5 ms·

Their thesis is that code quality does not matter as it is now a cheap commodity. As long as it passes the tests today it's great. If we need to refactor the wh

Their thesis is that code quality does not matter as it is now a cheap commodity. As long as it passes the tests today it's great. If we need to refactor the whole goddamn app tomorrow, no problem, we will just pay up the credits and do it in a few hours.

multisport 9mo ago
Yes agreed, and tbh even if that thesis is wrong, what does it matter?
- whatever1 9mo ago
  The whole point of good engineering was not about just hitting the hard specs, but also have extendable, readable, maintainable code. But if today it’s so cheap to generate new code that meets updated specs, why care about the quality of the code itself? Maybe the engineering work today is to review specs and tests and let LLMs do whatever behind the scenes to hit the specs. If the specs change, just start from scratch.
  - andrekandre 9mo ago
    > let LLMs do whatever behind the scenes to hit the specs assuming for the sake of argument that's completely true, then what happens to "competitive advantage" in this scenario? it gets me thinking: if anyone can vibe from spec, whats stopping company a (or even user a) from telling an llm agent "duplicate every aspect of this service in python and deploy it to my aws account xyz"... in that scenario, why even have companies?
    - whatever1 9mo ago
      The business is identifying the correct specs and filter the customer needs/requests so that the product does not become irrelevant.
      - ehnto 9mo ago
        Okay, we will copy that version of the product too. There is more to it than the code and software provided in most cases I feel.
      - majormajor 9mo ago
        I think `andrekandre is right in this hypothetical. Who'd pay for brand new Photoshop with a couple new features and improvements if LLM-cloned Photoshop-from-three-months-ago is free? The first few iterations of this cloud be massively consumer friendly for anything without serious cloud infra costs. Cheap clones all around. Like generic drugs but without the cartel-like control of manufacturing. Business after that would be dramatically different, though. Differentiating yourself from the willing-to-do-it-for-near-zero-margin competitors to produce something new to bring in money starts to get very hard. Can you provide better customer support? That could be hard, everyone's gonna have a pretty high baseline LLM-support-agent already... and hiring real people instead could dramatically increase the price difference you're trying to justify... Similarly for marketing or outreach etc; how are you going to cut through the AI-agent-generated copycat spam that's gonna be pounding everyone when everyone and their dog has a clone of popular software and services? Photoshop type things are probably a really good candidate for disruption like that because to a large extent every feature is independent. The noise reduction tool doesn't need API or SDK deps on the layer-opacity tool, for instance. If all your features are LLM balls of shit that doesn't necessarily reduce your ability to add new ones next to them, unlike in a more relational-database-based web app with cross-table/model dependencies, etc. And in this "try out any new idea cheaply and throw crap against the wall and see what sticks" world "product managers" and "idea people" etc are all pretty fucked. Some of the infinite monkeys are going to periodically hit to gain temporary advantage, but good luck finding someone to pay you to be a "product visionary" in a world where any feature can be rolled out and tested in the market by a random dev in hours or days.
        fragmede 9mo ago
        OK, so what do people do? What do people need? People still need to eat, people get married and die, and all of the things surrounding that, all sorts of health related stuff. Nightlife events. Insurance. actuaries. Raising babies. What do you spend your fun money on? People pay for things they use. If bespoke software is a thing you pick up at the mall at a kiosk next to Target we gotta figure something out.
    - nl 9mo ago
      You can do this for many things now. Walmart, McDonalds, Nike - none really have any secrets about what they do. There is nothing stopping someone from copying them - except that businesses are big, unwieldy things. When software becomes cheap companies compete on their support. We see this for Open Source software now.
      - gf000 9mo ago
        These are businesses with extra-large capital requirements. You ain't replicating them, because you don't have the money, and they can easily strangle you with their money as you start out. Software is different, you need very very little to start, historically just your own skills and time. Thes latter two may see some changes with LLMs.
        TeodorDyakov 9mo ago
        How conveniently you forgot about the most impotant things for a product to make money - marketing and the network effect....
        gf000 9mo ago
        I don't see the relevance to the discussion. Marketing is not significantly different for a shop and a online-only business. Having to buy a large property, fulfilling every law, etc is materially different than buying a laptop and renting a cloud instance. Almost everyone has the material capacity to do the latter, but almost no one has the privilege for the former.
        nl 9mo ago
        This is exactly my point.
    - mskogly 9mo ago
      It’s all fun and games vibecoding until you A) have customers who depend on your product B) it breaks or the one person prompting and has access to the servers and api keys gets incapacited (or just bored). Sure we can vibecode oneoff projects that does something useful (my fav is browser extensions) but as soon as we ask others to use our code on a regular basis the technical debt clock starts running. And we all know how fast dependencies in a project breaks.
  - majormajor 9mo ago
