354 points by pandemicsoul 3 days ago | 260 comments | View on ycombinator
joshdata 3 days ago |
timbert 3 days ago |
Here's my most-recent real-world problem: I wanted to plug in my midi controller keyboard and just play a piano, so I can practice my choral music. Very basic.
This turned into many hours over several days trying multiple DAWs, trying to get tehm to install, get them to run at all, get them to recognize my controller, get them to play a VST plugin...changing system settings, completely jacking up my sound, restoring it all again.
It was a nightmare.
So instead I fired up Claude, worked with it for about 3 hours, and now I have an *8-channel* VST player that supports all my hardware, and includes a snappy UI, with outstanding performance and sound. It's amazing!
I literally spent less time creating a custom solution than I did trying to get the off-the-shelf stuff to work. I could make my player OSS too, but the effort to do that is huge, for all the reasons this forum would understand.
Bespoke software avoids all of that and allows me to do what I want, with the environment I have. I don't have all the features of an FL Studio, but I have what I need, and that works for me, beautifully, at low cost and effort. What's not to like?
So maybe we should share OSS "templates" instead of full products, projects that are the barebones of a solution that you can clone, and have your LLM modify to suit your needs, on your platform of choice.
For example, perhaps my barebones VST player is mostly good enough for you, but you also want to play sound fonts? Cool, clone my project, instruct your LLM to add sound font support, and add support for your controller or your output device. Or convert it from Gnome to KDE.
The time and cost of doing this is so little now, why not? I'm getting more out of my computer than ever, and I'm happy about that :)
linsomniac 3 days ago |
Beware: Your best path forward there may be, contrary to part of the thesis in this post, be to vibe code a replacement. There are a couple reasons why:
Expect to encounter significant push-back due to AI contribution. For example, I've had bug fixes to significant problems be rejected because "we are in the future going to be adopting a no-AI policy". Annoying because I spent my time working on it because they did not have a no-AI policy. Amusing to have another contributor contribute the exact same fix, like character for character, under a new PR (referencing my issue, so not done in isolation) and get it accepted.
There may also need to be sweeping changes to the project you are fixing -- which are going to be a burden to the maintainer and likely wouldn't be accepted because of that. A fork may be the way to go here, but also "vibe coding" a replacement might be the best option.
Earlier this year I build a replacement apt cache called apt-cacher-ultra. Part of it was because the existing apt-cachers had reliability issues, but a bigger part was I wanted something that would survive ubuntu.com being under a DDoS for a week or two. apt-cacher-ng was totally useless at that. The path I wanted to try was to have verified known good snapshots of the upstream, background refreshes, pre-fetching of known "hot" packages when the upstream refreshed. HUGE changes.
I've put 100 hours or more into this "vibe coded" alternative (mostly enhancements and fixes for other users use cases). I'm pretty confident that this wouldn't have been something the apt-cacher-ng or apt-cacher maintainers would have been interested in accepting.
delichon 3 days ago |
This was just a sci-fi dream in 2022. Now I wonder whether apps are collapsing into an inference wrapped singularity.
ValdikSS 3 days ago |
https://dri.es/license-only-versus-stewarded-open-source
The truth is: open-source doesn't come with any support, maintenance or feature request fulfilment by default.
By definition, that's just a code with no strings attached.
Do not expect maintenance or bug/security fixes by default from any project, unless stated otherwise, or there's at least some metrics which hint of maintenance.
We need SOCIAL.md to describe expectations from this project.
If that mindset is new to you, read this: https://raccoon.onyxbits.de/blog/bugreport-free-support/
shmolyneaux 3 days ago |
This is the primary perspective behind this post, and that's going to look much much different than the devops roles a lot of us have.
> Too many apps need a database, a cache, a background worker, a reverse proxy and a long config file before you see a login screen
These apps need a database/cache/proxy/config because creating a fault-tolerant, performant, flexible system needs these things. The author wants everything to be a single binary running on a $5-a-month server. A lot of these open source tools could be updated to run in this mode, but that's not the configuration that's sustainable when there's a disaster.
