ravenstine

Man, this year has been nothing but bad news for me. This is disappointing, and really makes me question what is the point if even open source is now this easily enshittified or "killed by [insert big tech co here]." I've spent years now using Deno nearly exclusively for new projects, and saw it as the most sane JS server runtime. This also makes me hesitant to use anything with Ryan Dahl's name on it ever again.

Yes, I will never give up my old Brother laser printer. Prints fast like a champ. It's one of the most no-BS devices I've ever owned. I only print occasionally, and I don't think I've ever had to change the toner cartridge in the 13 years I've owned it. I don't care if newer printers are "better". I'm not gonna pay any of these companies for their ink-wasting DRM-protected cartridges.

I'd be hesitant to share anything I wasn't willing to lose. However, I started feeling that way long before the dawn of the current AI age.

The excitement over the little wins has largely evaporated. I think this is more of a reflection of what AI has done to the coding process itself. It used to be that you could take pride in relatively little things, like improving the performance of a single function that makes an application more responsive. It didn't necessarily have to be exciting to everyone, but you could feel that satisfaction. Now we really don't even get to have that sort of satisfaction. AI forces us to focus only on the big picture. When people don't feel the satisfaction over their own little wins, it's hard for them to imagine why anyone else would.

HN is still one of my favorite sites, though. Even when I get downvoted, at least it's primarily by people who can understand the subject matter. Every other site has people absolutely talking out their ass, especially when it comes to... ugh... AI.

If I was a much younger man, I would gladly spend weeks or months out of my life to try and break a Denuvo protected game. haha

(again, for the mere intellectual challenge, not to take profit away from someone's work)

I wouldn't dismiss Denuvo. It is intended for a particular use case, which is games where a week or more delay between the release of a game and the appearance of a crack would make a meaningful difference. The payoff wouldn't be nearly the same for a lot of other software.

Both fortunately and unfortunately, this situation may change as AI models and workflows tailored around reverse engineering improve.

Just what I need in my life; more YAML! /s

I'm not sure why anyone would want this. Is there something this actually does better than any other harness? Do current harnesses suck that bad at orchestrating things like Docker? If I do all my work inside a Linux VM and tell an agent inside of it what I need done, it figures out everything, including container orchestration.

The very job that AI gents are meant to do should mean that most of the documentation for this Docker Agent is obsolete/unnecessary.

So glad I never had to deal with Angular, especially v1. I tried learning it and it just seemed like it would actually make applications more incomprehensible than before. React originally addressed the actual problem that most frameworks mistook for not being that important. I still don't particularly like React, but it makes sense why it became popular.

It's all about what story you can tell.

Cursor is easily the worst experience I've had with an AI harness, yet it was acquired for billions of dollars in spite of being a middleman tied to a ripoff of VS Code. None of the colleagues of mine who were touting it last year are still using it. But if you can make yourself look like the next big thing, you'll get money thrown at you. Just look at Omarchy.

A good experience necessarily takes time and careful thought. Nobody these days can do that while also remaining relevant, or even surviving.

Yeah, I'm sure they're out there, but it's very rare in my experience. I'm a micro-brain when it comes to cracking software, and I've almost never encountered this, whether it's small fry software or something from Microsoft or Autodesk. There was some software with debugger detection I came across once (can't remember which it was), but that's the kind of thing where lots of existing workarounds often exist by other crack-ers.

Techniques like what you describe made a little more sense when the means to even figure them out were less feasible for the average person, especially before the web had much information on reverse engineering, when powerful debugging tools weren't as accessible or free.

Today, there is little point in trying to slow down software pirates. At best, adding an arbitrary piracy detection only adds anywhere between mere minutes and a few days to the effort to crack software. This is true absent AI assistance or even a meaningful understanding of ASM outside of logical JMP instructions. The author will likely waste more of their time implementing anti-piracy techniques than a software pirate would figuring out which function call results in the program exiting abruptly. I've yet to encounter a program where a single flipped JE/JNE or NOP couldn't unlock most or all capabilities. This is in spite of various licensing and contextual checks throughout.

