jdw64

hmmmm, what I mean is this. There's been backlash about Reddit's ecosystem, but fundamentally they've transitioned into a business model that monetizes proprietary data for AI training. From that perspective, it's been successful. The problem, though, is that the quality of that data continues to decline. So it looks successful now, but I question whether that will hold in the future.

In other words, if openness and user experience are compromised, the real experts who provide deep, professional insights will disappear. Most experts have a desire for recognition. In Korea, we have a saying: 'Plankton and whales' meaning whales can only exist when there's enough plankton.

Anyway, I think Reddit itself is a successful exit. They sold their past data at a high price. But I question whether this is sustainable in the long run.

It's a difficult trade off: short term profit versus long term sustainability. Given the AI era, the timing seems right in terms of value. But separate from that, any new communities that emerge will likely suffer from long term value depreciation because of this precedent

Maybe I'm just caught up in the illusion of 'sustainability.'

I'm watching how Tao uses AI, and it's interesting.

Expand the entire expression, then change the representation to find the core axis. You can't see the axis from just one perspective, so you change the representation. In programming terms, it's like applying multiple domain models. Then break it down into small contract units. Why is it a Jacobian monomial? Why does x satisfy a cubic equation? And so on.

Then swap out the modeling under a hypothesis, assemble it all back together, and verify it through the equation.

This feels similar to modeling in programming.

Observe the whole -> explore better modeling -> decompose local problem -> verify independently -> reason about the highre level structure -> integrate back into the original problem.

This feels similar to when I receive work from a client and write a programming proposal

I wish you'd share some conversations from experienced programmers too. How do they ask questions?

American-style words are really hard these days. I thought 'open' meant source code is available, but at some point, starting with OpenAI, it seems like 'open' has come to mean 'we'll open your wallet.

Yes, exactly. Bootstrap homogenized layout, and now AI seems to be homogenizing both layout and imagery at the same time.

There are a lot of these on websites too. There was a design that was pretty good back in the day, but lately I've been getting quite a few requests to change it to the trendy Linear-style AI design. As more sites use AI-generated images, I think some people have come to see that as the trend

Personally, if you need a moderate level of connection, there's something called the IndieWeb. Of course, sometimes you just need your own world without any connection at all.

Either way, I think creating your own digital garden is a valid approach. Thanks for agreeing with me, and have a good day :)

I think it's not so much a problem with social media itself, but rather a phenomenon that occurs when consumption points converge. Anger is great for grabbing attention, and for newcomers trying to get noticed at a converged consumption point, generating anger or outrage is just the easiest way.

At first I thought it might be a problem with social media, but it seems more like a problem that arises when groups converge in one place. So these days, I just write my thoughts on my own homepage. That way I can write my own thoughts without getting angry at other people's posts.

I'm a programmer with about 7 years of experience in programming work. I appreciate hearing this from someone with far more experience than I have. senior sir.(In Korea, the word senior is sometimes referred to as one of the positions, but it is also used as a respected adult.

I also think it's about the workload. A senior from my country (Korea) once told me something similar. He had 25 years of programming experience. He said to me:

'Back in the day, inefficiencies reduced the workload. People weren't used to email, or getting approvals took a day or two. Those inefficiencies actually meant people worked less. These days, there's just too much work.'

Hearing your words reminded me of what that senior programmer who hired me said.

It's true—once you start using AI code, it becomes hard to write your own code. Because the machine's abstraction and your abstraction are different. So there are points where maintainability becomes harder.

It feels like we're chasing efficiency so much that the resting points humans actually need are disappearing.

Anyway, I agree with you, senior.

It seems like the community business itself is dying. Not just Reddit, but other communities and forums are all in decline. I thought it was because of LLMs, but they were already dying before LLMs appeared. Though LLMs did accelerate the decline.

The business model where a company owns a public community and monetizes through ads and search traffic seems to have just been a failed model.

The value of a community comes from the conversations and knowledge that users accumulate for free. But when the content company lowers accessibility under the pretense of protecting that content, new members stop coming in.

Platforms try to protect their assets, but in doing so, they end up damaging the ecosystem that created those assets in the first place.

