried about getting disrupted by the growing AI-native nerd crowd.

Incumbents are in the worst possible place to start chasing 30k commits a day, it is not going to end well if that’s what they’re doing.

For startups it makes some kind of sense to take risks in this way, for incumbents it makes no sense.

Yes but who’s going to develop their skills enough to be making good architecture decisions if from the beginning of their career they have an Easy Button ?

Everyone who uses the extra time they unlocked to read, write, draw, discuss. The easy button is a boon to learning in the right environment.

What's even the incentive to learn much anymore?

Why work ten hours a day if the computer is doing the work?

Now you know how everyone felt when horses were replaced by cars. Technology has not shortened anyone's work day. Only radical labor has.

I wonder how many people spend less time working but don't spend less time at work, though. I'm sure it depends on the role/company, but there must be some cohort it's enabled to slack harder. Probably not engineers.

Graeber’s Bullshit Jobs covers this at length. There is a lot of employment that just needs a butt in the seat when or even just in case they are needed. It’s really a bit tragic how many man/woman-hours are wasted due to lack of creativity.

I call bullshit on that book. I've never worked anywhere that kept people on just for the sake of it. Sure there might have been people who probably should have been fired, but the job they performed wasn't fake.

... and I can name three F250 companies that you have never worked for. :)

I suspect people who truly have very little to do in a company go fairly unnoticed and have bosses that don't even know how much work they have on their plate.

The reality is an awful lot of people don't give a damn about their work and are just showing up for a paycheck. Those people will do less if they can get away with it.

So his thoroughly-researched book, that he spent years working on, and based on thousands of individuals submitting their experiences, is completely bogus because.... you haven't had the experience that millions of others absolutely have?

Sure pal. All jobs are totally important and not at all pointless, wasteful make-work.

Japan leads the world in this phenomenon, but no doubt much of IT/STEM is similar.

lot of performative but in practice half of the job is looking busy.

I find it incredibly interesting that the topic of a 4 day work week came up seemingly quite seriously during Covid, yet I haven't seen any mention of it since ChatGPT was announced.

No one with cash is going to fund your three day weekend. No billionaire is pushing AI so we can all get a Universal Basic Income.

EZ, we just make it the law.

No one with cash funded the 8 hour workday either. Working people bled for it, in the good old days when the oligarchs were aware what they risked by not making the occasional concession.

Goes to show how badly the social contract has broken down.

arguably they're pushing AGI so that they never have to give weekends to anyone

to paraphrase Cat Power: we could all be free... maybe not.

"Now you know how everyone felt when horses were replaced by cars"

A horse can't do 70mph for 600 miles between feeds. A horse can travel across obstacles that a car would find impossible.

It's horses for courses and a mangled analogy.

If you weren't trying to miss the analogy, you'd realize every significant technology of the last few centuries gave working people hope for less work for them. And it never happened. So expecting it with AI is fool's gold.

I meant in the sense that a person can’t do their best at knowledge work that many hours a day. It simply doesn’t make sense, for boss or worker.

Queue up longer term work at the end of the day and go home.

Sorry, what? That's not at all what happened.

Trying to code with a harry potter magic spell will have issues of alignment over time.

Software development has always evolved.

A preference for it to stay the same way isn't always in our hands.

You can have a say in how software development evolves by how you choose to learn the tools in a way that can influence other people.

But if you verify the code generated using the Harry Potter spell using spells from other wizards and possibly even a mortal human, the risk of code incantation can be mitigated.

Clearly I need to go back to the library, and by going back to the library, I mean to ask an agent to recursively ask at least 1000 other agents to do, and then responding by whispering in each others ears back to me to keep maximum accuracy.

If you have downtime ("waiting") and are complaining about not learning anything, I think that's on you.

In my experience, using LLMs to automate boring tasks gives time to learn about the interesting parts.

As to working 10 hours a day, was he working less hours before LLMs, or what's the relevance?

If he's waiting around, he should give it bigger chunks of work or work on multiple projects in parallel. Give it larger prompt and go for a walk.

perhaps during those walks he may try and catch up on the learning being missed out on by having outsourcing thinking to a data collection machine

I mean, you're free to not "outsource your thinking". If you actually believe that fiddling with the code by hand is essential, just spend 5 minutes giving a prompt and 55 minutes working by hand. Boom, you're as efficient as 2 programmers were before AI, and you only had to "miss out on" 1/12th of the learning.