It would slow down a pirate more to have a program modify or decompress itself in memory, but that class of techniques is still more trouble than it's worth. Experienced pirates already know how to deal with those traps. The timeout thing you mentioned is clever, but the type of person who knows enough to disassemble software would think to themselves "wtf does it crash after 5 minutes?", immediately investigate, and identify the source of the crash.

Having a license check is the only thing authors of software should bother with. It provides most people a framework to consider whether they should pay for a product. Most people won't download potential malware from a sketchy website if you offer your product at a fair price. Those who either know how to crack apps or refuse to pay will keep doing what they're doing.

tl;dr Don't fool yourselves into thinking you'll outsmart a kid with Ghidra installed by throwing a glorified if-statement in their path.

EDIT: I'm speaking in the general sense. The same principles apply to an app that runs almost all of its logic in the browser.

I'm not sure where you're getting the "worshiping" part from. Though I respect your overall opinion while still disagreeing from it, it's been a very long time since HN came anywhere near worshiping Apple or even most big tech companies. Most of us simply live in the real world where using big tech wares is a soft-requirement. Using an Apple device or having interest in them makes one no more or less a "hacker", which is already a pretty subjective and contextual term to begin with.

The problem with giving precedence to documentation over memory is that the documentation has to actually exist and it has to be the most accurate representation of reality. As soon as documentation becomes fragmented and out of date, memory of some kind becomes a necessity, even if this means an agent writing their own documentation as memory. I've yet to work on any team where the documentation was even close to being reliable enough without the need for cross-checking, asking those with intimate knowledge, and ultimately interim memory to make sense of it all. This is why "the code is the documentation" can often work better for agents than actual documentation written by/for humans. Humans have proven time and again to not care about documentation unless there is a profit motive for said documentation.

> I have no idea what type of job mini can do that nano couldn't or vis-à-vis.

In my experience, Nano won't reliably handle complex open-ended tasks and is mostly suited for very explicit instruction that it can't screw up. It's no different from how there are some chores you can give to kids and there are other tasks you need at least a teenager for. If the decision tree of the task is very clear and conventional, Nano can be cheaper than giving the task to a relatively overpowered model, especially if it's something where the output is rigidly structured. This makes it well suited for skills that essentially run CLI commands and generate output, especially because it is usually faster. Mini is more like a discount version of the base model, and Nano is the dollar store version. Mini is more of a generalist and a fairly good deal if you have a moderately complex task that is conventional, but can be less conventional that what Nano can handle. I mostly used gpt-5.4-mini this year for my side projects because it's a pretty good generalist while significantly saving on costs. It is, however, somewhat dumber than the base model and more prone to ignore or forget rules you give it. I'd have just used a base model, but the low cost of Mini and Nano made them appealing to me. Maybe I'm a cheapskate, but I have hundreds or possibly thousands more in my pocket than many other users because of that.

This workflow I settled into with Mini and Nano didn't map cleanly on to the current generation of model tiers. With the price of Luna, you'd think it would be a replacement for Nano. In a sense it is, yet I didn't find that Terra became the new Mini. Terra is more powerful, better at explaining its own decisions, yet I've also found it to be relatively stupid while charging me more to use it. On the other hand, Luna with its reasoning set to "high" is what I consider to fill the role of Mini, and is good enough such that I no longer use Mini. Sol and Astra are great, but they're pricey. It could be my own brain and its bad perception, but so far I don't get the point of Terra. Luna succeeded at reverse engineering some abandonware with a very complicated licensing and virtualization scheme, and did so over SSH into a Windows VM with only PowerShell on the other end. Terra did such idiotic crap to my flashcards app that I stopped using it for anything after that.

This is why I find OpenAI's naming unhelpful and kind of pointless. I don't really care about the benchmarks that all these models are commonly run against. They're not that useful, IMO. OpenAI could easily give early access to these models, get a ton of feedback, and provide better insight to customers on how these things behave. Even calling Terra "gpt-5.6-overpriced-cheating-dumbass" would be better than wasting my time and money figuring it out myself. But that wouldn't make OpenAI as much money.