Hmm, I've also worked with professors (from top Korean universities). them hiring me and me getting a degree from them are different things.

Well, I don't believe that just because you work hard, you'll always succeed. Most people just see me as a subordinate worker. Of course, if you come from a wealthy family, it's probably different.

Why does the AI unprecedent always unfold in ways that only benefit AI companies? You never see things like putting all code and weights on GitHub, or plastering employees' personal info all over LinkedIn. That alone shows AI is definitely smart

Seeing that the series aired on Apple TV+, it seems like Apple knows exactly what its employees are doing. Maybe they just turned what their employees want into a drama? hehe

I don't want to have a staring contest with reality.

Of course. Apart from tight deadlines and the damages I'd have to pay if I fail, the work itself is enjoyable

But I do think you can't enjoy life if you don't work. Pain is the seasoning of life. Of course, my life is full of seasoning and nothing else. Both pain and pleasure are necessary. It's just that there's only pain right now, which is the problem. If AI could just handle the economic activities for me so I could have a little bit of 'pleasure,' that would be great. I just want to do the programming I actually enjoy

Sometimes I wish AI could do my economic activities for me. There's a conspiracy theory that says if you get vaccinated, your body gets taken over by a millionaire. I actually like that conspiracy theory. Sometimes I wish someone would control me. Having a self and consciousness is just too painful.

Everyone lives in anxiety. Whether they use AI or not, people shout that their way is the right way. Right or wrong, I think it's ultimately about finding the organization that fits the workflow you want.

I use AI, but I don't feel the need to convince those who don't to use it. Conversely, if someone who doesn't use AI tells me not to, I'll just agree and move on.

I see the ongoing AI driven transformation as part of the early adoption phase of a technology, a period where it's often 'overused.' I think there will come a time when we acknowledge its limitations and adopt a more mature approach. Right now, I feel we're still in the early stage.

I don't know if the two examples in the post are real or fabricated, but even if they are, I don't see a reason to think relying on that technology is necessarily bad. Because I've seen a 30 year old woman with the intellectual capacity of a 15 year old use AI effectively. She asked about everything she didn't know, and it actually helped prevent a crime, a case of voice phishing.

Ultimately, technology is value neutral. But society's demands can reshape themselves around technology.

If that's the case, we should first question the nature of the anxiety itself. Why does a husband earning $550,000 still cling to something despite that? Why does a startup founder want even his dates to be evaluated?

It's because society demands those values and norms. From an Eastern perspective, we often say to step away from those value frameworks and realize who you truly are. But even that value framework is itself a kind of frame and a shackle.

The real question is where most anxiety comes from. It arises whenever individuals feel the need to conform to societal norms.

Both examples show that. Society demands self optimization through performance and competition, and AI simply makes that demand cheaper and more granular. The first example is about maximizing competition for promotion, wealth, and status using AI. The second example, the startup founder, treats even human relationships as improvable metrics.

We always feel anxiety in the gap between norms and our current state.

Even with a $550,000 salary, if the surrounding norm is 'a better house, higher AI productivity,' it will feel insufficient.

Even if a date was enjoyable, if the question is 'Was I empathetic and attractive enough?' then evaluation becomes necessary because you care about how you'll be perceived on her social media.

'Never Enough' isn't because desire is infinite. It's because the evaluation criteria are constantly updated externally, so we can never feel satisfied.

I understand that as a non-profit, they don't accept vibe code, and there are also server load issues. But once you start using Gen AI code, as you'd know if you've tried it, you can't stop the entire codebase from becoming Gen AI generated, because the machine's abstraction and my abstraction are different.

After receiving Gen AI code, if you try to modify it all and integrate it into your codebase, it can sometimes take longer than just writing the code yourself.

When I think about the structure, I automatically recall that when MCP calls are made, they do perform a search, but the model's tool descriptions and input/output are included together. So the list of tools becomes longer. Reflection might reduce the cost of writing wrappers, but it seems likely to use more tokens. There are also security concerns, and there are MCP host issues as well. I don't know exactly how the agent calls tools, and there's caching involved.