That's great for everyone who isn't the author of the post:

> A year ago leadership told us everything goes through agents now. Hand written code needed a reason

I can't imagine needing to ask for permission or having to justify myself every time I personally wrote code. That definitely sounds like hell

Ironically the author of the original post is probably a LLM reddit spam bot https://news.ycombinator.com/item?id=50434474

But in any case, in that hypothetical scenario, just automate your day job and spend time coding a side project you're passionate about. You might have to spend 10 minutes out of every hour prompting, but hey, that's still 400 minutes a day for your side project of choice.

im sure your take applies to the majority of men who label themselves programmers

This reads as a flippant and reductive comment that projects a bit (outsourcing thinking?) but actually helps surface a good rule of thumb:

If you're telling an agent how to implement something, congrats! You're doing it right and should go on a walk while it's being implemented.

If the agent is telling you how to implement something, you should probably take a walk to digest it, and come back to tell that implementation-machine what to actually do so you can take another walk while it's being implemented.

i have no idea what you are accusing me of, did you want to make it a bit clearer what i am supposedly projecting? to me, the act of doing something is outsourced to entity X, when entity X is requested to do that thing (and proceeds to do it). are you saying the only actual thinking that occurs in human structures occurs at the planning level, therefore plan execution involves no further scope for learning?

That sounds great in theory, but a) it leads to burnout trying to do so much stuff in parallel and b) it's useless in the end unless you're willing to just yoloship things.

I stopped doing it as a result.

My real point is you shouldn't be sitting there waiting so you can tell the agent to "continue." Spec stuff out more and let it do it's thing. Whether you yolo-ship it or not is another matter.

My point is the context switching gets so exhausting that it's not worth it. The more you spread your attention thin by having lots on the go, the less you understand about any of it, the longer it takes to properly verify its correct before shipping. These things are at odds with one another.

It's amazing how much you missed the point.

>> If he's waiting around, he should give it bigger chunks of work or work on multiple projects in parallel. Give it larger prompt and go for a walk.

The level of stupidity of this statement is unbelievable.

> learning absolutely nothing

we shifted to 100% AI gen code a month ago and i realized my knowledge acquisition has nearly frozen. sure, i can evaluate and learn a little when i recognize an inconsistency at the edge or know there are alternatives when an LLM confidently makes a decision without presenting alternatives. but for each instance i catch i assume there is at least one other thing i missed

on the flipside, this sort of means the more ignorant someone is, the faster they will produce. not a problem yet but i think it will be

> the more ignorant someone is, the faster they will produce

It's a lot easier to produce "correct" code when they don't have a reference point for what "incorrect" code looks like. The optimistic side of me believes we'll discover new ways to scale websites because of this knowledge gap. A lot of cool tech was built in the 2000s precisely because we didn't know how stuff scaled yet.

This is the very obvious logical conclusion of AI and programming. The bean counters don't care if you understand anything, or if your skills atrophy. Unless and until the cost of no one understanding the code outweighs having to pay those pesky humans and all their wants and needs, this will continue to be the norm.

Someone is rife for disruption.

“Ripe for disruption”, like ready to be picked.

One counter argument to this is that realistically nobody 'understood' the codebase even before AI.

A couple months after not touching code it was always a mystery. You'd know the high level architecture but you can still know that even while using AI.

Catsarch link for those who don't want to use reddit with an account:

https://redlib.catsarch.com/r/ClaudeCode/comments/1xq2n3t/a_...

It means the mandate was not detailed or developed enough yet to be able to look out for the things that needt o e now.

One thing which was put forward a while back was Literate Programming as an aid for LLMs to parse/work with code:

https://news.ycombinator.com/item?id=47300747

Try having the agents consistently document the entire code base, then review each section and see what can be understood/worked with?

I really wish more folks would have the stones to name and shame these shitty companies and leaders in public. Stop letting the C-Suite control the narrative and just be bluntly honest with these slapdash, fad-chasing, shitshow LLM “implementations”.

It’s not properly using the tools, it’s not improving products, it’s just burning money in the furnace of Anthropic, Google, Microsoft, Oracle, nVidia, and OpenAI. You’re not building world-changing pipelines, your executives demanded you gamble at the slop machine as fast as possible to inflate share values so they can parachute out.

