I think the saddest part of the LLM infection of open-source software development is the loss of trust in each other.
-
I think the saddest part of the LLM infection of open-source software development is the loss of trust in each other. Beforehand, if somebody submitted some bad code, it could be assumed it was either a mistake or they weren't well-versed enough with the codebase. Nowadays, any code submitted by an unknown party brings immediate scrutiny and a "guilty until proven innocent" mentality. Its a valid defense mechanism, but I will never forgive the corpos for this.
@justsoup I suspect it's their way of fucking up open source software. I mean, free software? How does that create billionaires?
In all seriousness, though, how can you trust software to be clean of slop? I'm extremely wary of open source as a result. It's like unknown people are pissing in the only fresh water well around.
-
@justsoup I suspect it's their way of fucking up open source software. I mean, free software? How does that create billionaires?
In all seriousness, though, how can you trust software to be clean of slop? I'm extremely wary of open source as a result. It's like unknown people are pissing in the only fresh water well around.
@Oyu_Fka @justsoup It goes much further than that. Open source is at least inspectable, and most have a minimum of awareness that they're at least disclosing LLM usage.
Now, even lead maintainers occasionally fall for the LLM-slop-gambling machine.
But I would trust proprietary software even then way less, because checking that for slop is close to impossible, except by guessing upon encountered out-of-the-usual regressions.
-
@Oyu_Fka @justsoup It goes much further than that. Open source is at least inspectable, and most have a minimum of awareness that they're at least disclosing LLM usage.
Now, even lead maintainers occasionally fall for the LLM-slop-gambling machine.
But I would trust proprietary software even then way less, because checking that for slop is close to impossible, except by guessing upon encountered out-of-the-usual regressions.
@Oyu_Fka @justsoup In particular, in the context of Open Source, individual people have a lot of trust to lose by trying to hide LLM usage. Additionally, the fear of copyright/license issues keeps many in check.
For proprietary software, the fact that copyright might be unenforcable is just not a concern unless/until the code gets leaked.
-
I think the saddest part of the LLM infection of open-source software development is the loss of trust in each other. Beforehand, if somebody submitted some bad code, it could be assumed it was either a mistake or they weren't well-versed enough with the codebase. Nowadays, any code submitted by an unknown party brings immediate scrutiny and a "guilty until proven innocent" mentality. Its a valid defense mechanism, but I will never forgive the corpos for this.
-
@Oyu_Fka @justsoup It goes much further than that. Open source is at least inspectable, and most have a minimum of awareness that they're at least disclosing LLM usage.
Now, even lead maintainers occasionally fall for the LLM-slop-gambling machine.
But I would trust proprietary software even then way less, because checking that for slop is close to impossible, except by guessing upon encountered out-of-the-usual regressions.
@fogti @justsoup Don't get me wrong, I'm not suggesting in any way that proprietary software is the way to go.
I appreciate Open Source is inspectable - my point is about the innocent user...the person who doesn't know how or where to inspect the software - me to a great extent, lol!
It's just worrying what slop may be running within your machine.
-
I think the saddest part of the LLM infection of open-source software development is the loss of trust in each other. Beforehand, if somebody submitted some bad code, it could be assumed it was either a mistake or they weren't well-versed enough with the codebase. Nowadays, any code submitted by an unknown party brings immediate scrutiny and a "guilty until proven innocent" mentality. Its a valid defense mechanism, but I will never forgive the corpos for this.
@justsoup i'm seriously starting to question the notion that you should update software as soon as a new version is available.
-
@justsoup i'm seriously starting to question the notion that you should update software as soon as a new version is available.
@fishidwardrobe @justsoup Some of us never accepted that notion to begin with. I've considered it malicious for 20+ years. I patch or otherwise mitigate something when it has a vulnerability that actually matters, and otherwise, only update something when it has serious new functionality I want. Otherwise it's whatever version it was when I found it.
-
I think the saddest part of the LLM infection of open-source software development is the loss of trust in each other. Beforehand, if somebody submitted some bad code, it could be assumed it was either a mistake or they weren't well-versed enough with the codebase. Nowadays, any code submitted by an unknown party brings immediate scrutiny and a "guilty until proven innocent" mentality. Its a valid defense mechanism, but I will never forgive the corpos for this.
@justsoup Yes...
-
@justsoup i'm seriously starting to question the notion that you should update software as soon as a new version is available.
@fishidwardrobe @justsoup software updates eventually kill your portable devices. Each New version works more slowly
-
I think the saddest part of the LLM infection of open-source software development is the loss of trust in each other. Beforehand, if somebody submitted some bad code, it could be assumed it was either a mistake or they weren't well-versed enough with the codebase. Nowadays, any code submitted by an unknown party brings immediate scrutiny and a "guilty until proven innocent" mentality. Its a valid defense mechanism, but I will never forgive the corpos for this.
@justsoup A side-effect (or intentional outcome) of LLMs that is often ignored regardless of if you use it or not is the erosion and breaking of basic societal expectations of trust and veracity. Bring those into question and you're undermining what allows society to function.
-
@justsoup i'm seriously starting to question the notion that you should update software as soon as a new version is available.
@fishidwardrobe updating quickly is only important in security contexts - so, world-exposed server stuff, and probably your kernel and your web browser. Maybe a few other things. Everything else? Yeah, don’t update if you don’t have to.
-
J jwcph@helvede.net shared this topic