You need to look at why these nonprofits are using open source projects in the first place. They probably want something free, and they probably want something with clear documentation. Maybe it's something that they're already used to. I would argue that AI-assisted (or vibe-coded) apps are cheap enough to fill this niche. Documentation is less important if the features of the app are much more narrowly-scoped. If they're already used to another app, then the vibe-coded one could be designed to provide a similar experience.
I've worked with non-profits that have struggled under the weight of the software they're using. Most often the problem is the software being too general in a way that doesn't model the domain well. Cheap custom software would help a great deal.
gwt4life 3 days ago |
poisonborz 3 days ago |
People post a bunch of throwaway microprojects. So what.
There are wide fields, industries, branches of society, communities, groups of friends left without proper apps to do "that thing". Not to do it better, but to do it at all. It is incredible that a much wider amount of people can engage in creating those apps now. That should be encouraged, and not gatekeeped into existing projects that are already drowning in PRs.
cheney_2004 3 days ago |
listenfaster 3 days ago |
blamestross 3 days ago |
dataminded 3 days ago |
My apps are native and fast. They run on all my devices, connect to each other through APIs and MCPs, and look the way I want them to.
kaishin 3 days ago |
Ericson2314 3 days ago |
maxdo 3 days ago |
Future Software Factories will have extra large static harness, that will be have a very few dependencies, e.g, reactjs, bundled into the package. Even nodejs probably will be more of a runtime feature, patched by the cloud.
open source was mostly a feature of not enough software eng resources. Now , since this is in abundance, instead of a shortcut it becomes a risk.
cush 3 days ago |
jph 3 days ago |
1. Friendly editor: Vix @ https://github.com/vixide
2. Broader web UI: Lily @ http://github.com/lilydesignsystem
3. Smarter web service helpers: Axa @ https://github.com/publicprojects-io/axa
I tried many open source software projects before starting these, and I'm a longtime open source developer so I highly value all the work from people working on vi and emacs, Daisy UI and shadcn, Rocket and Loco, and more. These projects are all aiming for public nonprofit social good.
_nobody 3 days ago |
Kuyawa 3 days ago |
You may not need it, may not like the colors, the UI, the depth, features, etc. That's the beauty of AI it can build whatever you want however you want it, all for less than a dollar in less than an hour
Yes, we need more apps, tailored to every user. From now on I refuse to install any app not customized to my needs
And yes, we should share more, so AI learns from all of us and builds stuff for us with all the shared knowledge of all
Ersho 3 days ago |
Also, its hard for me to trust their AIs and feed again and again my information to many different apps and their AIs.
If you think about it, you need one claude or codex subscription and you can generate your own apps with your own AI models that I trust more than ones used in the apps.
Recently I started working on open source local app playstore. Would appreciate your feedback and any input - (https://github.com/Ersho/sayn)
jstanley 3 days ago |
Yes fix bugs and add features etc. for your own personal use, but don't get your hopes up about contributing back, it's usually a waste of time.
mschuster91 3 days ago |
AI-assisted means that someone who can actually write code does the piloting of the AI.
Vibecoding however, that's usually people with sometimes zero understanding of tech, often enough not even capable of executing basic CLI commands, "creating" projects, without having anywhere near the skill required to see if they're getting shafted by their AI tool of choice.
rishankg 3 days ago |
coretx 3 days ago |
FinnLobsien 3 days ago |
But I don't think there's a big overlap between the people who could even name one open-source library and the people flooding the market with vibe-coded to-do list apps.
apitman 3 days ago |
gortok 3 days ago |
The problem with Open source presently is that it provides no way to sustain the energy used to support it. I have a limited amount of time in my life; I currently spend roughly 1/3rd of it supporting my family. If I use another chunk of time to ‘support’ open source, I’m looking at my work being taken and used for the profit of others without being able to profit from it myself. If I can’t profit from it, I can’t use it as leverage to make it so I don’t need to spend 1/3rd+ of my life working to support my family.
