lolinder

This also accounts for the author of TFA's sense that the smartest people they know are skeptics. Assuming they're being used well, those people spend far more of their time in the high complexity work than they do in the low complexity stuff, so LLMs seem to be more flashy toys than serious tools to them.

Which is exactly why no one's hiring juniors anymore. It made sense back when the market for hiring engineers was super competitive and it was easier to gamble on being able to keep a junior than it was to try to snag a senior. But now that there are seniors galore on the market who would bother with a junior?

> but OTOH "never improve much" seems unrealistic, given the insane progress of the field in the last 3ish years

The point is that no one should hire an intern or a junior because they think it will improve their team's productivity. You hire interns and juniors because there's a causal link between "I hired an intern and spent money training them" and "they joined my company full time and a year later are now productive, contributing members of the team". It's an investment in the future, not a productivity boost today.

There is no causal link between "I aggressively adopted Claude Code in 2025" and "Claude Code in 2026 functions as a full software engineer without babysitting". If I sit around and wait a year without adopting Claude Code that will have no measurable impact on Claude Code's 2026 performance, so why would I adopt it now if it's still at intern- or junior-level skill?

If we accept that Claude is a junior-level contribution then the rational move is to wait and watch for now and only adopt it in earnest if and when it uplevels.

I'm trying to figure out what sci-fi not crossed with something else even means. Even most of the great works of classic sci-fi of the 20th century draw tropes and plot points from other genres.

Aside from what others have pointed out about fixability in post: People have pretty low expectations for most shots in most films. Every film has many mistakes in it that film buffs have fun cataloguing and no one else ever notices. Part of the art of filmmaking is knowing what must be fixed and what can be ignored, which is awfully similar to the job of a product manager.

> I don't want to have to dig into a config file for every little aspect of my OS to be in working order

When I see comments like this about NixOS I feel like I must have been using a different breed of distro before and switched to a different NixOS than you did.

My experience has been that other distros—even the supposedly stable ones like Debian—are disasters waiting to happen with no clear way of understanding what goes wrong when you do an upgrade and it breaks. Instead of one config file to troubleshoot I have hundreds of config files that I don't even know about that might be at fault. Or it could be not even a config file, it could be that there's a version mismatch that can't be resolved.

For me NixOS's config file is a breath of fresh air. If something on my system breaks it's because I made a change to that single config file. If something on my system isn't working yet it's fixable in that single config file.

The system is far from perfect, but it's much better than the black magic that other distros lean on.

This is similar to the Minecraft version of this from a few months back [0], but it does seem to have a better time keeping a memory of what you've already seen, at least for a bit. Spinning in circles doesn't lose your position quite as easily, but I did find that exiting a room and then turning back around and re-entering leaves you with a totally different room than you exited.

[0] Minecraft with object impermanence (229 points, 146 comments) https://news.ycombinator.com/item?id=42762426

The most compelling argument from that other thread wasn't that AI isn't improving productivity, it's that companies never hired Junior engineers to increase productivity in the first place! Unless you win the lottery in hiring having a Junior on board slows you down as you teach them and ramp them up until they stop being a Junior.

Companies hired Juniors as a way to invest in someone who could be a valuable asset later, without having to compete quite so hard for people with strong potential. What's changed is that companies don't have as much financial flexibility as they did before (interest rate and tax code changes) and the first thing to go is risky long-term investments in hiring and retaining talent. That takes the form of cutting back on cushy benefits and it also takes the form of cutting jobs for Juniors.

> I prefer llamacpp because doing things "seamlessly" obscures the way things work behind the scenes, which is what I want to learn and play with.

And that's a fine choice, but some of us actually just want to hack with the models not hack on them. Ollama is great for that, and SSHing into my server to reboot the process every time I want to change models just doesn't work for me.

I'd rather wait a few weeks for the newest model and be able to alternate easily than stay on the bleeding edge and sacrifice that.

> The example scenario isn't "person asks the waiter for a hot dog; waiter just keeps walking past them, blanking them as if they don't even exist".

