Just to add to this. Open source for money has been a dead end for a long time, except for the (increasingly rare) situations where people accidentally convert their open source _contributions_ into employment (I accidentally did this back in 2015). Open source for recognition/reputation makes a bit more sense, but it is also becoming increasingly rare. LLMs are super-charging the extinction, but this was also observable in 2021, when I wrote this: https://news.ycombinator.com/item?id=29714929 .
Even before LLMs, I have seen people (shamelessly) re-implement code from open source project A into open source project B, without attribution (IIRC, a GPL C++ project [no hate, I use C++ too these days] basically copied the very distinctive AVL Tree implementation of a CDDL C project -- this is a licensing violation _and_ plagiarism, and it effectively writes the C project out of history. When asked about this, various colleagues[1], just shrugged their shoulders, and went on about their lives.). LLMs now make this behavior undetectable _and_ scalable.
If we want strong copyright protections for open source, we may need to start writing _literate_ programs (i.e. the Knuthian paradigm, which I am quite fond of). But that probably will not happen, because most programmers are bad at writing (because they hate it, and would rather outsource it to an LLM). The more likely alternative, is that people will just stop writing open source code (I basically stopped publishing my repos when the phrase "Big Tech" became common in 2018; Amazon in particular would create hosted versions of projects without contributing anything back -- if the authors were lucky they would be given the magnanimous opportunity to labor at Amazon, which is like inventing dynamite and being granted the privilege of laboring in the mines).
The fact is, if we want recognition, we need to sing each others' praises, instead hoping that someone will look at a version control history. We need to be story-tellers, historians, and archivists. Where is my generation's Jargon File?
[1]: Not co-worker, which is someone who shares an employer, but colleague, which is someone who shares a profession.
That's a big reason why FOSS is going to crumble. If AI succeeds and decimates the tech labor industry, people won't have the luxury to "code for fun". Life isn't a bunch of comfy programmers working on stuff in their spare time anymore.
We already see a component of this with art, but art actually needs to be displayed unlike code to show its vslue. So they adapt. Tools to keep the machine from training on their work, or more movements into work that is much harder to train on (a 2d image of a 3d model does the job and the model can be shared off the internet). Programming will follow a similar course; the remaining few become mercenaries and need to protect their IP themselves.
A little hungover, but some foggy thoughts: If a company can differentiate itself by loving its customers, then that is an indictment of the rest of the entire industry. Loving one's customers should be the standard. But alas, the people with their hands on the money-faucet tend reward profitable (zero-sum) behavior over virtuous (positive-sum) behavior. One can only hope that after they alienate enough of the population, they will run into problems that they cannot buy their way out of (so that they can learn what it feels like when those-who-cannot are at the mercy of those-who-can).
By way of analogy, the result of the theorem prover is usually actionable (i.e. we can replace one kind of expression with its proven equivalent for some end like optimizing code-size or code-run-time), but mathematicians _still_ endeavor to translate the unwieldy and verbose machine-generated proofs into concise human-readable proofs, because those readable proofs are useful to our understanding of mathematics even long after the "productive action" has been taken.
In a way, this collaboration between the machine and the human is better than what came before, because now productive actions can be taken sooner, and mathematicians do not have to doubt whether they are searching for a proof that exists.
Entire companies have been built around synchronizing the WAL with ZFS actions like snapshot and clone (i.e. Delphix and probably others). Would be cool to have `zpgdump` (single-purpose, ZFS aware equivalent).
I could believe that 90% of code will be generated by LLMs, because it takes almost no effort to generate some common boilerplate using an LLM. It's kind of like saying that 90% of code is BSD/MIT/GPL licensed because many people compulsively fork repos on github.
This just means that a smaller percentage of code will be useful/must-have/must-read code. Kind of like the market for films and books and games -- many more get made every year, which means that many more will get ignored every year.
> Sadly this won't be the wake up call that we need to bring back some sort of meritocracy to American society.
Genuine question: are we 100% sure that this society was ever meritocratic? How would we measure `meritocracy` so that we can compare year-to-year and decade-to-decade?
Of course not. It’s impossible. Take the presidency. Let’s say a very meritocratic person gets elected. There would be a hundred equally meritorious people who do not hold that office. Same for many things.
Meritocracy is a measure and some sort of idealized concept used for comparisons.
that's a utopian vision. we don't need a perfect meritocracy. majority of parents would do anything to put their kids ahead of others. i would help people I know and have relationships with first. it doesn't work out all the time.
reply