I am currently writing software with the expectation to sell it. I may even give away copies of it, I haven’t decided yet, but the source code won’t be online, and the support documentation will be behind a registration wall, as will the API. Nothing will be visible on the public internet, not for any other reason than AI scrapers and the large companies that use them.
There may be other ways for small software businesses to be sustainable, but at this moment it’s hard to see them.
mmorrell0 3 days ago |
florieger 3 days ago |
komali2 3 days ago |
mipo256 3 days ago |
x3haloed 3 days ago |
The fact that free software is benefitting at all from more coding power is a win. We have bottlenecks to solve though.
Ampersander 3 days ago |
Xirdus 3 days ago |
Because this is the joke. AI agents didn't change who "gets" to write software at all, it's still the same middle-level and senior devs it's always been. The term is intentionally derogatory towards the people who think otherwise.
hypfer 3 days ago |
This reads like Hacktoberfest all over again.
What one person might find "annoying", another maintainer might've purposefully put there. Not all friction is bad, and telling people to go out, remove all friction and then report back to receive a pat on the head is probably going to end poorly.
This reads more like another attack on the commons than someone genuinely invested in FOSS. Which.. uh.. checks out when you check out the author.
__
Please do not take advice on how to interact with FOSS from anyone but the actual people building it.
If you fundamentally disagree, that is fine. You can become the person that builds their own thing and makes their own rules within that space. But only there.
tclancy 3 days ago |
Because I am a masochist I also decided to make this a way to revive the 2 (!) Stadia controller I have. The initial results were less than promising in the same way hitting your head on bricks is. But with the wireless debugging port wide open on my tv now, I have a small app I am cleaning up that could help you work through the absolute vagaries of connecting Bluetooth game controllers to Google tv (once you go into promiscuous mode, something like three separate services on the tv start fighting for you, last man wins and usually the last man is a service that doesn’t even consider game controllers as a case). I’ve no idea how well it would work or where even to offer it, but in spite of me building new stuff, I mainly do it along these lines too. See also, lintarr a tool I am polishing up which tries to do something like formal verification for all the rules you can set up in different services across the arr stack so you can’t have rules that clash.
micimize 3 days ago |
When someone vibes out a new little todo-like app, they are usually thinking of exactly _one_ user: themselves. They get what they want with very low executive effort. That is what's so cool about it!
Good Open Source contrib takes effort, they don't want randos slopping out PRs. Total apples-to-oranges comparison.
faangguyindia 3 days ago |
A very common project that most guys who find this "vibe coding" do is make a calorie tracker, so they ask Claude to build them a calorie tracker, then they add some data from an Apple Watch and Garmin. Now they believe they've built a "dieting app."
It's just funny how many of them actually take data from Garmin and an Apple Watch and add a bunch of activities to work out a "calorie burn of the day" and how inaccurate it is.
Apple and Garmin are not accurate for this, read this: https://macrocodex.app/knowledge/rethink/how-many-calories-d...
But very few of them actually bother to check whether their "calorie burn estimate" even works to achieve the rate of loss or gain they set.
Not everyone is "doing better with AI", many of them are doing worse, they just don't know!
Woodi 2 days ago |
Nice.
> The curl project ended its bug bounty in January 2026 after a flood of fake, AI-generated security reports
Can be not so nice...
But we already have method of dealing with project being overloaded with work: distros, like in Linux distros, and less much Space|Doom|etc Emacs configs but still.
Real problem in those days is requirement of compiling stuff yourself. Eg.:
configure --prefix=/aux --no-gtk && make && make test
then checking and then making installing is black art :)
O, I forget to paste unpack command there... ;)
meerita 3 days ago |
deepsquirrelnet 3 days ago |
Lots of people make things now and that's cool and fun. But what people want to see is a track record of support. They want to know that the things they come to use and depend on won't exit the scene just as quickly as they came.