This has not been my experience. Obviously there's no way I can disprove a placebo effect, but I don't find myself to be particularly prone to placebo effects or magical thinking. The vast majority of the time,, I don't notice much effect if at all from the "amazing" psychoactives, nootropics, pharmaceuticals, supplements, etc., short of full blown psychedelics and stimulants. Perhaps I'm prone to the nocebo effect. Then again, if I nearly always have the nocebo effect, it would seem odd to me that suddenly I would have the placebo effect with patches. Is it like adderall? No, but I would be stunned if researchers performed a double-blind study of patches on me as the sole subject and couldn't find a measurable difference in my behavior in terms of task persistence.

They'll shut down servers and close your account with no reason provided outside of demanding you add a phone number. It's happened to me a few times, and I was squeaky clean on there even when compared to the way I talk on HN. Didn't matter whether I provided a phone number. I'm not the only one it's happened to. Anyone with friends on Discord should make sure to write down their contact information in case Discord one day decides that they or their community doesn't exist anymore. But do they delete your data? No, of course they don't... they keep it all to themselves!

> Helps if/when polluting context with a zillion (mostly unnecessary) tools.

Ugh, I can't help but respond to this one point. The fact that this is even an issue in the current year just tells us how screwed the software field is in a lot of ways. I don't mean that in existential terms, but of how divided we've become in terms of what's happened to human reasoning. On the one hand, you have people who apply deep thinking to develop the sort of approaches you described, and there's the exponentially growing segment of not-even-programmers who seem to never ask themselves whether any of their ideas have any sort of consequences.

Take MCPs for instance. Sure, I guess it can sometimes make sense to have a stateful API that is optimized for agents. Yet, more often than not, these MCPs frontload a ton of context where it's not needed, and solve problems where none existed. Merely sticking an API (MCP) in front of an API (CLI, REST, GraphQL) without a benefit that can be explained in a single sentence is lunacy and demonstrates a real lack of complex thinking.

I totally agree. At the same time, nobody should be using Discord anyway, for a variety of reasons. They're one of the least trustworthy online platforms I can think of, and have unfortunately encouraged online ecosystems where you have to "Join our Discord" in order to maybe be heard. I would be delighted if age verification encourage more people to abandon Discord. In a better world, companies like that wouldn't be allowed to become zombie corporations but instead be forced to go out of business.

I've come to a similar conclusion.

On one hand, yes, nicotine in many ways does deserve the reputation it's received. Then again, a lot of the blame should go more towards smoking rather than nicotine the substance.

Nicotine is highly addictive, but as you mentioned, the delivery method plays a huge role in its addictive potential, and most people don't realize this. The chances that you'll become addicted to patches is incredibly low. If they were that addictive, then they'd likely be way more effective in helping people quit smoking; in reality, it's somewhere between 10 and 20 percent of patch users who successfully end up quitting smoking that way. Even with the highest dose patches, the addiction potential is still relatively low because the patches release the nicotine over a very long period of time and make it difficult or impossible to associate the use of the substance with a ritualistic activity like lighting and puffing on a cigarette.

And yeah, it's not a subject you can really bring up with anyone besides recreational drug users. Very few will be receptive to the idea that nicotine has legitimate uses and can be used without getting addicted.

I, too, have experimented with the patches over the last year. Nicotine is great in a lot of ways, but whether it's a tool or a crutch depends on both the person and the delivery method.

The patches are different, not just because it lacks that same connection to compulsive and ritualistic activity, but it simply doesn't provide the same quick "hit" that makes smoking far more addictive.

On the other hand, someone prone to addictive tendencies probably shouldn't even use the patches. Even the patches can be a gateway to vaping and smoking for such a person. I have simply never been the type to become addicted to any drug or substance, other than the occasional overeating. This isn't to say that I don't have the potential to get hooked on more powerful drugs, but my innate desire to use them is a lot less than I think is average. Even during times where I've habitually drunk copious amounts of alcohol, I've always been able to realize "this kinda sucks so I think I'm gonna stop" and then I just do without really much struggle other than some brain fog for a few days. Similarly, I sometimes use phenibut, and people talk about how easy it is to become dependent on it and how hard it is to get off it, and that hasn't been my experience in the slightest. Someone without a clear objective around how they use substances like that will be much more likely to end up suffering.

