ideas can be make-or-break but it's much more common that execution is the bottleneck, and AI is the king of high-speed low-quality execution. High-quality execution can only come from running a human feedback loop through a huge number of iterations.
I like to think of the evolution of Factorio. It's been 15 years with the same core team playing their own game in a LAN party every week, as well as individually. And it's high quality.
Dogfooding led them to counter-intuitive decisions like, if a flying robot runs out of battery, it should obviously crash into the ground and explode, but it turned out way better for gameplay if it just limps slowly to the nearest charging point instead. This fact and many more can only be discerned from actual gameplay experience. No amount of pure linguistic reasoning will tell you them. Another one is how the burner inserters (coal-powered robot arms) fuel themselves - in the first version, they didn't. They could have made running around to refuel your robot arms a part of the game, but that isn't where they discovered the statue within the marble. You had to try the game both ways and see what feels more fun. A third: filter inserters and smart inserters were removed, and their features were moved to all inserters. A fourth: trains originally ran on strict time-based timetables, but it was frustrating so they were made able to move based on logistics needs instead. A fifth: basic oil processing used to produce all three oil products, and now it doesn't. A sixth: the basic oil refinery has its liquid I/O ports set up in a way that's compatible with switching to advanced oil refining later, instead of requiring you to redo your pipes.
All these little things you get from actually using the software and streamlining it, those are the quality. The list above obviously doesn't translate directly to any other project, but something like Microsoft has its own list, probably starting with things like common controls, dialog tabbing, and drag'n'drop. Did you know the reason you click "Start" to shut down in Win95 is because they did actual user testing and that's where the users looked first?
And most importantly, it doesn't matter how much AI you throw at a problem, it's not going to be able to figure out things like this. Because it doesn't have the experience of being a human.
... and politics! To me LLMs are just another piece of evidence that it was never the lack of resources that made Microsoft produce bad user experiences.
Stack overflow was one of the best ways to learn and mature as an engineer “back in the day”. The
meme is that devs would copy and paste and hope it worked. The reality was more like you had an obscure bug, you wrote a detailed and cohesive question, and you connected with people who were more knowledgable and would help you.
It also sharpened your writing and thinking. By responding to questions you solidified your understanding of topics.
It got a lot of hate for the moderation but I think that was mainly in huge communities where literally “dumb” questions got bombarded every minute. Smaller tech like Docker when it first arrived on the scene had communities that learned together.
Spent a lot of time there when I was a few years out of school and still trying to figure out what it meant to be a software eng. It’s sad to see it become a ghost town and I think it’s a net negative for young engineers to have an all knowing entity spit out every answer.
You touched on what made it great, along with other forums where devs helped each other: It was a pleasure to read and try to answer other people's questions. NOT for the points! Because you found out so much about the problems other people were trying to solve, parallel to your own. The good questions forced you think what would I do in that situation and become a better coder yourself.
I may have asked 20 questions on Stack Overflow in 20 years. But I probably answered several thousand. Whatever I was deeply interested in at the moment, there was no better way to learn more than to try to answer those tough questions. It was like a hobby, during a few minutes of downtime, to just cruise around looking for database optimization problems... the way some people might pick up a Sudoku puzzle. It would keep me in the zone.
A lot of the mods sucked, and they had to deal with deluges of spam from kids asking for answers to CS exam questions... but hunting for the really juicy questions that couldn't be answered with a search was part of the fun.
I mourn for the generations that will never get to use their brains.
> I mourn for the generations that will never get to use their brains.
If you aren't using the incredible technology we have around us right now to accelerate your learning then it is us that should be mourning your loss of imagination.
Can you name anything you've learned from an LLM? Start your essay with what you've learned from AI as your opening paragraph. Write by hand as if you had been educated to do so. List one or more specific instances in which an LLM gave you knowledge which you would take forward and use productively, as your own integrated knowledge, in such a way that you would be consistently able to apply it to novel problems and circumstances on your own.
I won't down vote you, because you simply misunderstood me. Of course I'm using this incredible technology to accelerate writing code.
If you never learned the value of learning itself, for its own sake, you have everything to gain from a world in which skills like learning and thinking for yourself are useless. I can hardly blame you for that.
But as someone who has learned by teaching others, people who in turn will learn by teaching others, I don't find your belief in someone else doing the work for you to be particularly sustainable. Or convincing. Regardless of whether I think it's contemptible, lazy, or damaging to your mind, which isn't my problem.
> Can you name anything you've learned from an LLM?
Not GP, but here’s some of the things I have learned in the last couple weeks from LLMs while asking LLMs to review parts of my code and documentation. I wrote the code and documentation without LLMs.
Linux user permissions is a fake because by design, the owner has permissions to write these 3 bits. A good way to reliably deploy a readonly file is owner=root, assign the read permission to the group.
chown command follows links by default, -h switch disables that.
In Linux, resetting environment variable in a long-lived server process doesn’t erase much data because /proc/{pid}/environ is a snapshot taken on process startup.
If my server crashes (or manually killed by a root to test things on staging VM) and Alpine’s supervise-daemon restarts it, the start_pre() function from the OpenRC service script won’t be called again. A fix is use OpenRC to run a launcher shell script as root, in that script manually downgrade the account with `exec su-exec svc:svc /the/server/elf`
This last one is irrelevant to my use cases but still interesting. AES GCM may fail catastrophically when different message encrypted with same password and nonce. For 12 bytes of random nonce, the stuff is only safe up to approximately 4E+9 messages using the same password, beyond that birthday paradox ruins things.
