However, it also highlights a key problem that LLMs don’t solve: while they’re great at generating code, that’s only a small part of real-world software development. Setting up a GitHub account, establishing credibility within a community, and handling PR feedback all require significant effort.
In my view, lowering the barriers to open-source participation could have a bigger impact than these AI models alone. Some software already gathers telemetry and allows sharing bug reports, but why not allow the system to drop down to a debugger in an IDE? And why can’t code be shared as easily as in Google Docs, rather than relying on text-based files and Git?
Even if someone has the skills to fix bugs, the learning curve for compilers, build tools, and Git often dilutes their motivation to contribute anything.
> Buying millions of ebooks online would take a lot of effort
I don't understand.
Facebook and Google spend billions on training LLMs. Buying 1M ebooks at $50 each would only cost $50M.
They also have >100k engineers. If they shard the ebook buying across their workforce, everyone has to buy 10 ebooks, which will be done in 10 minutes.
> ask why so much engineering time and money is being wasted on [Linux]
Most likely the corporations will say "because that is much cheaper than developing our own; 10 of our devs on Linux an 99% of devs from other corporations are much cheaper than 1000 of our devs on our own OS". And the shareholders will likely accept that.
The key thing is that for most corporations that contribute to Linux, Linux is not the product (except Red Hat, SuSe etc). Google, Facebook, etc, just need a good OS to run their billions of servers on.
> so we can fork when their interests stop aligning with ours
You can fork but you likely cannot maintain Linux as-is. Where do the 1M hours/year come from? That's hard to do in free time.
That is also fine from the perspecitve of Free Software. The 4 freedoms do not include "the program must be maintainable with little enough manpower for people to do it in their free time, free of independence on corporate interests":
The GPL is great in that it gives large power to users of the software, no matter if those users are corporate or personal, and even if the makers of the software are mostly corporations.
> You can fork but you likely cannot maintain Linux as-is. Where do the 1M hours/year come from? That's hard to do in free time.
Maybe a question too radical for Hacker News, but why does Linux need 1M hours/year? How much "worse" would Linux be for the end user if this year, that time spent dropped to 100k or even 10k? And which "type" of Linux user was benefiting most from that time spent (person or corporation)? Why is more automatically treated as better than less?
> The GPL is great in that it gives large power to users of the software, no matter if those users are corporate or personal, and even if the makers of the software are mostly corporations.
Indeed, I respect the wisdom and forethought of Stallman greatly in drafting the GPL.
I think it's a reasonable question to ask, and there's surely some churn that we could do without. But don't underestimate the effort that goes into device drivers.
A. How do you contribute to the kernel in a way that only benefits the contributing organization? That's quite literally impossible in this kind of project. Even the more niche stuff like virtualization support is used by homelab enthusiasts. It's also not like Linux has 100,000 APIs for every customer under the sun.
B. Most of the effort goes into hardware enablement; CPUs, GPUs, power management, etc. Without corporate interests, try running Linux 2.2 from 1999 on a modern PC (which came out just before the $1B IBM investment). See how well it works. Fork modern 2025 Linux, and try running it on a computer that comes out in 2028. See if it even boots. If Intel's only done a minor refresh, it might work; if it's something bigger like the split to P cores and E cores, expect a brick. Even if it does boot, don't be surprised if it crashes, acts unstable, has borked performance, broken sleep/wake, broken audio, broken USB, you name it, it's probably broken.
C. A forked Linux would not have the same level of security research behind it. For example, the Linux 4 era was marked by the introduction of fuzzers and the fixing of countless bugs. A C codebase with handwritten Assembly is rather unlikely to ever become bug-free. How well would your forked non-corporate codebase handle Spectre and Meltdown, just an example, with Google's experts contributing the technique for fixing these problems efficiently (Retpoline)?
D. Added to the above, this isn't a hypothetical: The FSF didn't like the practice of proprietary firmware blobs being in the kernel; so they made their own commercial-interest-free version of Linux called Trisquel. It still uses commercially written code if it's open source; so even it can't be called completely free of commercial influence. The kicker: It runs on almost nothing, and people were complaining about how it works on nothing 13 years ago.
