“If you don't want to write software then don't.”
-
@carlosefr I didn't reduce it nor equated it. Again, read my posts. And yes, code is a liability, the same way a bridge with people walking, driving or riding trains on it is a liability. So, build good bridges, and build good software. Which is NOT what LLMs do. It's the exact opposite. It's creating a much bigger liability and the “I don't even review code any more" crowd are dramatically increasing that liability, future cost, both direct (fixing it) and indirect (consequences of failure).
@arroz @carlosefr There appears to be a gap between you two in the definition of programming. On one side it sounds more like the mechanical entry of instructions into the computer and on the other it is the art of converting a problem into discrete steps that a computer can understand. I see it as the later. No one likes typing instructions into a box with a glowing screen, but people do like solving challenging problems that require skill and experience to turn into digital solutions.
-
@arroz You won't see me disagree on that...
But, outside the Dunning-Kruger crowd that seems to be oozing out of every manhole crack these days, I don't think that has much to do with engineers (dis)liking programming, and a lot to do with the problems being solved.
If you're writing garbage software to solve make-believe problems, why not use a garbage generator?
I think this is one of the big drivers of LLMs among software engineers. Not the only one, bug a big one.
@arroz And I think you're right, this is creating a much bigger liability. It's another AI bubble inside the financial AI bubble.
I predict that the burst of that liability bubble is what's going to cause massive destruction of jobs in engineering in the future.
Right now garbage software has value. It provides the appearance of usefulness, creating tokens for VCs to trade, etc.
Once enough garbage exists, value will plummet, wiping out everyone working on those areas of the industry.
-
@arroz You won't see me disagree on that...
But, outside the Dunning-Kruger crowd that seems to be oozing out of every manhole crack these days, I don't think that has much to do with engineers (dis)liking programming, and a lot to do with the problems being solved.
If you're writing garbage software to solve make-believe problems, why not use a garbage generator?
I think this is one of the big drivers of LLMs among software engineers. Not the only one, bug a big one.
@carlosefr My thinking is there’s broadly two groups: the ones who just just care about the “what” and those who care about the “how”. First ones just want to ship something no matter what. Second ones care about quality and how the problem is solved. Naturally for the first ones who see software engineering as writing crap code that barely does what’s needed, that code being written by humans or machines doesn’t matter. That’s the majority of the industry, sadly.
-
@arroz @carlosefr There appears to be a gap between you two in the definition of programming. On one side it sounds more like the mechanical entry of instructions into the computer and on the other it is the art of converting a problem into discrete steps that a computer can understand. I see it as the later. No one likes typing instructions into a box with a glowing screen, but people do like solving challenging problems that require skill and experience to turn into digital solutions.
@fast_code_r_us @carlosefr This. Very good point. Because very often I see these things being evaluated by how fast code can be written. This is an old trend, not just LLM. I spend 0.5% of time, or less, actually doing the writing. So who cares.
This doesn’t exclude not writing boilerplate code should be a goal. But this is “not writing”. By using better frameworks and libraries that avoid such code. Not sweeping it under the LLM rug.
-
@carlosefr My thinking is there’s broadly two groups: the ones who just just care about the “what” and those who care about the “how”. First ones just want to ship something no matter what. Second ones care about quality and how the problem is solved. Naturally for the first ones who see software engineering as writing crap code that barely does what’s needed, that code being written by humans or machines doesn’t matter. That’s the majority of the industry, sadly.
@arroz I think the vast majority of engineers care as much about the "what" as they do about the "how" and will do the right thing when they can.
I'm more optimistic than you on that, maybe.
But the vast majority of engineers *can't*. They must live within the dystopia that the industry has become.
The industry does not care or want people that do the right thing. The industry wants people that helps it meet its goals, and those goals are just: money, and quickly.
-
@arroz I think the vast majority of engineers care as much about the "what" as they do about the "how" and will do the right thing when they can.
I'm more optimistic than you on that, maybe.
But the vast majority of engineers *can't*. They must live within the dystopia that the industry has become.