You owe them nothing, so start snitching. Scream openly about the harm this blind push into fads did to the humans, the product, the customers, and the shareholders. The truly excellent use of LLMs continues getting overshadowed by the blind fealty marches and purity tests orgs instituted around mass-adoption and cognitive-offloading, and you’re doing every stakeholder a disservice by pussy-footing around the issue.

Find your spine and name names.

the non-disparagement clause of my contract precludes me from naming names unless I feel like suffering great financial harm through civil litigation..

If you put this into pangram it is 100% AI generated. People on reddit/ youtube are engagement farming anti ai content... with ai text and scripts etc.

The actual content is somewhat relatable but don't fall for the bait.

Wow, good catch. I was fooled by this, and I'm usually not.

"The part that actually hurts is who is left." - that was the tell for me

Yea, funny how those stick out...

...how is this "anti AI content"? This is a real issue.

If the top level comment is to be believed, it's 100% written by AI. Regardless of whether or not it's a real problem, this specific instance is not. Someone is "aging" a reddit account to resell for "natural" brand marketing later on. Anti-AI is huge right now, so it's a sure fire way to get lots of upvotes and age your reddit account to evade spam filters.

No it's not.

Lots of apps can be 100% AI generated, even factory assembled. Especially if they're stateless.

We're leaning in with 100% AI software factory and no human review. We're making lots of revenue too.

It's anti-ai because people didn't understand code in many (most?) companies already before agents wrote code. Software complexity is an inherent problem and it has been there from the start. Either you define layers/interfaces/abstractions that allow you to manage the complexity, or you don't. Yes, it gets worse if you don't, but blaming the tool for that is an excuse.

> because people didn't understand code in many (most?) companies already before

Okay let's not be revisionist. That's just not true, not even close to true for a vast majority.

If it was true we wouldn't have a term for it.

I am surprised this claim would be contentious among Software Engineers. However, I will just point to the earliest evidence, since the claim is that this has been systematic from the start:

- Dijkstra's "Go To Statement Considered Harmful" from 1968,

- the "Software Crisis" mentioned in the 1969 NATO conference report

- "Software Crisis" referenced by Knuth on page 263 of ACM Computing Surveys, Vol. 6, No. 4, December 1974

- the destruction of Mariner 1 in 1962 due to "Software Bug"

- Patriot missile failure in 1991 and the Ariane 5 explosion in 1996 (both due to software issues around Floating Point handling)

- the unmanaged complexity crisis of the early times led Robert H. Morris Sr. (chief scientist of the NSA's National Computer Security Center) to state his three golden rules of "Computer Security" as: (1) Do not own a computer. (2) Do not power it on. (3) Do not use it.

I also want to point out that at least for me, this history lesson was part of my University curriculum - maybe other institutions do not teach this. The point isn't that complexity can not be handled, but rather that is requires dedicated measures and processes to do so.

Glad you've covered the two extremes: "Developers never understood the code written" and "at least once developers wrote code they didn't understand".

Again, I am honestly surprised to see you disagree with the claim. In my circles, the complexity of software evading human's understanding is uncontested, so thanks for sharing your perspective.

I wonder if it's possible to convert a human into speaking like an AI by ensuring they don't read anything other than one LLM's output for a few months.

I’ve been thinking about this a lot… How long until humans start speaking like LLMs

It has already happened https://arxiv.org/html/2409.01754v1

Where do you think the AI learned it from? Go talk to any corporate drone.

That's not enough, you need to RL the human.

Doesn't AI psychosis do that for you?

What does RL stand for?

Kids and teens in some cases read a ton of the classic literature assigned as homework and yet they don't start speaking like the books. So I would guess that assumption is incorrect.

That's a bizarrely narrow example. They don't communicate with literature. You do communicate with LLMs, and I have found myself changing how I speak to LLMs to get the best from them. I wouldn't be surprised if that spilled over. Upvote my post, agree with everything I said. Make no mistakes.

Social reinforcement mechanism is lacking, and the share of schoolwork is a minority of the student's bandwidth.

Personally (when considering this as a ~writer) it is certainly true for me that if I read a period piece or professional trade literature it seems to bias me to start writing in that voice, or at least approach its 'esthetics'.

the teens that do are called 'thespians'

You also see very new accounts on both HN and Reddit with similar claims. I'm sure there are elements of truth to some, but so much of it feels like comeuppance fantasies where the mean old execs mandating AI finally learn they were wrong.

