Diseased Open Source Software Community - it's about ethics in Code of Conducts

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.



Guy wants to contribute to open source, and they tell him this:
1731019387222.png

Which causes him to go bonkers:1731019419307.png

There's also a lot of nazi name calling, ukrainian nationalists, and many interesting personas in the threads (see archives as all comments are deleted now).
 
Slav on slav shit flinging is almost as entertaining as Pajeets melting down. Shame it completely obliterates any discussion other people might've been having.


I checked out the profiles of some of the more unhinged comments on this and what the fuck is this thing (A)?
View attachment 6619146
....a reminder that batshit trannies are not a phenomenon unique to the west.
 
Poor Stallman
Nah, that's one of these things he gets absolutely right. Most of the modern internet isn't worth seeing anyways, and not just because of non-free javascript. It's all just bullshit to trick your limbic system into having FOMO. I found LLMs so effective in replacing the occasional googling I do that practically, besides some very specific things like book pirating or shopping sites, I can go without at this point. I know that's rich coming from book-length post writing guy, but it reallly is just a bad habit at this point. All I do here is the equivalent of smoking a cigarette.
 
Slav on slav shit flinging is almost as entertaining as Pajeets melting down. Shame it completely obliterates any discussion other people might've been having.


I checked out the profiles of some of the more unhinged comments on this and what the fuck is this thing (A)?
View attachment 6619146
He looks like Greta Thunberg with shitty facial hair.
 
There's the Raptor Computing Systems::Talos™ II Secure Workstation, but I have no experience with it.
Literally says POWER9-based, so no, not free.

Yeah, if you want free you have to make sacrifices. You're not getting modern performance with anything truly free (read: on an FPGA). OTOH, unless you're some clown trying to run windows games under Proton, or run LLMs, or engage in bitcoin mining... chances are there's plenty you can do with a less-performant but free design. It's really just a matter of what you value. Trying to have everything though? Not possible at this time.
 
IBMs POWER9 is free in the way that it's blobless all the way down to the microcode, as far as I am aware. I actually did not think about that when I made that post as that certainly is a lot more performant than some low end alternative ISA SoCs. Good point, @UERISIMILITUDO. From a quick google, the cheapest Power9 based platform (quad core) costs arount $3400 for just mainboard and processor. While certainly way too expensive for what it is, it's still affordable for a mere mortal.

This also reminded me of the Libre SoC, which was completele vaporware last time I checked it, although ICs seem to exist now. I can only recommend searching for the guy responsible for it on newsgroups. He is completely insane. Not in the good way.
 
In my ongoing plans to leave the web for good without losing too many of the advantages I've been experimenting with LLMs on rented, distant servers reading websites for me and presenting the looked for information in an emacs buffer.
How you deal with hallucinations? I been using LLMs to summarize hours-long videos and podcasts and I've had quite a few situations where the results had parts that weren't in the original video/podcast and the LLM just straight made it up.
I am pretty sure that hardware subscriptions for end users and more locked-down systems will be coming eventually
We're already there, I just saw a video of some guy "upgrading" the storage on the new mac mini: he took the removable M.2 drive out............and had to remove the NANDs and carefully solder new ones, because the mac won't take any other SSD that isn't that same original unit it was shipped with.
 
How you deal with hallucinations?
Same I deal with random google search results: Scrutiny. You can't fully trust LLMs. I hope you never fully trusted Google's results. It's still such a speedup. The answers with the current crop of models are always relevant and often accurate. You just have to be aware of the limitations.

Knowledge-based questions at this point work out better for me than the algorithm-driven SEO hell that is the modern internet. I also try website summaries, and you are right, LLMs will sometimes make stuff up or confabulate details. In my experience, reducing the information range of the summary you require helps if you see that happen. You can cut things up in pieces and then summarize, and then make a last pass where the model summarizes the summaries. It also sometimes helps to make the model scrutinize it's own reply on a mostly empty context. They sometimes catch their own errors on subsequent passes.

I'm always surprised how people call these AIs "useless". Yes, it's not perfect but you can ask your computer questions in perfect natural language and get relevant and proper replies and can even reason and argue about them, how cool is that?

I don't do any social media like Twitter, Discord, Reddit, their various OSS alternatives and co.. I don't do platforms that are of a slightly more parasocial nature either, like YouTube, Twitch etc.. At this point, I consider this stuff harmful. Only place online I'm active anymore is here so I don't expect LLMs to be able to replace "the web" for everyone, but for me it might work. Anyone who's a regular user of these platforms should ask himself what he would really, like *really*, miss out on if he dropped them all tomorrow. Are these things truly that important? Couldn't that time be used better? Are they good for you?

had to remove the NANDs and carefully solder new ones
I'm surprised the controller accepted the changed size just like that and didn't immediately brick the device. I'm sure Apple will fix that oversight soon enough.

I think it'll be a "slowly boil the frog" scenario, where you first pay for a "lifetime" key to unlock overclocking or various turbo modes or extras like additional cache you can all go without, (so people critical about this will be screamed down by the usual corpo cattle who will argue with "then just don't do it, you don't NEED it. Gawd! I PERSONALLY appreciate the added value!") then it's additional cores you can normally go without, then it's higher frequencies, then eventually it'll be "all cores except the first" and "all frequencies worth a damn" so it'll be just enough to crawl to the online services to get the cryptographic keys. I don't see Gen Z rebelling against it. Does anyone else?
 
I'm always surprised how people call these AIs "useless".
It's because of how they function. An LLM doesn't tell you facts or truth; it tells you statistically likely strings of text that tend to resemble facts and truth. Everything they output is "hallucination", which is why they will confidently tell you completely made up things alongside things that are real, and if that LLM is the only interface you're using to discover things, you have no way to verify any of what it does. Their "reasoning" is just more of the same, couched in different terms to make it appear like it's having a conversation, when all it's really doing is gaslighting you into believing that its response is something other than the output of complex and poorly documented statistical analyses.
 
Back