The industry does not care or want people that do the right thing. The industry wants people that helps it meet its goals, and those goals are just: money, and quickly.
@arroz I don't think this is fixable by complaining about software quality, sadly.
I don't even know if it is fixable at all. But it sure as hell could be slowed down if enough engineers stopped cheering their way to the gallows.
-
@arroz I don't think this is fixable by complaining about software quality, sadly.
I don't even know if it is fixable at all. But it sure as hell could be slowed down if enough engineers stopped cheering their way to the gallows.
@carlosefr Yes. Again, something I also keep saying: as someone coming from this country where the minimum wage is what it is, I laugh very loudly when people getting paid 6 digits, where the most significative one is often >1, argue they can’t afford to risk saying no to their management hierarchy. lol.
-
@carlosefr Yes. Again, something I also keep saying: as someone coming from this country where the minimum wage is what it is, I laugh very loudly when people getting paid 6 digits, where the most significative one is often >1, argue they can’t afford to risk saying no to their management hierarchy. lol.
@arroz Maybe it isn't about saying "no" to management. Nobody's prompting engineers to do the wrong thing. There is no one to say "no" to.
It's the incentives that are in place everywhere. Saying "no" is about fighting against those incentives and that can only be done collectively.
That's why I call it a dystopia, and that's why engineers should at least stop cheering.
They fear they'll lose those salaries if they don't cheer, but that's not true. Silence is safe (yet effective) resistance.
-
@arroz Maybe it isn't about saying "no" to management. Nobody's prompting engineers to do the wrong thing. There is no one to say "no" to.
It's the incentives that are in place everywhere. Saying "no" is about fighting against those incentives and that can only be done collectively.
That's why I call it a dystopia, and that's why engineers should at least stop cheering.
They fear they'll lose those salaries if they don't cheer, but that's not true. Silence is safe (yet effective) resistance.
@arroz To be specific: LLMs can be useful tools, but I look around and everyone is overselling them under the watchful eye of the very same people that *will* take those salaries away.
It's absurd and nauseating.
-
@arroz To be specific: LLMs can be useful tools, but I look around and everyone is overselling them under the watchful eye of the very same people that *will* take those salaries away.
It's absurd and nauseating.
@carlosefr I agree with all you said, but I think on top of the general dystopia there’s concrete things individual engineers and teams can and should say no to. Take Liquid Glass. Even if the design team crapped that out (assuming they weren't pressured to release it one year earlier to distract from Apple not being in the AI train), the engineering team should have said nope. Other examples are all the software that comes on hardware we recommend our parents not to install in any way.
-
@carlosefr I agree with all you said, but I think on top of the general dystopia there’s concrete things individual engineers and teams can and should say no to. Take Liquid Glass. Even if the design team crapped that out (assuming they weren't pressured to release it one year earlier to distract from Apple not being in the AI train), the engineering team should have said nope. Other examples are all the software that comes on hardware we recommend our parents not to install in any way.
@arroz The industry is now so massive that for every engineer that says "no", there are three engineers that will say "yes". And two of those engineers don't yet understand why they should say "no" (they will eventually, when it's too late), and the industry is exploiting this heavily.
But also, we're all on the same boat. Engineers, designers, managers, all are victims.
-
@arroz The industry is now so massive that for every engineer that says "no", there are three engineers that will say "yes". And two of those engineers don't yet understand why they should say "no" (they will eventually, when it's too late), and the industry is exploiting this heavily.
But also, we're all on the same boat. Engineers, designers, managers, all are victims.
@carlosefr No, I reject that premisse. If we actively contribute to the degradation and possibly destruction of our industry, we are not victims, we are collaborators.
-
@carlosefr No, I reject that premisse. If we actively contribute to the degradation and possibly destruction of our industry, we are not victims, we are collaborators.
@arroz It is possible for victims to be collaborators at the same time. We have seen this throughout history.
In fact, barely any crime of any significant size is possible without collaboration of a significant part of it's inevitable victims.
-
R relay@relay.infosec.exchange shared this topic
