cross-posted from: https://programming.dev/post/54810980
You must log in or register to comment.
- Repple (she/her)@lemmy.worldEnglish22 hours
This to me seems to be a very odd interpretation of the phrase “coding was never the hard part”. First, that doesn’t mean that it’s not hard, just not the hardest. Second, and more importantly, they jump to deciding what to build being the hard part as if those are the only two things?
I have used a variation of this phrase (I usually say coding was never the bottleneck at work) but I’ve never even considered deciding what to build as to be what the bottleneck is.
First, I should say by coding I mean taking a concrete design for a change and writing a single pass at implementation. Modern models to this extremely quickly with up to good performance depending on exactly what you ask them to do.
The things that take up more time in software development just for engineers, leaving aside things that design and pms do.
- Coming up with that concrete design for a change (humans may do this in many ways, one of which is writing the code without a plan and refining)
- Architecting the system at all levels. This one is continuous and huge. A good architecture will easily adopt changes as they come, but you can never foresee all possibilities, and even if you could, there’s usually other people around you who can’t, which brings me to
- Communicating the system design and coming to consensus about aspects of it. This is mostly applicable to teams and is the primary reason individuals can often outperform entire teams. It’s also one reason why throwing a second LLMs at a problem doesn’t solve it at twice the rate. You’ll never get linear scaling
- Upkeep tasks: what needs to be replaced and fixed now, how do we best monitor and alarm to figure this out in the future.
There are “what to build” aspects in here and many engineers also spend time on the overall what to build questions, but the way the article was talking about it seemed very high level.
Coding isn’t easy, it’s hard. But it’s also not usually the bottleneck in software development.
- Zedstrian@sopuli.xyzEnglish21 hours
Whoever you are, don’t outsource your understanding, judgement, empathy and taste to AI. Don’t abdicate your responsibility. Don’t be a meat proxy.
By expanding the ability to program to people without the knowledge of how that code works in the first place, it’s naive to expect that most uses of AI in programming will be informed uses by senior developers able to recognize errors and security flaws.
Such a pro-AI position also ignores the enormous amount of theft that these tools are built on.
- 23 hours
Such a pro-AI position also ignores the enormous amount of theft that these tools are built on.
Or, less likely, it could be that person knows and just doesn’t care. I certainly don’t because I believe intellectual property is bullshit.
- Zedstrian@sopuli.xyzEnglish23 hours
It’s one thing for individuals (particularly those without the means of paying) to pirate content that its original creators have already been compensated for and now exists to milk money for an IP holder for a century; it’s another matter entirely when AI companies with multi-billion dollar valuations scour the internet for content to include in their models to then profit from without permission from or compensation to the millions of people whose work they depend on.
- curbstickle@anarchist.nexusEnglish20 hours
The problem in there is corporations. AI is just the means of the dickery, not the cause.
I would also go further and say the root cause is capitalism, and that someone being able to eat or be housed should not be based on whether or not their art has met the criteria for corporate purchase. But I digress.
- Xeno@lemmy.dbzer0.comEnglish21 hours
I’m baffled I just downvoted a comment saying “intellectual property is bullshit”. There’s a first time for everything I guess. And yes it is bullshit but gosh, what a weird place to take that stance.
21 hoursYeah, it can be both, like, if you want to argue that in our current capitalist society people need to be compensated fairly for their work, I’ll hear that argument, but I don’t think it’s the best or most accurate framing. I think it’s just as obnoxious reddit made a deal with google to train on my (and everyone else’s) comments throughout the website as I do the possibility it was trained on my artwork, and I’m not alone in that.
- 22 hours
The data theft is another story, but there’s clearly an IP theft angle to anti-AI arguments, as seen in the other person response.
His other comment applies just as easily to to stolen medical images, like this woman’s , as it does more creative works. It feels violating either way. Virtually no one is worried about Mickey Mouse appearing in data sets, but lots of people do feel wronged when their emails are trained on without permission or compensation, even though they’re not really IPs.
- Jrockwar@feddit.ukEnglish22 hours
I appreciate the intent behind that sentence, and the problem is that is written (or said) by people who don’t understand coding.
Syntax was never the hard part. Code syntax is a set of rules that you can teach a machine and it’s “easy” enough for an LLM to grasp it. The problem here is that this phrase does a “code = syntax” but ignoring all the difficult things that actually are in the code. Architecture, design, tradeoffs, translation of vague requirements into concrete software components…
All of those things are terribly hard, and the reason why many LLM-first coding projects get abandoned at 70% and then companies need to hire a Software Development consultancy to mop up their mess.
- Jrockwar@feddit.ukEnglish22 hours
Interestingly what I’ve said here ☝️ ️comes with an interesting implication. There used to be a type of software “engineer” that couldn’t actually think through a problem and implement a good software solution. They were valuable because they could refactor, write code, but not think through a problem. These are the people who have the most to lose at the moment, and also the ones that find the least value on AI, because AI only does really well the single thing they could do.
- it_depends_man@lemmy.worldEnglish22 hours
Remember that story about the chalk mark engineer with Ford? Chalk mark 1$, knowing where to mark 1000$ ? Same with code. “Code” is easy. Knowing what to write, why and where, that’s the hard part. LLMs have somehow managed to automate the code part, without the abstraction, at least in an error proof way. I don’t think “coding isn’t that hard” is insulting, and I don’t use C++ because I think it’s hard. But it’s not because writing something is hard, it’s because writing precisely what is needed is hard.
Anyway, this debate doesn’t matter, negotiate better, be active politically to ban stuff or not, complaining on the internet won’t do anyone any good. I know you’re emotionally invested in this, just… breathe and do something else? Idk. Also I’m not your enemy on this, so no need to attack me.
- UnfortunateShort@lemmy.worldEnglish22 hours
I won’t miss writing code. In part because I still do when I feel like it, but also because most of the time it is extremely boring. A mere mean to a goal for which you probably have the path mapped out in your brain already. It is basically just busywork at that point.
There are these moments where you need to squeeze out performance, where you change something about the UI or generally visuals and get a pleasing result, or where you discover a small tweak that collapses a lot of complexity you originally planned for - but that is, in my humble experience, not the norm in programming. The vast majority of code is boilerplate or stuff you have already written a thousand times in some variation. The more experience you have the worse it gets.
I was ecstatic when I realized how good agents have gotten, realized how I could focus on design and architecture and with small corrections, get more than adequate software. Because for me personally, programming is the least fun part of software development. Most of the time that is. And you can hate it all you want, but programming is a skill worth nothing on its own now, and it will never be again. You can take solice in the fact that you can still do it for fun if you want to.
idk, code was always written only once, but it will always be read many times. Writing code with LLM centric way and just focusing the high level gets hazardous pretty quickly, since somebody actually needs to understand the codebase deep down (or at least the relevant part) when something goes wrong. When you have 100k+ line codebase with multiple contributors with LLM it can very quickly turn into a black box, which nobody understand anything deep down anymore and everybody are more and more incentivized to use LLMs to interact with the thing to get anything real done. Then it’s couple hallucinations away from turning into a fragile mess, where things break unexpected and nobody knows why.