    "Write the specs and let the outsourced labor hit them" is not a new tale. Let's assume the LLM agents can write tests for, and hit, specs better and cheaper than the outsourced offshore teams could. So let's assume now you can have a working product that hits your spec without understanding the code. How many bugs and security vulnerabilities have slipped through "well tested" code because of edge cases of certain input/state combinations? Ok, throw an LLM at the codebase to scan for vulnerabilities; ok, throw another one at it to ensure no nasty side effects of the changes that one made; ok, add some functionality and a new set of tests and let it churn through a bunch of gross code changes needed to bolt that functionality into the pile of spaghetti... How long do you want your critical business logic relying on not-understood code with "100% coverage" (of lines of code and spec'd features) but super-low coverage of actual possible combinations of input+machine+system state? How big can that codebase get before "rewrite the entire world to pass all the existing specs and tests" starts getting very very very slow? We've learned MANY hard lessons about security, extensibility, and maintainability of multi-million-LOC-or-larger long-lived business systems and those don't go away just because you're no longer reading the code that's making you the money. They might even get more urgent. Is there perhaps a reason Google and Amazon didn't just hire 10x the number of people at 1/10th the salary to replace the vast majority of their engineering teams year ago?
  - PunchyHamster 9mo ago
    It's all fine till money starts being involved and whoopsies cost more than few hours of fixing.
- lacunary 9mo ago
  in my experience, what happens is the code base starts to collapse under its own weight. it becomes impossible to fix one thing without breaking another. the coding agent fails to recognize the global scope of the problem and tries local fixes over and over. progress gets slower, new features cost more. all the same problems faced by an inexperienced developer on a greenfield project! has your experience been otherwise?
  - tjr 9mo ago
    Why not have the LLM rewrite the entire codebase?
    - rcoder 9mo ago
      In ~25 years or so of dealing with large, existing codebases, I've seen time and time again that there's a ton of business value and domain knowledge locked up inside all of that "messy" code. Weird edge cases that weren't well covered in the design, defensive checks and data validations, bolted-on extensions and integrations, etc., etc. "Just rewrite it" is usually -- not always, but _usually_ -- a sure path to a long, painful migration that usually ends up not quite reproducing the old features/capabilities and adding new bugs and edge cases along the way.
      - rectang 9mo ago
        Classic Joel Spolsky: https://www.joelonsoftware.com/2000/04/06/things-you-should-never-do-part-i https://www.joelonsoftware.com/2000/04/06/things-you-should-... > the single worst strategic mistake that any software company can make: > rewrite the code from scratch.
        nl 9mo ago
        Steve Yegge talks about this exact post a lot - how it stayed correct advice for over 25 years - up until October 2025.
        9mo ago
estimator7292 9mo ago
The fundamental assumption is completely wrong. Code is not a cheap commodity. It is in fact so disastrously expensive that the entire US economy is about to implode while we're unbolting jet engines from old planes to fire up in the parking lots of datacenters for electricity.
- babelfish 9mo ago
  [citation needed]
  - fragmede 9mo ago
    you mean https://www.tomshardware.com/tech-industry/data-centers-turn-to-ex-airliner-engines-as-ai-power-crunch-bites https://www.tomshardware.com/tech-industry/data-centers-turn... ?
- whatever1 9mo ago
  It is massively cheaper than an overseas engineer. A cheap engineer can pump out maybe 1000 lines of low quality code in an hour. So like 10k tokens per hour for $50. So best case scenario $5/1000 tokens. LLMS are charging like $5 per million of tokens. And even if it is subsidized 100x it is still cheaper an order of magnitude than an overseas engineer. Not to mention speed. An LLM will spit out 1000 lines in seconds, not hours.
  - rectang 9mo ago
    Here’s a story about productivity measured by lines of code that’s 40 years old so it must surely be wrong: https://www.folklore.org/Negative_2000_Lines_Of_Code.html https://www.folklore.org/Negative_2000_Lines_Of_Code.html > When he got to the lines of code part, he thought about it for a second, and then wrote in the number: -2000
  - leptons 9mo ago
    I trust my offshore engineers way more than the slop I get from the "AI"s. My team makes my life a lot easier, because I know they know what they are doing. The LLMs, not so much.
throwaway173738 9mo ago
It matters for all the things you’d be able to justify paying a programmer for. What’s about to change is that there will be tons of these little one-off projects that previously nobody could justify paying $150/hr for. A mass democratization of software development. We’ve yet to see what that really looks like.
- inopinatus 9mo ago
  We already know what that looks like, because PHP happened.
  - qwm 9mo ago
    ahahahaha so many implications in this comment
  - throwaway173738 9mo ago
    php was still fundamentally a programming language you had to learn. This is “I wanted to make a program for my wife to do something she doesn’t have time to do manually” but made quickly with a machine. It’s probably going to do for programming what the Jacquard Loom did for cloth. Make it cheap enough that everyone can have lots of different shirts of their own style.
    - jasonfarnon 9mo ago
      But the wife didn't do it herself. He still had to do it for her, the author says. I don't think (yet) we're at the point where every person who has an idea for a really good app can make it happen. They'll still need a wozniak, it's just that wozniaks will be a dime a dozen. The php analogy works.
    - inopinatus 9mo ago
      What the Jacquard machine did for cloth was turn it into programming.
  - oenton 9mo ago
    Side tangent: On one hand I have a subtle fondness for PHP, perhaps because it was the first programming language I ever “learned” (self taught, throwing spaghetti on the wall) back in high school when LAMP stacks were all the rage. But in retrospect it’s absolutely baffling that mixing raw SQL queries with HTML tag soup wasn’t necessarily uncommon then. Also, I haven’t met many PHP developers that I’d recommend for a PHP job.
sksishbs 9mo ago
[dead]
Ancapistani 9mo ago
> Their thesis is that code quality does not matter as it is now a cheap commodity. That's not how I read it. I would say that it's more like "If a human no longer needs to read the code, is it important for it to be readable?" That is, of course, based on the premise that AI is now capable of both generating and maintaining software projects of this size. Oh, and it begs another question: are human-readable and AI-readable the same thing? If they're not, it very well could make sense to instruct the model to generate code that prioritizes what matters to LLMs over what matters to humans.