The LLM is citing it's sources and I follow up on them. Your attempt at a gotcha is interesting in that I think it says more about your relationship with AI than mine.
I find asking a question about whether an application should be doing what it is doing, and getting someone's braindead response of "Claude says it does this..." shows that people aren't learning. I know what it is doing, I can read an IF statement, I don't need someone to spit out 3 paragraphs of unrelated nonsense.
With Claude people aren't learning. They aren't even reading their copy/pasta response to see if it relates to the question.
It makes me really sad that my brother, a bit of a celebrity lawyer and one of the smartest people I know, now frequently responds to texts with long-form garbage he gets from ChatGPT. it's somehow both smug and braindead simultaneously, and it's cost a lot of my respect for him.
To be honest, he used to paste a lot of links to articles that were only obliquely related to the topic at hand, and which I often suspected he hadn't fully read.
Thinking about it now, I think the new thing of copying LLM output is worse for a couple of reasons. Although both are a sort of vampirism that sucks all the oxygen out of a text chain, at least pasting articles implies that one has read that article and agrees. Someone can be held responsible ror the selection. With LLM output, there's an implied shrugging off of any responsibility... it's not "here's something I read and agreed with," it's just "here's what ChatGPT said." There's less implicit effort, and less risk of staking one's credibility on something, because saying "hey, that article is junk science and these studies disprove it" would require a response beyond "well, ChatGPT can be wrong".
Secondly, and maybe more metaphysically or epistemologically, there's a certain decline in cognitive function that I associate with pasting LLM answers that goes beyond what was previously considered a lazy pasting of articles someone had speed-read or skimmed. (It also doesn't help interpersonally that my brother is a speed-reader who can form eloquent opinions on a document within seconds, although apparently it's fabulous in trial).
So you're right to point that out, but it's also a red flag to me that some sort of mass lead/heavy metal poisoning is happening to our society, to see both credulity and shirking of responsibility from one of the smartest people I know.
I'm just rambling now, but I also have a cousin who was an engineer for years on the Supercollider in Texas. Also a very intelligent guy. But during covid, he had retired by then, he started spamming the whole family with emails insisting that covid was actually caused by 5G radio waves and that vaccines were poison. I outright attacked him in a full family-wide email, telling him that I had lost all respect for his once-great mind, which I had formerly looked up to.
Somehow I feel like the lack of analytical discrimination, logic, and above all effort is the common thread here.
My experience was different. I'd post on StackOverflow as an absolute last resort when I couldn't solve my problem any other way. The knowledge there was great, the kindness was sorely lacking. That's what ultimately did it in if you ask me. It got a (deserved in my opinion) reputation for unkindness way before LLMs came on the scene.
I think that’s why Reddit and discord haven’t been harmed nearly as much. They have a reputation for being a lot kinder than stack overflow for basic questions. Meanwhile stack overflow was dead pretty much immediately after people got an alternative.
I think StackOverflow was actually dead before ChatGPT appeared. But ChatGPT was the final nail in the coffin, for the reason you mention.
And there was a time when I enjoyed answering question on StackOverflow but that time ended long before the end of the site. It started out kind of competitive but ended insanely competitive to the point that answering questions was a waste of time.
I can say that easily over 90% (pretty close to 100%) of the most useful answers I would find in stackoverflow back in the day, were from questions/threads that were locked as bad questions (forget what term they used) by those horrible moderators.
It was literally a joke. Every good question with thoughful useful answers, was always poo poo'ed by the moderators.
The only reason stackoverflow remained useful is that those questions and threads remained available and were a treasure trove of useful information. If they'd actually deleted the threads moderators declared useless, there would have been zero useful content on the site.
It is really a monument to failed moderation policies.
>The meme is that devs would copy and paste and hope it worked. The reality was more like you had an obscure bug, you wrote a detailed and cohesive question, and you connected with people who were more knowledgable and would help you.
Au contraire, the reality was more like you had an obscure bug, you wrote a detailed and cohesive question, and it was immediately closed as a duplicate of something unrelated but with a superficially-similar name from 5 years prior, and then you were threatened with a ban.
StackOverflow was easily one of the most toxic communities on the internet and I don't miss it.
> I think it’s a net negative for young engineers to have an all knowing entity spit out every answer.
I agree with the rest of your comment, and I sympathize with what you're saying here, but really this only applies when measuring engineers against a yardstick that frankly is rapidly in the process of no longer existing.
As an industry rather than an art form, software is measured by its output.
You sound like someone who is enjoying the retirement benefits and full social safety net. One which does not have the funding to continue on for the youth of today. You have a certain smugness about your current situation but maybe you should look at how things are actually coming along today for young people staring down the barrel of higher inflation and lower wages.
“we should never try building new things because they get peoples hopes up” this is such a moronic attitude to have that seems more and more common. its the harrison bergeron-ification of the world
We can rewrite existing libraries and software all we want, but the true test is if we can create something new.
The thing I learned about myself with AI is that now I have time to work on every idea that pops into my head —- my ideas aren’t that great.
reply