TL;DR: A forked Linux, is a broken Linux, that will never run well on newer hardware, and will quickly become insecure.
> Most of the effort goes into hardware enablement;
Which is why Linux is primarily corporate funded. The corporates want to sell their hardware. To sell their hardware they need to get it running on Linux, so they fund that. As hardware support constitutes most of the code contributions, most of the code that goes into Linux was because corporate actors made a decision to pay for it's development - purely in their own interests.
That's nice - this symbiosis between FOSS and corporates works well for both sides. But it's a stretch to say Linux would not exist today without it. The most you can say for certain is Linux would not have it's great hardware support without it. The BSD's don't get anything like the amount of corporate funding Linux does. They are doing fine, and notably work on common modern hardware. So could Linux even without corporate funding.
In particular, most of the interesting stuff that happens in the kernel, the stuff that determines what the kernel will look like in 10 years time, stuff like adopting Rust, is driven by people scratching itches in the FOSS tradition. Not all of it - pKVM is Google initiative. But eBPF was scratching an itch. Jens Axboe developed io_uring probably as a consequence of wanting storage to run faster at Meta - but it was definitely an itch of his. It's nice that Meta to paid him while he developed it, but saying its creation was "driven by corporate interests at Meta" is a bit of a stretch.
I believe most of this is hardware support, followed by refactoring. The core kernel functionality certainly doesn't change to the tune of 1M hours/year.
Could somebody with a good understanding of microchip production economics explain what the price difference would be if all 8-bit chips (such as the Atmega328) were replaced by otherwise equivalent 32-bit or 64-bit chops tomorrow (that is, same production capacity and unit counts sold)?
How much would it cost, in cents, to just have those extra bits?
Supply-chains are more complicated than merely the unit price. Things like existing stock, talent, software, product design, certifications (both chip and products based on chip), and hundreds of other things play just as substantial role.
a cheap 32 bit RISC-V chip like the CH32V003 goes for about $0.10, and the cheapest MCU I've heard of (PMS150C, barely useful for anything) goes for around $0.03
Even without special knowledgeon production economics, I dare say there would be a race to produce replacement compatible 8-bit microcontrollers to get people's stuff working again.
It doesn't have much to do with "China being an actual country".
Germany alone is also an actual country and the trains inside it's borders are super slow.
In China and Japan, many long distance trains get 320 km/h average speed!
The German ICE "machine" could theoretically also do that speed but there are barely any tracks where this is possible, so the average speed is around 3x slower.
In France and Italy it is much better. TGV and Frecciarossa trains usually operate much closer to their specced speeds.
germany is making the big mistake of mixed use tracks. in china high speed tracks are dedicated to high speed trains, and a high speed connection means dedicated high speed tracks for the whole trip. germany is creating a patchwork of high speed routes thinking that this is enough to make high speed trains work.
The slowness regarding rolling out things like ETCS, ATS and standardised European infrastructure as everything to do with Europe not being an actual united political entity. As you rightfully pointed, some EU members rail strategy is very dubious.
Yes, if your delay is caused by DB, and the next train that continues your journey is an hour later, you can take that even with the super saver ticket.
The DB Navigator app can now "even" show that your you are no longer bound to specific trains when that happens, with a little banner in the top.
This is quite common in Blighty; there is no need for another ticket (outside some time restricted fares where the ability to stop is limited). Indeed, it can be nice hopping off at a new city for a mooch around if you have time to do so.
(Only slightly off-topic; We have the phenomenon of 'split tickets', where it can be cheaper to buy two tickets from A --> B --> C rather than one ticket from A --> C even when the train stops at B anyway! You don't even need to get out of the train at B if you don't wish to do so. Our ticketing system is a mess.)
The UK is the odd exception here. But it also doesn't have real high speed trains, except from the one that leaves the country, where hopping off is impossible (unless you want to end up in a dark tunnel under the sea and be arrested afterwards).
Advances are also extremely common in/to/from England and sales are on the up. The government is working hard to replace off peak singles with advances.
Flexible tickets are becoming unaffordable for certain journeys such as those on the ECML with LNER (aka The Government)
https://news.ycombinator.com/item?id=42814509