multisport 9mo ago

Yes agreed, and tbh even if that thesis is wrong, what does it matter?

whatever1 9mo ago
The whole point of good engineering was not about just hitting the hard specs, but also have extendable, readable, maintainable code. But if today it’s so cheap to generate new code that meets updated specs, why care about the quality of the code itself? Maybe the engineering work today is to review specs and tests and let LLMs do whatever behind the scenes to hit the specs. If the specs change, just start from scratch.
- andrekandre 9mo ago
  > let LLMs do whatever behind the scenes to hit the specs assuming for the sake of argument that's completely true, then what happens to "competitive advantage" in this scenario? it gets me thinking: if anyone can vibe from spec, whats stopping company a (or even user a) from telling an llm agent "duplicate every aspect of this service in python and deploy it to my aws account xyz"... in that scenario, why even have companies?
  - whatever1 9mo ago
    The business is identifying the correct specs and filter the customer needs/requests so that the product does not become irrelevant.
    - ehnto 9mo ago
      Okay, we will copy that version of the product too. There is more to it than the code and software provided in most cases I feel.
    - majormajor 9mo ago
      I think `andrekandre is right in this hypothetical. Who'd pay for brand new Photoshop with a couple new features and improvements if LLM-cloned Photoshop-from-three-months-ago is free? The first few iterations of this cloud be massively consumer friendly for anything without serious cloud infra costs. Cheap clones all around. Like generic drugs but without the cartel-like control of manufacturing. Business after that would be dramatically different, though. Differentiating yourself from the willing-to-do-it-for-near-zero-margin competitors to produce something new to bring in money starts to get very hard. Can you provide better customer support? That could be hard, everyone's gonna have a pretty high baseline LLM-support-agent already... and hiring real people instead could dramatically increase the price difference you're trying to justify... Similarly for marketing or outreach etc; how are you going to cut through the AI-agent-generated copycat spam that's gonna be pounding everyone when everyone and their dog has a clone of popular software and services? Photoshop type things are probably a really good candidate for disruption like that because to a large extent every feature is independent. The noise reduction tool doesn't need API or SDK deps on the layer-opacity tool, for instance. If all your features are LLM balls of shit that doesn't necessarily reduce your ability to add new ones next to them, unlike in a more relational-database-based web app with cross-table/model dependencies, etc. And in this "try out any new idea cheaply and throw crap against the wall and see what sticks" world "product managers" and "idea people" etc are all pretty fucked. Some of the infinite monkeys are going to periodically hit to gain temporary advantage, but good luck finding someone to pay you to be a "product visionary" in a world where any feature can be rolled out and tested in the market by a random dev in hours or days.
      - fragmede 9mo ago
        OK, so what do people do? What do people need? People still need to eat, people get married and die, and all of the things surrounding that, all sorts of health related stuff. Nightlife events. Insurance. actuaries. Raising babies. What do you spend your fun money on? People pay for things they use. If bespoke software is a thing you pick up at the mall at a kiosk next to Target we gotta figure something out.
  - nl 9mo ago
    You can do this for many things now. Walmart, McDonalds, Nike - none really have any secrets about what they do. There is nothing stopping someone from copying them - except that businesses are big, unwieldy things. When software becomes cheap companies compete on their support. We see this for Open Source software now.
    - gf000 9mo ago
      These are businesses with extra-large capital requirements. You ain't replicating them, because you don't have the money, and they can easily strangle you with their money as you start out. Software is different, you need very very little to start, historically just your own skills and time. Thes latter two may see some changes with LLMs.
