Hacker Newsnew | past | comments | ask | show | jobs | submit | gregoriol's commentslogin

How is it really related to AI? there have been issues with open-source and maintainers for a long long time

In the post, the ladybird maintainers say that they trust pull requests less than they used to, because many pull requests are authored by AI now. A big pull request no longer signals that the submitter put in a lot of work into it and it's committed to developing and maintaining quality code.

Not sure if this happened to ladybird, but the amount of junk vibecoded AI-slop pull requests has been putting an immense amount of strain on many open-source maintainers. Reviewing stuff like that is intensely energy draining an most of the time your comments will just be copy-pasted into claude code and the "contributor" will put in 0 effort themselves to try to make the code readable or maintainable.

Before AI, being open source and having to manage issues and PR's was already a huge task, burning out maintainers left and right. Now with AI, anyone with a terminal and a claude code subscription can open PR's...


Maintainers can also use the exact same tools to help review and validate PRs

So the solution is for FOSS developers to pay large sums of money to AI companies, to solve the problem that the AI companies created, for profit? ... and you typed this out as though extorting charities were a solution, instead of a grotesquely immoral and unethical systemic failure?

I'm on your side, but don't you think 'grotesquely immoral and unethical' is a bit of an exhaggeration? It's just software.

No, these things don’t exist in a vacuum. Most LLMs were trained on huge amounts of copyrighted works that they then regurgitate portions of without regard to license or copyright holder, and without attribution.

Why do you believe morals and ethics do not apply to software? Would you say nuclear weapons are "just physics"? What about the software involved? In that context, software is a tool that can make or break civilizations.

Honestly, this kind of disregard and reductive reasoning comes across as corrupt and sociopathic. The sentiment is a recurring theme I see on HN more than any other forum, and representative of the moral and ethical systemic failure permeating modern business and governance; how most politicians, investors, and leaders treat their users/customers/employees/constituents adversarially, as though they are marks to be fooled, manipulated, and exploited without conscience.

We should always hold each other accountable and ensure our beliefs and actions are conducive to improving everyones quality of life and standard of living. Software is no exception. These should not just be virtues to be signaled through marketing, or while standing on a pulpit, or being recorded. They are how we should strive to live our lives in private, even when nobody is watching.


Software has an ethical dimension, and it's ignorant or disingenuous to dismiss that.

So 50 years ago people who created calculators were at fault for people not calculating from memory? or 100 years ago people who created engines were at fault for people not doing manual labor anymore?

Interesting. I don't remember calculators or engines being required to solve the problems they created... nor monthly subscription fees to maintain protection from those problems... nor the companies responsible profiting off the amplification of labor, cost, time, and effort inflicted on non-profits; nothing remotely similar to the problems AI companies have forced on FOSS.

Perhaps you could provide some examples?


AI reviewing AI. Or in other words, having non-deterministic systems review the creation of non-deterministic systems, hoping for a deterministic result. Good luck with that.

My friend, the very article we are talking about this mentions this directly

This is direct result of AI as you can see in many other public repos.

before AI like 1 in 1000 would spend their time fixing something they had no idea about and even then considering how much time you spent and how few of those happened it made sense to review/talk about it.

now every "dev" with claude submits prs having absolutely no idea what they are even doing. most of them would not even be able to create PR without AI in the first place.

and on top of that add slop bots that "fix" issues in the loop and create hundreds of PRs daily


Read the post

This is only because of AI. In the past the barrier for contribution was high because you had to know what you are doing and put effort.

Nowadays any AI lunatic with a couple of tokens to spare can spare no effort, have no understanding and still flood you with a wall of code that on first look might even look okay (spoiler: it's actually trash). That is tiring for maintainers.

This is all about AI


You do understand that cars are designed to be in situations where they don't have network access for some time? parking underground and stuff?

Well my thinking is maybe the telemetry module (which has no power) might be doing the logging itself.

Which model is it? There has been no zero connectivity cars produced in the last 10 years or so, except maybe some niche manufacturer?

I've got a couple older cars (2010, 2003), but the only new one I'm excited about - the only actually new car I've ever considered buying - is the Slate truck:

https://www.youtube.com/watch?v=5_G4OfXTlvs

It has an LCD for the gauges and backup camera, but no modem and no surveillance tech. Bring your own phone/tablet if you want navigation or audio.


claude package has ten new versions published per week, and one new model every few months, one should definitely not rely on some undocumented tricks around it: it'll change, it'll break deep ultra-specific configurations

in my experience, "undocumented tricks" break as often as documented features

like when they removed "clear context and execute plan" option after releasing 1M opus because "context window is not a problem anymore"


I so miss that clear context and execute plan mode! Now i have to keep clearing it manually again.

fyi you can re-enable it with `{ "showClearContextOnPlanAccept": true }` in ${CLAUDE_CONFIG_DIR:-$HOME/.claude}/settings.json

also find `"disableAutoMode": "disable"` useful, since I'm typically switching between yolo and plan


Thanks!!