First: that scenario is actually still just fine when the establishment is a soup kitchen. When the people providing the service are providing it for free funded entirely out of their own pocket then no user of the free service is entitled to anything, ever.

Second: that isn't an accurate description of what happened, as you should know because you read the thread in full. What actually happened is that a maintainer responded to the issue back in January with a soft timeline. OP began harassing that maintainer less than halfway through the provided timeline.

When they got ignored for just over a week (which is, again, entirely reasonable especially given that a timeline was already offered) OP proceeded to write not one but three "FUCKUJimmy" acrostics further harassing the maintainer who had provided the initial timeline.

Then, on May 9, which is well within the initial estimate, the maintainer provided an update saying that the timeline had to be pushed back. OP's reaction to that was to try to bring the project hate on HN.

In what universe is any of this behavior by OP acceptable?

Thanks, I hadn't read that before and it's a great piece!

> If most things on the menu disgust you, what on earth are you still doing at the table? The door is right over there, try not to slam it on your way out.

No, I was very thorough: I saw every comment and pulled out the obscene, aggressive, offensive acrostic embedded in months of harassing tagging of the target of that acrostic because that recent pattern of behavior overshadows everything else. There's no excuse for that entry no matter how frustrated they get.

Vercel owes them nothing—and the specific employee OP chose as a target owes them even less—and this entitled behavior is a long-standing plague on open source that needs to be called out and soundly condemned. If they need the feature so desperately they can add it themselves (Vercel already said it's something they want to implement!) or switch frameworks.

This was your contribution to that thread last month [0]. Now a month after that insightful contribution you're on HN posting editorialized titles trying to get sympathy? Given your behavior in that thread I'm just impressed at how well the Vercel folks handled your aggression.

Maybe it's time to take your own advice and try a different framework instead of wasting time trashing on one you don't like.

[0] https://github.com/vercel/next.js/discussions/64660#discussi...

> Since you're on HN, I'd recommend skipping Ollama and LMStudio.

I disagree. With Ollama I can set up my desktop as an LLM server, interact with it over WiFi from any other device, and let Ollama switch seamlessly between models as I want to swap. Unless something has changed recently, with llama.cpp's CLI you still have to shut it down and restart it with a different command line flag in order to switch models even when run in server mode.

That kind of overhead gets in the way of experimentation and can also limit applications: there are some little apps I've built that rely on being able to quickly swap between a 1B and an 8B or 30B model by just changing the model parameter in the web request.

Thanks, somehow I missed that.

https://docs.openwebui.com/license/

To be fair to our modern world this is how the Victorian gentleman's clubs were too—it's even in the name. If you weren't a gentleman you had the pub.

That he was able to go back to sleep with caffeine in his system doesn't show anything at all about the quality of that sleep.

"Aforesaid" is a very specific word that means that the "unequal impositions or discriminations" refers back to specific concepts previously referenced in the law. He can't (legally) just invent his own interpretations for what "unequal impositions" and "discriminations" entails, he has to convince a court that the specific actions he's retaliating against are covered by the "aforesaid" definitions.

Here's the complete text [0]. The act authorizes imposition of tariffs on any country that:

> Imposes, directly or indirectly, upon the disposition in or transportation in transit through or reexportation from such country of any article wholly or in part the growth or product of the United States any unreasonable charge, exaction, regulation, or limitation which is not equally enforced upon the like articles of every foreign country; or

> Discriminates in fact against the commerce of the United States, directly or indirectly, by law or administrative regulation or practice, by or in respect to any customs, tonnage, or port duty, fee, charge, exaction, classification, regulation, condition, restriction, or prohibition, in such manner as to place the commerce of the United States at a disadvantage compared with the commerce of any foreign country.

This is pretty specific. The tariffs/customs/dues/whatever don't even have to be unfair relative to what the US charges on that country's imports into the US, it's specifically targeting cases where a foreign country discriminates against US trade over and beyond the dues it charges on other countries' trade.