      - TeodorDyakov 9mo ago
        How conveniently you forgot about the most impotant things for a product to make money - marketing and the network effect....
        gf000 9mo ago
        I don't see the relevance to the discussion. Marketing is not significantly different for a shop and a online-only business. Having to buy a large property, fulfilling every law, etc is materially different than buying a laptop and renting a cloud instance. Almost everyone has the material capacity to do the latter, but almost no one has the privilege for the former.
      - nl 9mo ago
        This is exactly my point.
  - mskogly 9mo ago
    It’s all fun and games vibecoding until you A) have customers who depend on your product B) it breaks or the one person prompting and has access to the servers and api keys gets incapacited (or just bored). Sure we can vibecode oneoff projects that does something useful (my fav is browser extensions) but as soon as we ask others to use our code on a regular basis the technical debt clock starts running. And we all know how fast dependencies in a project breaks.
- majormajor 9mo ago
  "Write the specs and let the outsourced labor hit them" is not a new tale. Let's assume the LLM agents can write tests for, and hit, specs better and cheaper than the outsourced offshore teams could. So let's assume now you can have a working product that hits your spec without understanding the code. How many bugs and security vulnerabilities have slipped through "well tested" code because of edge cases of certain input/state combinations? Ok, throw an LLM at the codebase to scan for vulnerabilities; ok, throw another one at it to ensure no nasty side effects of the changes that one made; ok, add some functionality and a new set of tests and let it churn through a bunch of gross code changes needed to bolt that functionality into the pile of spaghetti... How long do you want your critical business logic relying on not-understood code with "100% coverage" (of lines of code and spec'd features) but super-low coverage of actual possible combinations of input+machine+system state? How big can that codebase get before "rewrite the entire world to pass all the existing specs and tests" starts getting very very very slow? We've learned MANY hard lessons about security, extensibility, and maintainability of multi-million-LOC-or-larger long-lived business systems and those don't go away just because you're no longer reading the code that's making you the money. They might even get more urgent. Is there perhaps a reason Google and Amazon didn't just hire 10x the number of people at 1/10th the salary to replace the vast majority of their engineering teams year ago?
- PunchyHamster 9mo ago
  It's all fine till money starts being involved and whoopsies cost more than few hours of fixing.
lacunary 9mo ago
in my experience, what happens is the code base starts to collapse under its own weight. it becomes impossible to fix one thing without breaking another. the coding agent fails to recognize the global scope of the problem and tries local fixes over and over. progress gets slower, new features cost more. all the same problems faced by an inexperienced developer on a greenfield project! has your experience been otherwise?
- tjr 9mo ago
  Why not have the LLM rewrite the entire codebase?
  - rcoder 9mo ago
    In ~25 years or so of dealing with large, existing codebases, I've seen time and time again that there's a ton of business value and domain knowledge locked up inside all of that "messy" code. Weird edge cases that weren't well covered in the design, defensive checks and data validations, bolted-on extensions and integrations, etc., etc. "Just rewrite it" is usually -- not always, but _usually_ -- a sure path to a long, painful migration that usually ends up not quite reproducing the old features/capabilities and adding new bugs and edge cases along the way.
    - rectang 9mo ago
      Classic Joel Spolsky: https://www.joelonsoftware.com/2000/04/06/things-you-should-never-do-part-i https://www.joelonsoftware.com/2000/04/06/things-you-should-... > the single worst strategic mistake that any software company can make: > rewrite the code from scratch.
      - nl 9mo ago
        Steve Yegge talks about this exact post a lot - how it stayed correct advice for over 25 years - up until October 2025.
        9mo ago