This is just like 'killed by Google' on steroids. Yes you can write way more things things than before, but if you can't sign up maintain them, then they aren't providing value.
Focus is becoming the commodity. Doing one thing really well is what becomes hard when you could be doing 1,000 things (poorly). To be successful, you can't just throw shit at the wall and see what sticks anymore, because nothing is sticking that way. The walls are already covered in shit. You have to have good product taste and a willingness to apply yourself to it for a long time.
royraghav28 3 days ago |
It's a cycle which breaks without either component missing.
pjmlp 3 days ago |
Rhetoric question of course, especially in the AI coding age.
jmillikin 3 days ago |
> Pick one open source tool you use. Find the most annoying thing about
> installing or running it. Spend a weekend fixing it, with the people who
> maintain it. Then tell me about it.
Flex: The most annoying thing is that it's written in C (and is therefore a pain to cross-compile), and it has a runtime dependency on GNU M4, which depends on gnulib (the GNU anti-portability library). Flex is currently being rewritten to be more complicated and worse by ESR for reasons beyond my comprehension.I had an LLM write an m4 implementation in Rust based on the POSIX specs, then fuzz-test it against GNU M4 until it was mostly the same (modulo all the regex bugs it found in GNU M4). Then I had an LLM rewrite Flex in Rust, using rust-m4 as an in-process library.
QEMU: It's written in C, has a bunch of huge dependencies, automated test coverage that rounds to zero, their main contribution path is a mailing list that sometimes swallows patches, and the maintainers might ask for a risky patch to be made more risky before they'll accept it. I went through that process once for a few relatively minor fixes and never want to do it again.
I had an LLM write a multi-architecture emulator with a JIT, high test coverage, and a device plugin system that uses WebAssembly.
Busybox: Written in C, huge pain to build. I tried switching to Toybox instead but it's incomplete, the build system involves a pile of shell scripts that require Bash, not any other POSIX-compliant shell (including its own), and the maintainer's response was basically to tell me to fuck off: https://lists.landley.net/pipermail/toybox/2024-October/0095...
I had an LLM write a replacement in Rust that has enough shell builtins to run an autoconf script from within a hermetic build environment, and its only build-time dependency is rustc.
---
You know why LLMs are so great? Because they save me from having to interact with open-source maintainers who intentionally choose to make their software baroque and their contribution process obnoxious.
You want your project to be written only in ANSI C89 and built with Meson and accept patches only via GPG-signed email to a mailing list that silently drops HTML? Fine, have fun, an ASIC somewhere in Iowa is going to spend the next 3 hours letting me forget your project exists.
nadis 3 days ago |
sreekanth850 3 days ago |
ing33k 3 days ago |
edstarch 3 days ago |
deckar01 3 days ago |
barelysapient 3 days ago |
jayd16 3 days ago |
You don't think it was easier to scratch your own itch instead of host OSS before AI? Of course it was.
Gone are the days of the hippie techno revolution, I guess. It really is "fuck community, I got mine" and its sad to see.
apitman 3 days ago |
junixx 3 days ago |
- Old proverb
rvz 3 days ago |
Even worse, most of open source is now majority slop and it is highly likely that they will be abandoned.
mithril_ 3 days ago |
gsoff 3 days ago |
qwoprunner 3 days ago |
OSS devs were sinking a mountain of time into relatively unpaid work, doing unglamorous work. Some of it is in backend systems and console utilities that few people will ever appreciate, but a lot of effort went into user-facing things. And yet, so far I found just one program that replaced one of my mainstays - I went with AI-coded SquirrelDisk to replace FileLight. Everything else is the same, and I just don't understand that. Surely not EVERYBODY rolls their own? Surely some OSS people did embrace AI rather than quitting in disgust?
A small example: tagging music files on Linux is horrible. Kid3 is.. it works, but the UX is almost like a parody. Mp3tag runs on windows and it has a macos version and it's a vast improvement. You can run it through bottles but I refuse to believe everybody does that. Surely people that use Linux listen to music, which leads itself to occasionally needing to tag or rename your audio in batches.