The reason everyone is using similar techniques for these tools is probably that the harness environment surrounding agents differs from company to company, making it hard to specialize. It's good to come up with something new, but ultimately, industry standards are what matter

LLMs are trained on large-scale text and code data, and fundamentally, they predict the probability distribution of the next token given the preceding tokens. There are studies showing that even with this simple learning objective, grammatical and semantic features, as well as relationships between concepts, emerge in the internal representation space in a structured geometric form. However, I understand it's more accurate to say that rather than the model retrieving sentences from a separate 'meaning space,' it converts the prompt into a contextual vector representation and then sequentially generates the next tokens from that state. I could be wrong about this.

Since code is fundamentally built on 'patterns' and mostly follows established conventions, it becomes easier to predict, which makes LLMs very effective for programming. After all, programming is designed to converge toward specific patterns. Frameworks with IoC are a good example of that structure.

Recently, even Linus Torvalds has acknowledged AI as a useful tool, so it's probably good to use it. That said, a lot of people still hate LLMs because they're afraid of losing their jobs

No, it depends on the workflow

I actually think bugs have decreased.

Because the programs being released these days are much larger than they used to be. With old games and old software, when they first came out, they were often unplayable or unusable. At the same level of complexity, I think there are fewer bugs than before.

Take Cyberpunk 2077 as an example. Huge projects have always had a lot of bugs. I think the overall increase in bugs is due to a combination of factors. But I don't think it's fair to say that LLMs are the reason bugs have increased. I think it's simply that as projects get larger, the number of bugs naturally increases.

In fact, as IT technology matures, projects that aren't of a certain scale are no longer commercially viable, so things are becoming increasingly complex

my homepage https://www.makonea.com

Actually, if you use an LLM translator to post on Hacker News, it gets flagged immediately. HN doesn't even allow LLMs as translators. So I use Google Translate when I need to translate sentences I don't know. (I think HN itself probably uses something like fangram on the database side to check AI similarity for new comments.)

Even translation gets flagged immediately if it's done with an LLM, so I think actual LLM-generated content would get flagged even more easily.

I don't think it's that there are more LLMs. It's that since LLMs appeared and people have recognized that LLM quality is often low and bad, more people are now calling anything that disagrees with them an LLM.

I used to use DeepL, but after getting flagged once on HN, I stopped using LLMs. HN is surprisingly strict about LLMs.

I don't know the recipe for uranium cake. Since uranium is already high-calorie enough without whipped cream, maybe I'd just recommend eating it as is

A lot of people choose Rust because of UB and memory safety, but honestly, the language structure itself doesn't have great DX. And compared to other languages, there's a lot more you need to know.

With C or C++, implementing a simple tree structure is straightforward, but in Rust, even writing a linked list is painful. The ecosystem is also fragmented. Most people use Tokio, and while AI can solve a lot of problems, AI often generates bad code just to work around the Borrow Checker.

That said, the UX experience is good—but the developer experience is terrible, and there's so much to learn. Things that would just work in other languages require explicit handling in Rust. A lot of people like Rust's technical purity, but I don't like it because of the terrible experience that comes with it.

That said, if I had to rewrite a large project from scratch, I think I would choose Rust. The benefits are that big.

Personally, I think it has both pros and cons

I use both. Claude I use up to 5, and Codex I use up to 20 for PRO.

Personally, I think Codex is better. Aside from the frequent resets, since the Fable incident, there have been cases where connections drop. For writing code, Codex generally does a better job. That said, I feel Fable is better for overall structure, but ultimately, it varies from person to person. I use both as my coding assistants

It's because of the structure of Korean. I think in Korean first, so I end up translating it directly.

When I want to emphasize something, I tend to repeat it

Wow, it displays Korean properly without breaking. But there are still a lot of typos. Haha, it's good that Korean displays properly, but there are a lot of incorrect sentences

Rather than that, the quality of human-written documentation is just too inconsistent. Because it's not written by a single person.

AI's verbosity is definitely a problem, but I think it's something you can just check once more. Of course, opinions vary.