It'd be very difficult to prove that discriminatory treatment for each and every one of the 180+ countries caught up in Trump's tariffs.

[0] https://www.law.cornell.edu/uscode/text/19/1338

This isn't quite the framing I'd use—the key thing that would make it make sense for the deal to go this way is if there's competition bidding for the slot. Say OpenAI also wants in and was willing to pay for it. Absent competition for the slot and assuming that Telegram actually wanted the deal, we'd expect to see at most a no-cost exchange.

There's no reason for Grok to pay this much for the deal unless either Telegram sees it as a net negative that needs reimbursement or they're competing with other bids.

More work than what? More work than sharing HEAD in the monorepo, certainly. But it's definitely not more work than versioning across multiple repos because it's literally the same thing. When you're exporting code from a monorepo you follow all the same patterns you would from a small single library repo.

The argument is rarely that their lives are worse, it's that they're somehow making other people's lives worse by not paying attention to X, Y, or Z injustice. But even that argument doesn't really hold water.

I know people who are so incapacitated by their anger, frustration, and sadness about the Gaza war that they spiral into depression and are incapable of making any impact on the world directly around them. In their own words, they say that they have a hard time seeing how anything they do locally really matters when such terrible things are happening elsewhere. Their excessive amount of care about things outside of their control has actively hampered their ability to care about things that they actually can influence.

A caching proxy for Google is a real solution for a real problem. It might not be a solution to a problem you have.

Monorepo doesn't mean any of those things, nor does a polyrepo setup prevent any of them except for bazel.

It sounds like you worked in a dysfunctional organization that happened to use a monorepo. Their dysfunctions are not inherent in the monorepo model and they would have found other ways to be dysfunctional if not those.

But I think you're the first person to introduce the concept of a monolith to the conversation. How you structure your repo is an orthogonal question to how you break up your deployments, and this conversation is about the former not the latter.

A monolith that's broken up into 20 libraries in their own repos also prevents experimentation with new runtimes just as much as the monorepo version does.

Why can't you just use versioning in your external-to-the-monorepo APIs and use HEAD within the monorepo? Nothing about combining some projects into a monorepo forces you into dropping everything else we know about software release cycles.

Yep, I think this further illustrates OP's point—hobbyists building low-stakes projects get enormous benefits from LLM tooling even while professionals working on high-stakes projects find that there are lots of places where they still need something else.

This is exactly what I'm advocating for:

https://news.ycombinator.com/item?id=44103895

Why does it need to be the Internet Archive that takes big risks?

In the startup space it's common for a founder to, after a successful launch, hand off control of the organization to someone else who's better equipped to drive the company in slow and steady mode. If the people in charge of the Internet Archive feel like they're best equipped to be revolutionaries rather than maintaining the ground they won in the past, maybe it's time for them to transition into a new organization and hand off the IA to people who are able to run it in maintenance mode.

I did read that, but you don't know that that's the only way to trigger that kind of behavior. The point is that you're giving a probability drive that you don't have direct control over access to your system. It can be fine over and over until suddenly it's not, so it needs to be treated like you'd treat untrusted code.

Unfortunately, in the current developer world treating an LLM them like untrusted code means giving it full access to your system, so I guess that's fine?

> Internet Archive [important would be some pro bono legal help due to the recent turn of events, perhaps]

I have a really hard time with this one. Their current legal situation didn't just happen to them: they made an intentional choice to do something which ~everyone who knew anything about law warned them would be catastrophic and are now facing the catastrophe that they were warned about. They acted like a small activist group with nothing to lose rather than the essential piece of infrastructure that they have become.

I appreciate that the Internet Archive only exists because it took legal risks back in the day and those risks paid off. But from a strategic perspective you really have to stop being an activist organization once you've won enough ground that being ground to a pulp by the establishment would be a catastrophic loss for humanity. At that point it's time to let the little guys with nothing to lose take the mantle and just focus on preserving the ground that you've won.