I had some spare time so I rolled my own mp3tag clone that would work on Linux, MacOS and Winblows. I added audio playback to it because opening in external program annoyed me. Then I added some functionality for podcasts to it: being able to search for subtitles, playback speed and such. Then I added the ability to convert between formats, using ffmpeg. Then I added a way to measure loudness, some ReplayGain functionality and the like. Then a way to find duplicates, for making sure those random files you have on that old external HDD can be deleted safely, and a way to search through what you have on all disks, even if they're disconnected, so you don't buy some album you already own like a dumbass.
It took me about three weekends of tinkering with it until it was good enough to switch to it. I keep running into small issues but by and large it's been fairly reliable. It's fairly complex, with a lot of buttons in the view, but given that the 'competition' is mp3tag (which is okay but also ugly) or kid3 (which is both ugly and almost unusable) it's just flat out superior, unless you're happy with mp3tag and don't want all the extra complexity. I don't want to offer unpaid tech support to people on GitHub but I'm considering releasing it and just leaving it there with no further communication and maybe somebody will pick it up. If not, it's still better than alternatives even if there are undoubtedly many, many bugs and the codebase is undoubtedly a tangled mess.
Aside from the ugliness of having almost all functionality without any sub-menus that would give Apple designers heart attacks and perhaps more complexity than it should have (I personalized it for what I wanted, after all), it's genuinely a decent application. It's written in c++, a language I never used. It's fairly performant on my ~15k audio files library. Changing it is as simple as pointing a modern LLM at it and asking in english for some changes. It took me approximately 20 hours to make and the cost was about $10 in subsidized tokens from Claude.
If I, a random guy who doesn't even know c++, can write a usable program in three weekends, then where is all the new vibecoded software? There are so many tools that technically work but the user experience is straight up awful, usually on linux side, or flat out holes that aren't filled, usually on mac side. As best I can tell, very little is being developed. People who have no idea about coding are recompiling console games on pc, in admittedly dubious quality, but people that dedicated years of their lives to OSS... don't?
lifeisstillgood 3 days ago |
Open Source did an end run round that problem by producing
- better quality code, faster, and replacing the arguments over standards with a level playing field marketplace
As we are learning software has eaten everything but everything has not become software literate - there are hospital MRI machines that are only certified for windows XP. Our supermarket fridges think they are cheap IOT devices for teenagers bedrooms.
It’s going to take a decade or more for society to work its way through replacing all its devices (sewages plants are devices too) wit what is a software literate approach - secure, PKI, monitored updated and … open
No fucker will trust a fridge running a blob of binary - a complete SBOM and a repeatable build will be table stakes
And while we could wait for standards what is most likely to happen is each industry wakes up and starts to say “we want a beyond zero based approach, PKI setup this way, the builds are published and run here and publish the firmware”
Walmart might be thinking about this now - but open source will if we are senile be the backbone
And it will finally be funded as the answer is for industry players to fund the fridge-python api project, mostly by hiring the devs. Or the devs go full pococo and get sponsored
dankobgd 3 days ago |
NiceWayToDoIT 3 days ago |
drltaidev 3 days ago |
philipwhiuk 3 days ago |
ReDress 3 days ago |
ovnz 3 days ago |
ex1fm3ta 3 days ago |
dario_amodei 3 days ago |
My open source projects were barely sustainable when the few PRs that came in would fix actual, non-hypothetical, serious logic problems.
Now I've been seeing PRs like that, fixing annoyances, performance edge cases, or hypotheticals, across my open source repositories, and I don't have time for them. I especially don't have time to fix problems that probably nobody will ever experience, and yet they keep coming. I have not organized my life around triaging annoyances (let alone fixing them).
If the trend continues, I'll soon be forced to close repositories to contributions, or just stop maintaining the projects entirely. And I imagine that's the point a lot of maintainers are at.