That said, the patches are great for focus and anxiety when used in low doses in moderation. I don't necessarily advise that everyone use them as I have, but I also don't think they're anywhere near as harmful as one might assume. I don't feel the need to ever switch to smoking or vapes, or even gums and pouches.

I'm impressed that Luna got as far as it did on max reasoning. Don't get me wrong, that's one weird looking ass, but even the top GPT model a year ago probably would have done worse based on my experience (with asking it to create meshes in general, not modeling asses... yeah, that's the ticket). Then again, I'm also not, because I've found it to be a great deal at its relatively low price point when the reasoning is set to high.

Without fail, these vibe coded project websites all make themselves look way more ambitious and established than they actually are. A software project could be hours old and it will have a website that, just a handful of years ago, only projects that had been around for a while and with a revenue model would have had. To me, it comes off as so overly persuasive that the only thing I can think nowadays is "what sort of half-baked junk are they trying to sell me?" Today, a primitive website or a mere handwritten README.md is, ironically, a more reliable signal of whether an application is worth considering. At least to me, anyway.

The idea, I think, is to provide a sense of GUI unity that Windows and Mac mostly provide. While it's nice in principle, the reality is you're going to install other software, and it's often going to use GUI libraries that don't perfectly match or match at all with your Linux DE. Personally, I've never found the need for a system to be designed so heavily around a DE. Like you said, it's entirely feasible and relatively straight forward to download Debian or whatever distribution and install a DE and chances are it will work ask expected. Well, until you perform an update/upgrade and boot to a black screen, but that's a different story. Once you've got your kernel and init system of choice, you can pretty much do anything you want. A premade DE is hardly even needed if you have a modest familiarity with how Linux works. The only downside is the added work from setting it up and debugging it, but people who use Ubuntu, Gnome, KDE, XFCE, MATE, LXDE, and so forth, end up doing that anyway eventually.

Seriously! Though I prefer GPT models to other frontier models, this shit is confusing. They keep changing the names of these models and they often don't communicate anything meaningful about the model itself, especially with these latest iterations. At least with "mini" and "nano" you understood they generally had differing speeds and "reasoning" capability, but what the hell do "Terra", "Sol", and "Astra" really mean? Which one of them is the effective successor to gpt-5.4-mini? It's hard to tell since the only objective information you'll get is token pricing. Is Terra less capable than Luna because it makes me think of dirt and grass? Or is Luna less powerful because the Earth is bigger than the Moon? Apparently that's the real answer. And why do I even have to think about this? And what comes after Astra? Galactica? Or will they start naming the succeeding models after different candy bars? Should I even care since a new model will get farted out mere days after I figured out what differentiated the last one?

What's unclear to me is who OpenAI thinks they're marketing to with this form of branding. These different models don't really mean all that much to the vast majority of people using their products who aren't developers, and developers aren't helped at all by the way they've been naming said models. Are they merely scared that they'll become irrelevant because Anthropic decided to give their models quirky names like "Opus" and "Fable"?

If OpenAI really wants to give their models names, they should name the generation of model and then have the different sub-models named by purpose or capability level. After all, I wouldn't use Mini for a job that Nano could easily do, and I wouldn't use Nano for a job that the full version of GPT-* necessitates. Similarly, I've had to discover exactly how Luna, Terra, and Sol are appropriate for different complexities and task types. OpenAI could help me skip a lot of those steps and just tell me what each model distillation is good for without causing me to look through their pricing page and make educated guesses. After all, shouldn't they not want me to pay attention to how much they're charging me?

All of this makes the days of frontend framework churn seem quaint and actually preferable.

Yeah I hate the term "sideload", but I'm guessing they meant it in the sense that they installed a separate DE within a distro designed around its original DE. Ideally, the DE that ships with a distro should be able to solve problems like video streaming and KiCAD with relative ease in the year 2026. Although I'd be curious as to why exactly that person decided they needed to change their setup in that way, I also can't say I'm entirely surprised that they would end up needing to do something like that.

It's been ages since I've looked at System 76. They have quite an impressive lineup of products now! I'll definitely be checking them out the next time I want a personal computing device that's not tied to Microslop or Apple.