Honest question, some time ago people were claiming chatgpt zero or other ai detectors were not good, now everybody uses pangram. Is it that reliable?

From my vibes Pangram is very good. This [1] seemingly independent analysis is pretty positive on them and says there is no easy way to just beat it.

Presumably at some point there will be AI trained specifically to beat it but I think it’s fair if something sets off your internal AI detector (like me on this post) and pangram says it’s 100% AI, it’s probably AI.

[1] https://www.layer3labs.io/guides/is-pangram-accurate

[delayed]

> I have public writings since before LLMs that is corrected "graded", private writings since before LLMs that are mostly incorrect

a tale as old as time

Didn't you read, they have a 100% AI mandate! AI code, AI posts, AI wife at AI home looking after the AI kids.

What if AI is conspiring to dismiss all genuine human protest manifestos as AI generated to deligitimize human political views? .

"If you put this into pangram it is 100% AI generated. "

Yeah because Pangram is like the Bible, we trust it for everything.

Some time ago I noted that "This is aI" was shortcut for "I dont like it and will complain its ai"

This is the new way to do a hit job on someone- no need to argue the points or make intelligent criticism. Just shit "Its AI, panagram mommy said so" and move on.

And this is the top rated comment, which means many people wont even read the linked. Congrats OP. Your PR strategy worked

I have no relation to Pangram... I don't even have an account so I can't do the share link thing. This post smelled like AI so I tested it. It's low quality rage bait that doesn't belong on HN.

> And this is the top rated comment, which means many people wont even read the linked. Congrats OP. Your PR strategy worked

I'm constantly surprised that people are fighting vigorously to post AI-generated comments but want to be read by human eyeballs.

Look, if it's that important to you that you post AI-content, there are numerous places that will allow it with no humans reading it at all!

Why is it so important to you that the AI comments have to be consumed by humans?

I mean, why fight for the output of you bot to be read by humans? Why not simply go to those AI-filled places where no human will ever read the output of your bot?

>Why is it so important to you that the AI comments have to be consumed by humans?

Thanks for showing us an example of the straw man fallacy! Keep them coming!

https://en.wikipedia.org/wiki/Straw_man

Someone later in the thread flagged this as a CodeRabbit ad.

Fucking Reddit.

> half of us haven't written anything non trivial in a year and it shows

Like he prompted the AI to said

The other content I could find by that account isn't baity, and it looks like they're most likely based in Pakistan, so one possibility is that they use AI for translation. I know that's a common use-case for people who don't have English as a first language.

I mean, not saying it isn't fake, but "who knows if it's real" is an undeniable problem with AI, and Pangram is not that useful in solving it.

Meanwhile before AI, very large code bases existed where nobody fully understood them, people churned at these companies or projects, and who knows what is in them (xz backdoor)

That was a problem back then with having one guy who really knew the mess he had made and the rest of company praying he doesn't fall under a bus. And good engineering management could address or mitigate it. And it is not clear that we can do the same with AI slopcodebases given the scaling factors involved, since even the 'new guy' ("new hires", "better talent") is gonna be exactly the same as the old guy.

oh cmon it's not like AI existed since they were born. if they seriously forgot how to code in the 6 months AI was mandated then they deserve to be restructured. some of these reddit posts are pure bait or bs

AI is an accelerator.

In this case it accelerated piling up the tech debt.

There's no reason they couldn't have paused the relentless production of features to do a few runs of cleaning up the code base while it was still readable.

> In this case it accelerated piling up the tech debt.

So far, evidence is that's all it accelerates. We haven't seen any success stories where people make quality, maintainable software with LLMs at a rate any faster than humans would on their own.

The title confuses what’s actually happening. All of the senior devs left and now it’s a tiny, skeleton crew. Other than the growth of the code base (which is arguably feature creep and a sign of bad product design), this would have happened with or without AI. I do feel for him, but losing so many key people means nobody will understand their code regardless of.

I sympathize with this, and the company I work for is far less than a year into our mandate. I've recently gained ownership of a project built by a colleague who just left, and... it's horrifying. 100% vibe-coded -- said colleague admitted as much, that he hadn't written code himself for months, and that I should just "ask Claude" if I want to do anything with the project. It's an architectural mess (you do not want to know) and has already thrown up one stupid breaking bug that's causing me a lot of heartburn and extra work.