> like when they removed "clear context and execute plan" option after releasing 1M opus because "context window is not a problem anymore"

I then show me banner that I can clean up "xxx.x k" of context by typing clear right after it started on the plan.


This is true, but also "temporal hacks" can make or break "cutting edge" workflows. I don't re-architect my claude instructions every release. But some releases justify examining your existing instructions and making sure they still fit the current model. And it has made a noticeable difference.

It is possible to build automation that efficiently handles low level customization of new versions as they appear.

Katy Perry was not lost, the world can go on

I think a company has to be able to change its commitment, but should not screw users at the same time. For example, if they want to remove the free plan, why not, strategy can change with context, the world is moving around the company, so then remove it for new users not existing ones and it's all good.


Maybe that's fair but that's not what's happening here.


Maybe they should ban metal grinders as they allow people to steal things?


AWDL is such an amazing technology, it's understandable that Apple wants to keep it only for their devices as it gives them a noticeable advantage for quick stuff sharing.


They didn't. Apple contributed the core logic to the Wi-Fi Alliance to build Wi-Fi Aware, which they now also support.


Interestingly, it still took the EU to force them to actually adopt it (and open it up for apps to use) in iOS 26.



Apple docs say iOS 26/macOS 26, that's so brand new that no apps are using it right now, will have to check that again in a few years.


Kind of. When I looked, they added the api for devs to use on iOS, but it isn’t on macOS yet, and nothing uses it as far as I could see.

It’s a future promising tech though. A much better version of Wi-Fi Direct.


So, should there be apps that use it to transfer files between iOS, Android, Windows, and Mac without requiring them to be on the same network?


Google's QuickShare contains a reverse-engineered AWDL implementation that works on Pixel and a few other phones.

As for WiFi NAN: support for it seems rather limited outside of iOS and Android. From what I can tell the feature is barely tested on Linux and I can't find any generic Windows APIs for it either.


No WiFi cards for pcs support it.


That's not quite accurate, I don't think.

I've definitely used STA and AP modes concurrently on my Windows laptop with the operating system's built-in internet connection sharing function to help troubleshoot a problem in the field.

That was around a decade ago. It didn't take any extra effort on my part; I just told it to do the thing, and then it did that thing.


Researched this for a bit: there is some hardware for PCI supporting it, but Windows 10/11 not, and Linux is still work in progress, so no real support on OS level, only for some iOS/Android devices.


Afaik the hardware supports it for a while now, but there is no firmware to expose the functionality.


it might be interesting to use unused or extra wifi cards to support this. My pc motherboard has both wifi and ethernet and I only use the ethernet. That card does absolutely nothing at all.


The EU required they use an open standard https://arstechnica.com/gadgets/2025/11/the-eu-made-apple-ad...


AWDL really isn't that novel, neither as an idea nor implementation. What Apple did nail is the user experience on top of it.


Except 20% of the time it just doesn’t work. Hardly an advantage if most people default to texting because of airdrop’s failure rate


We have been using AWDL intensely (not via AirDrop but Network.framework) for 6+ years and it fails less than 5% of the time. It's pretty impressive for a non-connected link between devices. The most common problems we face are very high device density places (100+ device in 30sqm space) and device wandering out of reach quickly (sometime as low as 5m).


Agreed. It's been like that for years.


The law is not about you, but about everyone: 1) Apple doesn't have service centers everywhere: some countries/cities/small towns don't have them 2) Apple doesn't provide service for older devices 3) making it easier doesn't mean you'll be able to swap them live as we did in the 90s, but it means you could do it at home with a reasonable set of tools instead of sending the device to some shop that would need to unglue, unsolder, ...


Out of the 27 EU countries, there are only apple stores in 8 of them.


So you mean you are keeping full card numbers somewhere in your logs to... fix some potential security issue...?


>Hey mr processor, the cards for transaction numbers x...y are stolen.


I'm assuming there were transaction IDs provided that can be given to the processor. If they can't do anything with the IDs, then that's a pretty broken system.


If you are the processor, yes, I guess. If you aren't, then you can provide the transaction ID to the processor and let it handle that part.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: