Live data from Hacker News

I gave the AI arms and legs then it rejected me

grell.dev

391–400 of 419 posts

Re: I gave the AI arms and legs then it rejected me

#391

Hey, I'm the author of the blog post. Thank you for submitting this. If you have any questions feel free to ask and please let me know how the writing was. It's one of my first posts so I'd like to improve

You should change the license to AGPL and 'custom, contact for payment details', and provide a link to this as why you did so. Simply put, anything not a viral license like GPL allows parasitization by companies effectively living off FLOSS devs, with absolutely nothing to gain. Human rights under GPL were meant to apply to humans, not '3 lawyers in a trench coat' (corporations). They can make their decisions (snubbi…

I know a bunch of people have tried to argue the toss on this one with you but I'd just like to put it out there that I can't agree strongly enough! Anyone watching can see these big companies are happy to toss developers to the side and develop social harms for profit.

All of this is built on exploiting the open source movement. Delineating between closed source ventures and Free community efforts is just good sense at this point. If they're going to take they must give back.

Re: I gave the AI arms and legs then it rejected me

#392

Earlier quoted context omitted.

>The only exception has been JaneStreet Huh. I guess if you decide to make OCaml your company's primary programming language, you have to take what you can when it comes to devs.

They put an absurd about of effort into their recruiting, so I'm not really surprised. Pretty much every math-related content creator I listen to has advertised for them. I doubt anyone who works there is "take what we can get" calibre. They want to attract people who casually solve college-level math puzzles for funsies. So I imagine it's the opposite and if you get hired there, you're surrounded by people who are e…

I wasn't talking about the likelihood to get accepted or what makes a successful candidate, but the statement that they responded fast to cold applications.

Re: I gave the AI arms and legs then it rejected me

#393

Earlier quoted context omitted.

>The only exception has been JaneStreet Huh. I guess if you decide to make OCaml your company's primary programming language, you have to take what you can when it comes to devs.

Not accurate. They pay very well, and their interview process is supposedly quite challenging. A thing to consider, though, is ethical: they seem to have been involved in market manipulation. [0] [0] https://www.bbc.com/news/articles/c5y0zgrevl1o

Wasn't referring to the level of challenge but to the fact that they responded fast to a cold application. You don't have to do that unless you have a very hard time finding qualified candidates. And they certainly made it hard for themselves by going all in on OCaml.

Re: I gave the AI arms and legs then it rejected me

#394

Earlier quoted context omitted.

You should change the license to AGPL and 'custom, contact for payment details', and provide a link to this as why you did so. Simply put, anything not a viral license like GPL allows parasitization by companies effectively living off FLOSS devs, with absolutely nothing to gain. Human rights under GPL were meant to apply to humans, not '3 lawyers in a trench coat' (corporations). They can make their decisions (snubbi…

Some people want others to freely use their software and choose MIT precisely because it's more free than GPL. There's nothing wrong with just making something for free and giving it away if that's what you want. Not everybody has to be chasing money in all their activities. The author said he was proud of this outcome and nervous at how widely his hobby project will be deployed. That sounds like the ambition of many…

> choose MIT precisely because it's more free than GPL.

This has been debated a lot already. It depends whose freedom we are talking about. It is overly simplistic to define MIT as more free.

Re: I gave the AI arms and legs then it rejected me

#395

Unfortunately, the choice of license likely won't matter in the nearest future (if not already so). If a tech giant wants you open-source library, they will just point their agent to it and ask "to rewrite in the style of War and Peace". And more unscrupulous players won't even bother with a rewrite, as we've seen recently in the case of Cheatingdaddy/Pickle.

what has been your experience with asking AI to create a complex project from beginning to end?

Exactly zero, if answering your question as stated. But that is not the point. LLMs struggle with things that require tacit knowledge or are not found in the training set (i.e. haven't been done before). They excel at tasks that are described in the most verbose and explicit way, to the point that it looks humiliating from human POV. And what is more explicit than an existing codebase, especially with tests, comments and documentation lovingly put there by the proud author (remember, those github stars).