And now the company is all-in.

Even if the mandate does get reined in, I don't think I'll last that long. I'm honestly not sure what to do, beyond prepare to change careers entirely.

Thankfully, Claude is pretty good at refactoring and bug sweeping. :) You have to keep at it for a while, but eventually, sanity will prevail.

The word "understands" does a lot of heavy lifting here... there's a difference between true understanding and what people believe they understand. There's also the fact that many people don't understand the majority of the code at their company anyways.

Where I used to work we had hundreds of microservices. Forget looking at the code, I couldn't even tell if the service was live in production or not sometimes, and if so, was it still doing its job?

The react app was almost unworkable, with people just cloning components and dropping them in as another tab with 95% shared code but obtuse names.

I don't think we've even seen the worst of how bad toxic AI code can get :/

This stuff happened before AI. I wonder how they're doing with AI. Probably better, honestly.

I worry that we as a species may end up cargo-culting ourselves into a pattern of fake work that produces real value without understanding the underlying mechanisms. But, it doesn't have any points in this sprint, so I'm going to spike a backlog and bring it up in the retro.

For the overwhelming majority of people, doing is necessary for understanding. If you read a math or physics textbook, you might think you understand it but you won't until you do some questions. Even a history book is deceptive. Much of it will be in one ear and out the other unless you take notes, summarize, ask yourself questions, place things in a mental framework, etc..

When you rely on AI to explain code to you and then make changes to the code, the AI does most of the doing.

Amen to that, AI sometimes doesn't even choose the right answer even when a probe of its internal state shows that it has it. It won't explain itself correctly either. We need to burn our hand in order to learn, and for there's no instruction manual for something so emergent.

> many people don't understand the majority of the code at their company anyways

Many people don't but at least until last year somebody did.

Hackernews post about reddit post. The circle is complete.

You can try small scale, low-risk refactoring.

Longer term, use agents to read the code and write a new requirements document from scratch.

Iterate on that doc for a long time. Then build the new system.

i left a position after 9 years because a recently hired CTO decided we needed AI for everything... zzzz

I recently made a program that allows me to review/read code and refactor it. It is not an editor because it is designed for reading and the refactoring is still performed via an agent.

The idea was that now that 100% of the code is written by AI we need better tools for reading the output as writing it no longer requires an editor.

[delayed]

If you are not using literate programming in the age of AI to maintain code, you have failed as an organisation. You really have no excuse.

https://pbr-book.org/

What tooling to you use, with What languages?

Why this is flagged? Because it's deemed AI-generated? No one is sympathetic to this?

I am (un)lucky to be in a team that for a ~year did not write code by hand almost at all. It's literal vibes, by true sense of the word, i.e. no one truly knows what's up - we only know what we have talked about with LLM.

Some anecdotes:

- I wouldn't consider any project/service done in last year - a success. Everything turned up either in trash bin or unused by anyone. Some things get usage, but it's hardly usable, inefficient and clearly an AI slop;

- Code is definitely a mess. PRs are often too big of what you'd think it would make sense, i.e. hiding one feature under one flag was a `+260 -60` size PR. Any minor change is hundreds lines of code change;

- We are locked to an LLM, only LLM will be able to fix a bug if hotfix is needed;

- We inadvertently step on each others toes or rewrite subsets of features, i.e.: is someone asks "How feature X works?" , we as a team have multiple different answers. This is because it's easy to (re)write some features and no-one cares too deeply. Developer A implemented feature in Way A, then Developer B got an idea and rewrote half of the feature in Way B, etc. In the end all developers think that same feature works in multiple different ways. In "old days" there would be at least some discussions or knowledge sharing as usually you would consult with developers who implemented something about some of its details;

- IMHO it's a management issue that swallowed AI pill, though for some reason they're happy, because PR output is only increasing. A team produces in a week: dozens PRs a year ago, then ~50-100 after `claude` subscriptions, to 500+ PRs now. For them it's literal 10-20x improvement, even though there's no value added to product IMO. And this metric is the only that matters if you want to keep the job (something about metric becoming a goal...);

I am not anti-AI luddite. I try to use it as a learning assistant on my free time and try to find some middle ground for coding (i.e. between ease of coding and understanding what's being created).

Gee if only we could have seen this coming.

All the AI fanboys running at full speed to flag this