It's things like this that make me imagine what kind of Linux desktop we could have if one of them had the kind of budget that goes towards Windows and macOS. The kernel itself is great and tons of money is effectively invested towards it and its ecosystem. In my humble opinion, it's the Linux DEs that sometimes get really close to being as equally awesome but never quite get there. MacOS is by no means what I consider a perfect OS, but it's reliable enough that I haven't had to think to myself "what if my computer stops working tomorrow because reasons." I know it happens to people, but it seems relatively rare in contrast to the alternatives. The one big exception might be the removal of Rosetta, but developers and consumers have had an abundance of time to migrate to Apple Silicon. In general, macOS respects me enough such that it isn't constantly trying to make me adapt to new ideas. You know something? I hate that I'm practically kissing the rear end of a gigantic corporation that does other things that are anti-consumer and anti-developer. I don't worship Apple. I think GrapheneOS is better than iOS. But I appreciate that their desktop OS causes me the least number of frustrations. And speaking of GrapheneOS, it essentially benefits from Google's budget towards AOSP, and Linux DEs don't have this advantage. Imagine if they did and we had a Linux DE that was as solid as GrapheneOS, and without the BS that comes with Android.

It's funny because I wrote my comment this morning and then an hour later my manager told me unprompted about how his work Macbook updated over the weekend and got stuck in a boot loop! I have a feeling the blame might actually be on some IT department configuration going haywire, but it could be Apple's fault for all I know. That was an amusing coincidence.

And yeah, "It works – until it doesn't" is probably the most accurate statement that can be made about all technology. My original comment also can make it seem like I'm taking a dump on Linux, but my gripe is more about Linux desktop environments. Linux absolutely can be a really stable OS, but I've only been able to truly reach that for myself by compiling my own kernel and composing my own desktop environment from barely-related packages. This is where I totally relate to your sense of "fuck it, I don't care because it's mine." Maybe someday I'll end up completely switching to my own Linux OS.

> Also the OS mostly “just works”

A large portion of the HN crowd woefully underrates this aspect of macOS. The types who would get pleasure tweaking their OS and understanding it intimately are overrepresented here, though props to them.

I don't care what people say; macOS is the only OS I've ever used that has never booted to a black/blue screen with relatively cryptic error after performing an upgrade. Not once ever. Windows today is of course abusive to users and shouldn't be used, and as much as I absolutely love Linux as a server OS and for VM appliances, I get no joy whatsoever trying to figure out why the DE stopped working after letting the OS update itself or debugging subpar rendering from the Wayland compositor or X11. (Linux people still tell me this isn't a problem in 2026 but, guess what, my Ubuntu VM hit a black screen a couple days ago after upgrading with no other changes made to the installation, so they're wrong) That and so much graphical Linux software is written with the attitude of "I'm a freak who thinks context menus shouldn't exist so fuck the user and fuck anyone who tells me I'm wrong", yet so many DEs integrate these apps anyway.

The worst thing that macOS has ever overtly done to me is slightly change the look of its interface over the years, which is incredibly minor compared to all the changes that Windows, Ubuntu, Gnome, and KDE have gone through in the last decade. Does it pull lame crap like phone home? Probably, but these days I assume that just about all software thinks it knows better than I and constantly pings cloud servers. MacOS is not really unique in this regard. I will switch to my own Linux if Apple goes down the road of "use our AI or else", but for now I don't really see the venom towards macOS to be justified when it's incredibly easy to prove that the alternatives are usually very poor.

Oh yeah, and the hardware is excellent. (except for the keyboards which need TPU dam covering them or else dust kills them after a couple years)

It's kinda nuts to me how people can act like Claude is lightyears ahead of OpenAI models. Sure, it's one thing to simply have a preference or claim that Claude does some things better, but in reality they are both about as effective at doing the same job. I've long preferred GPT models because they know better how to shut up and don't seem to overthink as much as Claude, but I'm under no illusions that if OpenAI went belly-up then I couldn't do my job essentially the same with Claude. GPT models have also clearly improved over time in terms of programming. There haven't been any "big bangs" necessarily, but it's really not hard to give the same task to 5.3 and 6 and see which one has better output.