Comment by skydhash
6 hours ago
> I should not need to care about the implementation details of most software, only if it meets my retirements.
The only reason those details don’t matter to you is because someone has gone through the pain of ironing out every details that have not made it into the specifications. One one side you have the platform and on the other side you have the interface contract (requirements). Saying what’s in the middle doesn’t matter is strange. Because both the platform and the interface are dynamic and can shift drastically from their 1.0 version.
> The only reason those details don’t matter to you is because someone has gone through the pain of ironing out every details
I disagree. Very often the reason the details don’t matter is that they are irrelevant. There are a million ways an app might remember my personal settings, as a simple example. SQLlite db, json file, ini, cloud storage, registry, etc. The specific implementation matters very little so long as it’s sane.
> Saying what’s in the middle doesn’t matter is strange
I understand your point but do not agree. I think over the next decade we will get increasingly good at specifying rigorously the parts of the surface that matter while increasingly caring less and less about the rest. We will not find a way to write rigorous code in English because that would necessarily be less efficient than just using a programming language.
> There are a million ways an app might remember my personal settings, as a simple example. SQLlite db, json file, ini, cloud storage, registry, etc. The specific implementation matters very little so long as it’s sane.
It may not matter if you’re just an end user. But if you’re the one deciding the tools to be used, you may wonder about consistency (sqlite is better than a json file or ini file), availability (local storage is better than a cloud service), security risks,… Trusting an LLM to take care of that looks like negligence to me.
> Trusting an LLM to take care of that looks like negligence to me.
You are of course entitled to hold your opinion. How to work with LLMs successfully will be determined by those who believe it’s possible rather than those who argue it’s fundamentally negligent, though.
The arguments against AI coding have rapidly evolved from “it’s not possible” to “it’s possible but breaks down as soon as the system gets complex” to this “it works but it’s negligent” argument. The industry will continue to move on.
2 replies →
Yeah, I'm with you and I'm not seeing much sense in the argument above.
We should only need to specify observable behavior... but observable behavior is so broad that it includes common defects... but also we shouldn't need to specify the lack of defects even though they are observable.
It feels like, "the AI should read my mind and do the right thing without me fully specifying it."