971–980 of 1001 posts
Earlier quoted context omitted.
> No, you don't. When someone says "I feel like" and you answer "No, you don't", you're most certainly wrong :-). I do feel like the parent.
If you read parent's comment it's not "I feel like" comment even though he mentioned it. I have been in software engineering for long and the queries to stackoverflow/chatgpt combined haven't decreased for me.
Are you being serious here?
Earlier quoted context omitted.
The idea of one “accepted answer” there always bugged me. The correct/best answer of many things changes radically over time. For instance The only sane way to do a lot of things in “JavaScript” in 2009 was to install jquery and use it. Most of those same things can (and should) be done just as succinctly with native code today, but the accepted answers in practice were rarely updated or changed. I don’t even know if…
I noticed a similar thing for Python 3 questions, closed as a duplicate of a Python 2 response. Why they weren't collated and treated as a living document is beyond me.
They could go with "when in doubt, keep the duplicate", but they chose the opposite. Meaning that instead of happy users and duplicates, they have no duplicates, and no more users.
Earlier quoted context omitted.
I was writing like a robot before robots could write, dammit!
I've done that, too. It's a bit like a dream where it's not clear what's real and what's not.
Because it seems natural to me; that's how I've always written... Except, I'm also the bot. Just turn off part of my brain and an endless stream of verbiage emerges, vaguely centered around a theme... Then the real work begins: editing for relevance and imposing a coherent structure.
So, I don't really fault anyone who adopts these new tools for the task. But I have some strong feelings about the lazy editing.
Earlier quoted context omitted.
Code golfing on SO was fun, too! I just looked it up, and "Note: This tag is currently blacklisted and can no longer be used." lmfao, what a braindead site, so glad I left years ago, after many years of greatly reduced activity. Source (at the top): https://stackoverflow.com/questions/tagged/code-golf?tab=New... Look at fun stuff like this from 2010: https://stackoverflow.com/questions/2440314/code-golf-%cf%80...
You do realize that they made an entire site for CodeGolf that is reasonably active (and has its own culture... and lead to the creation of specialized languages for it... and even pushed the bounds of OEIS a few times - https://codegolf.meta.stackexchange.com/q/5318 )? https://codegolf.stackexchange.com The issue is that code golf didn't fit well into the intended design of the library and it split off 15 years ago…
>The issue is that code golf didn't fit well into the intended design
Exactly! The intended design of SO was to be a hellhole, even though it was able to stave off this fate in its infancy by virtue of having too many optimistic new users.
Some comments: - This is a really remarkable graph. I just didn't realize how thoroughly it was over for SO. It stuns me as much as when Encyclopædia Britannica stopped selling print versions a mere 9 years after the publication of Wikipedia, but at an even faster timescale. - I disagree with most comments that the brusque moderation is the cause of SO's problems, though it certainly didn't help. SO has had poor mode…
It will be the reverse I suspect. Eventually we will see that LLM quality is lower when it is training data from 2014-2020 and will chalk it up to human limitations and the data not being written with a laser-focused goal of training better AI.
Earlier quoted context omitted.
25k here, stopped posting cause you'd spend 10m on a reply to a question just to have the question closed on you by some mod trying to make everything neat. Maybe it was a culture clash but I came from newsgroups where the issue was is that someone needed help. However SO had the idea that the person who needed help wasn't as important as the normalisation of the dataset. I sometimes wonder how much time I could have…
> Maybe it was a culture clash but I came from newsgroups where the issue was is that someone needed help. However SO had the idea that the person who needed help wasn't as important as the normalisation of the dataset. Yes, because doing things that way was explicitly part of the goal, from the beginning. There are countless other places where you can directly respond to people who need help (and if you like doing t…
You say I should have stuck to newsgroups but SO sucked all the energy out of those spaces. I have 25k rep on the site so its not like I wasn't activately engaged and helped a lot of people on there, I just wish it had been more than what it was.
Earlier quoted context omitted.
Asking questions on SO was an exercise in frustration, not "interacting with peers". I've never once had a productive interaction there, everything I've ever asked was either closed for dumb reasons or not answered at all. The library of past answers was more useful, but fell off hard for more recent tech, I assume because people all were having the same frustrations as I was and just stopped going there to ask anyth…
I had the opposite experience. I learned so much from the helpful people on StackExchange sites, in computer science, programming, geology, and biology.
Earlier quoted context omitted.
Thinking from first principles, a large part of the content on stack overflow comes from the practical experience and battle scars worn by developers sharing them with others and cross-curating approaches. Privacy concerns notwithstanding, one could argue having LLMs with us every step of the way - coding agents, debugging, devops tools etc. It will be this shared interlocutor with vast swaths of experiential knowled…
Am I the only one that sees this as a hellscape? No longer interacting with your peers but an LLM instead? The knowledge centralized via telemetry and spying on every user’s every interaction and only available thru a enshitified subscription to a model that’s been trained on this stolen data?
I’ve seen this trend a number of times on HN that feels strawman-y. Taking the worst possible example of the status quo but also yada-yadaing or outright ignoring the massive risks of the tech du jour.
The comment I’m replying to hand waves over “legal issues” and totally ignores the fact that this hypothetical (and idealized) version of AI fundamentally destroys core aspects of community problem solving and centralizes the existing knowledge into a black box subscription all for the benefit of a clunky UX and underlying product that has yet to be proven effective enough to justify all the negative externalities.