estimator7292 9mo ago

The fundamental assumption is completely wrong. Code is not a cheap commodity. It is in fact so disastrously expensive that the entire US economy is about to implode while we're unbolting jet engines from old planes to fire up in the parking lots of datacenters for electricity.

babelfish 9mo ago
[citation needed]
- fragmede 9mo ago
  you mean https://www.tomshardware.com/tech-industry/data-centers-turn-to-ex-airliner-engines-as-ai-power-crunch-bites https://www.tomshardware.com/tech-industry/data-centers-turn... ?
whatever1 9mo ago
It is massively cheaper than an overseas engineer. A cheap engineer can pump out maybe 1000 lines of low quality code in an hour. So like 10k tokens per hour for $50. So best case scenario $5/1000 tokens. LLMS are charging like $5 per million of tokens. And even if it is subsidized 100x it is still cheaper an order of magnitude than an overseas engineer. Not to mention speed. An LLM will spit out 1000 lines in seconds, not hours.
- rectang 9mo ago
  Here’s a story about productivity measured by lines of code that’s 40 years old so it must surely be wrong: https://www.folklore.org/Negative_2000_Lines_Of_Code.html https://www.folklore.org/Negative_2000_Lines_Of_Code.html > When he got to the lines of code part, he thought about it for a second, and then wrote in the number: -2000
- leptons 9mo ago
  I trust my offshore engineers way more than the slop I get from the "AI"s. My team makes my life a lot easier, because I know they know what they are doing. The LLMs, not so much.

throwaway173738 9mo ago

It matters for all the things you’d be able to justify paying a programmer for. What’s about to change is that there will be tons of these little one-off projects that previously nobody could justify paying $150/hr for. A mass democratization of software development. We’ve yet to see what that really looks like.

inopinatus 9mo ago
We already know what that looks like, because PHP happened.
- qwm 9mo ago
  ahahahaha so many implications in this comment
- throwaway173738 9mo ago
  php was still fundamentally a programming language you had to learn. This is “I wanted to make a program for my wife to do something she doesn’t have time to do manually” but made quickly with a machine. It’s probably going to do for programming what the Jacquard Loom did for cloth. Make it cheap enough that everyone can have lots of different shirts of their own style.
  - jasonfarnon 9mo ago
    But the wife didn't do it herself. He still had to do it for her, the author says. I don't think (yet) we're at the point where every person who has an idea for a really good app can make it happen. They'll still need a wozniak, it's just that wozniaks will be a dime a dozen. The php analogy works.
  - inopinatus 9mo ago
    What the Jacquard machine did for cloth was turn it into programming.
- oenton 9mo ago
  Side tangent: On one hand I have a subtle fondness for PHP, perhaps because it was the first programming language I ever “learned” (self taught, throwing spaghetti on the wall) back in high school when LAMP stacks were all the rage. But in retrospect it’s absolutely baffling that mixing raw SQL queries with HTML tag soup wasn’t necessarily uncommon then. Also, I haven’t met many PHP developers that I’d recommend for a PHP job.

sksishbs 9mo ago

[dead]

Ancapistani 9mo ago

> Their thesis is that code quality does not matter as it is now a cheap commodity. That's not how I read it. I would say that it's more like "If a human no longer needs to read the code, is it important for it to be readable?" That is, of course, based on the premise that AI is now capable of both generating and maintaining software projects of this size. Oh, and it begs another question: are human-readable and AI-readable the same thing? If they're not, it very well could make sense to instruct the model to generate code that prioritizes what matters to LLMs over what matters to humans.