Easy can still be time consuming. I think trying to get an LLM to spit out the whole program is a flawed approach, but I understand why you'd want to be able to spitball 12 different data models quickly.
Man, these "hot takes" on the impact of AI are all becoming so tiring. I'm especially sick of all these "code was always the easy part" missives I see everywhere now, mostly because I think they're flat out wrong.
As another comment said, "easy can still be time consuming". I've seen plenty of projects that were well defined take months in implementation time (and then still sometimes fail for technical reasons). But most importantly, if "code were the easy part", why were top programmers receiving kingly wages for over 20 years? Because business people knew the difference between a successful tech company and an also-ran usually was, in huge part, due to the quality of their software engineers. If "code was the easy part", then you go write Google Maps in 2005, or Netflix streaming in 2007, or self driving cars in 2010, or, heck, ChatGPT in 2022.
Sure, good code for a bad product still fails, but this revisionist history trying to pretend coding was so easy, so LLM-assisted coding tools won't have a big impact, is nauseating.
Oh man, I was composing my own opinion, and I can see that Hacker News is brimming with opinions. But that’s perfectly fine, I believe. All the diverse perspectives that we engage in this intellectual sparring match is a positive thing, isn’t it? :)
> But most importantly, if "code were the easy part", why were top programmers receiving kingly wages for over 20 years?
The vast majority of developers are not paid to write code. They are paid to produce, deploy, and run bug-free and efficient software systems. That does not necessarily require you to type anything at a keyboard.
Ask yourself this: how come the more you progress in seniority the less code you write?
Writing robust, bug free, efficient, maintainable, and readable code has never been easy and still isn’t. It’s an extremely difficult skill.
Are we seriously pretending that it was always easy now that an LLM can spit out some mediocre code? This seems to me a coping mechanism in response to the industry shifting.
The truth is we’re just realizing nobody, even the SWE’s, care about the code much as long as an LLM can grind it out for free.
There’s going to be a lot more of this coping as more and more human thinking gets automated. I can hear it now: “gathering business requirements was always the easy part”.
At this stage the use this kind of rhetorical structure, whereby a series of affirmatory statements (sometimes alternatively a series of rhetorical questions) are used to hedge a following "but", is so regular and reliable a flagraiser that an article will be either propagandistic, blinded-by-science, or otherwise uncritically oleaginous towards AI, I know that I can close this article midway through already knowing that the title is both the substance of the argument and also incorrect.
Hire excellent programmers who spent their lifetime learning how to write good code and build systems, and then the code is easy. Hire some bad ones and watch them struggle and fail and learn why maybe coding is not that easy. AI doesn't change this, it just shifts failure modes a bit since the volume of code produced is now larger, and AI-enabled bad programmers are still bad.
I remember a tennis survey where something like 70% of respondents believed they could win a game against a Tennis pro player (I can't find the source anymore, but it was discussed on Andy Roddick's podcast). If you watch Roger Federer effortlessly and elegantly make beautiful shots, it's very easy to fall into this trap and think -- it must be so easy, right?? Are people falling into the same trap watching Claude Code regurgitate CRUD apps for them?
10 comments
[ 4.4 ms ] story [ 25.9 ms ] threadhttps://notes.mtb.xyz/p/your-data-model-is-your-destiny
As another comment said, "easy can still be time consuming". I've seen plenty of projects that were well defined take months in implementation time (and then still sometimes fail for technical reasons). But most importantly, if "code were the easy part", why were top programmers receiving kingly wages for over 20 years? Because business people knew the difference between a successful tech company and an also-ran usually was, in huge part, due to the quality of their software engineers. If "code was the easy part", then you go write Google Maps in 2005, or Netflix streaming in 2007, or self driving cars in 2010, or, heck, ChatGPT in 2022.
Sure, good code for a bad product still fails, but this revisionist history trying to pretend coding was so easy, so LLM-assisted coding tools won't have a big impact, is nauseating.
The vast majority of developers are not paid to write code. They are paid to produce, deploy, and run bug-free and efficient software systems. That does not necessarily require you to type anything at a keyboard.
Ask yourself this: how come the more you progress in seniority the less code you write?
Are we seriously pretending that it was always easy now that an LLM can spit out some mediocre code? This seems to me a coping mechanism in response to the industry shifting.
The truth is we’re just realizing nobody, even the SWE’s, care about the code much as long as an LLM can grind it out for free.
There’s going to be a lot more of this coping as more and more human thinking gets automated. I can hear it now: “gathering business requirements was always the easy part”.
I remember a tennis survey where something like 70% of respondents believed they could win a game against a Tennis pro player (I can't find the source anymore, but it was discussed on Andy Roddick's podcast). If you watch Roger Federer effortlessly and elegantly make beautiful shots, it's very easy to fall into this trap and think -- it must be so easy, right?? Are people falling into the same trap watching Claude Code regurgitate CRUD apps for them?
Until the AI orgs need to turn a profit.