Re: I gave the AI arms and legs then it rejected me

#396
While I do think it's easy (and justified) to be a little full of righteous indignation over this, I do think in general people (myself included, but not necessarily this author) tend to overvalue their code. It's an expression of our imagination and skills, and is our little baby, so it makes a lot of sense that we would ascribe higher value to it than someone else might, but I think the occasional re-grounding is healthy. Code in and of itself isn't usually very valuable, in fact it is often worthless. (Note I'm not saying Enigo is worthless, in fact it's clearly far from it, but the vast, vast majority of open source code does tend to be relatively worthless. Do a github search for various things if you don't believe me). Remembering this (at least for me) reduces the sting a little bit.

Now that said, I think there's an opportunity here for you. It's rare, but I've known people who spun open source like this into a revenue source, not directly from the company but from other ways. My advice (if you have the time/motivation/willingness): Embrace Anthropic's use here, even maybe reach out and help accomodate them for features! Offer to help get the Linux build working (you would be a personal hero of mine for that!). They may even hire you once they see that you're genuinely good and helpful. I have seen many times in the past (including personal experience) trying to hire an open source dev, and we discovered that they weren't actually very nice people, and in about half the cases they were outright arrogant, possibly even with a superiority complex. The beautiful thing about code is that it doesn't care about personalities, but rather "does it work." In a workplace though, personalities are (often) even more important than code. We almost hired one of the most brilliant open source guys I've ever seen - his code was like a work of art, and his domain knowledge was unparalleled. But he was an asshole to the HR person, was openly critical of our interview process (during the interview!), and had virtually no tact when talking to what would be his fellow coworkers. It's ok (and sometimes very good) to correct an interviewer if they ask or say something technically incorrect, but the way you do that matters. Don't immediately retort with stuff like "that's completely and utterly wrong" or "that's a really stupid way to solve that." Nobody wants to work with people like that. Of course I'm not suggesting that OP is that way, but rather trying to share some personal experience as to why a company may not be in a rush to hire an open source contributor. To be clear, that doesn't justify what they've done, but I do think remembering that Anthropic isn't just a giant monolith, but is rather made up of other humans that are just as human as we all are can be very helpful at understanding why things sometimes go the way they do.

On the license question, I'm a huge proponent of the GPL/AGPL (I would always suggest licensing end-user software this way), but I think for libraries like this MIT is usually the best option. Had you licensed it AGPL it may never have gotten traction in the first place, and you'd be worse off than you are now. It's possible it could have led to licensing, but having seen the inner-workings of big tech companies, I highly doubt it.

Apologies for the length of this. It struck a chord with me because I've had libraries used in a similar way (though at a much smaller scale) and had to do a lot of soul searching on it.

Regardless, awesome work! At a minimum you now have an awesome story to tell and you're part of a small club of people that have been successful enough to get hit :-D

Re: I gave the AI arms and legs then it rejected me

#397

I wonder if it was geolocation? Anthropic is based in SF, the author seems to be based in Munich, and maybe they're not open to hiring people who aren't based in the US right now? Given the state of US visas right now, this wouldn't shock me.

My company, which is significantly smaller, hires people in multiple countries across the world. You don't need an office to hire (I am sure there so exist countries where you do, but I expect they are the minority).

Having worked in such companies, switching to that mode requires very different processes.

Re: I gave the AI arms and legs then it rejected me

#399

Earlier quoted context omitted.

If the license were copyleft forking would not be a solution.

Fails the desert island test then. But maybe that's fine. The world is different from 1980's. Different license approaches are needed and Richard Stallman is not God.

It's crazy. I used think his ideas were completely unrealistic and him being a general loon. He has certainly had the last laugh (or first tear, really). I still think he's definitely not correct on all fronts, but I pay more attention to his opinions on tech.

Re: I gave the AI arms and legs then it rejected me

#400

Earlier quoted context omitted.

what has been your experience with asking AI to create a complex project from beginning to end?

Exactly zero, if answering your question as stated. But that is not the point. LLMs struggle with things that require tacit knowledge or are not found in the training set (i.e. haven't been done before). They excel at tasks that are described in the most verbose and explicit way, to the point that it looks humiliating from human POV. And what is more explicit than an existing codebase, especially with tests, comments…

Thanks, that does answer my question.
Post reply on HN