<<Prev Next>> Scroll to Bottom
Stuff goes here
L1[00:19:26] <luna​r_sam> Izzy: If only HDDs weren't 6x the price right now
L2[00:19:44] <luna​r_sam> i might have to bite the bullet and get a bigass HDD
L3[00:20:15] <luna​r_sam> might just get a 12TB HDD and split it between games and root
L4[00:20:23] <luna​r_sam> rather, home, not root
L5[00:21:33] <Mic​hiyo> I really wish these 6TB drives I got weren't SAS
L6[00:22:25] ⇦ Quits: Izzy (~izzy@210.1.201.103) (Ping timeout: 190 seconds)
L7[00:29:33] <luna​r_sam> wd more like wmd
L8[00:37:47] <luna​r_sam> aw sweet, my 6TB is destined to die soon too
L9[00:38:22] <Amanda> %choose haha waves
L10[00:38:22] <MichiBot> Ama​nda: Yes, but only if you do it right now.
L11[00:40:34] <Mic​hiyo> I've got 12 6tb SAS drives...
L12[00:40:48] <Mic​hiyo> that do me 0 fucking good, since I can't run the damn storage server because it costs too much to do so
L13[00:41:34] <Mic​hiyo> Which the timing on REALLY pisses me off... I bought them a month before the roommates decided fuck you we're leaving, and I'l STILL paying them off :D
L14[00:43:06] * Amanda flops atop @Michiyo meows about me pee-pee pants, and how he peed his pants again
L15[00:43:13] <Amanda> Mr*
L16[00:43:17] <Amanda> Ffs Google
L17[00:43:49] <Mic​hiyo> o_O
L18[00:55:50] <Amanda> @Michiyo https://m.youtube.com/watch?v=Q5RQQfg173U
L19[00:58:17] <Spider​EveryOS> %tonk'
L20[00:58:19] <Spider​EveryOS> %tonk
L21[00:58:20] <MichiBot> Wow! Spider​EveryOS! You beat your own previous record of <0 (By 5 hours, 35 minutes and 54 seconds)! I hope you're happy!
L22[00:58:21] <MichiBot> SpiderEveryOS's new record is 5 hours, 35 minutes and 54 seconds! No points gained for stealing from yourself. (Lost out on 0.0056)
L23[01:03:59] <luna​r_sam> yeah i ain't gonna buy a WMD Again
L24[05:08:06] *** Joins: Izzy (~izzy@210.1.201.103) ShadowKatStudios
L25[06:36:47] <Forec​aster> %tonk
L26[06:36:47] <MichiBot> Yeah! Forec​aster! You beat Spider​EveryOS's previous record of 5 hours, 35 minutes and 54 seconds (By 2 minutes and 33 seconds)! I hope you're happy!
L27[06:36:48] <MichiBot> Forecaster's new record is 5 hours, 38 minutes and 27 seconds! Forecaster also gained 0.00024 (0.00004 x 6) tonk points for stealing the tonk. Position #2. Need 0.294238 more points to pass Spider​EveryOS!
L28[06:47:50] <Spider​EveryOS> %sip
L29[06:47:50] <MichiBot> You drink a runny void potion (New!). SpiderEveryOS feels the need to smash. (%smash)
L30[06:47:56] <Spider​EveryOS> %smash
L31[06:47:56] <MichiBot> SpiderEveryOS smashes a box. It turns out to be a loot crate. It contained One Thing.
L32[06:52:58] <Va​ur> %chug
L33[06:52:59] <MichiBot> You drink a punctual pear potion (New!). As Vaur drinks the potion they seem to have become magnetic and A living tombstone flies towards them! They successfully evaded it with a 14 vs DC 14!
L34[07:25:58] <Spider​EveryOS> %give MichiBot an air fryer
L35[07:25:58] * MichiBot accepts the air fryer and adds it to their inventory
L36[08:52:07] <Izzy> lunar_sam: yeah that's why I went for used server drives. pair of 4TB drives, cables, and a controller card was $5 cheaper than a single 4TB SATA drive
L37[08:52:58] ⇦ Quits: Hawk777 (~Hawk777@2605:1700:1080:de00:27ee:10f3:bbd0:f2c1) (Quit: Leaving.)
L38[10:23:32] ⇨ Joins: vftdan (b5cb05d1b9@2a00:c70:1:178:170:40:189:1)
L39[10:48:34] * Amanda meows and looks around
L40[10:48:41] * Amanda orders MichiBot some %soup
L41[10:48:41] <MichiBot> Waiter!! Waiter!! Theres a document giving her ownership of the moon in my soup!
L42[11:33:53] <Amanda> #soup
L43[11:34:05] <Amanda> Alas, no 2michi3bot
L44[11:59:41] * Amanda curls up atop nadja, yawns a big toofy yawn
L45[12:11:19] <luna​r_sam> fuck it, welcome back toshiba drive
L46[12:59:17] <Forec​aster> %tonkout
L47[12:59:17] <MichiBot> Hooray! Forec​aster! You beat your own previous record of 5 hours, 38 minutes and 27 seconds (By 44 minutes and 2 seconds)! I hope you're happy!
L48[12:59:18] <MichiBot> Forec​aster has tonked out! Tonk has been reset! They gained 0.006 tonk points! plus 0.01 bonus points for consecutive hours! Current score: 1.495286, Position #2 Need 0.278238 more points to pass Spider​EveryOS!
L49[13:35:02] <Corded> > <as​ie> it just came out. and it is also heavily AI assisted so who…
L50[13:35:02] <Re​nno> idk what the deal is with the whole anti ai assisted when it comes to modding
L51[13:35:22] <Corded> > <Re​nno> idk what the deal is with the whole anti ai assisted when i…
L52[13:35:22] <as​ie> i have seen what the existence of adobe creative cloud did to art communities
L53[13:35:31] <as​ie> or what the existence of CAD subscription models did to design communities
L54[13:35:48] <as​ie> i do not want programming, which was and remains one of the cheapest hobbies to get into, to end up the same way
L55[13:36:17] <as​ie> i do not want hobby programming activities to split people into "haves" and "have-nots", especially as everyone expects OpenAI and Anthropic to further limit the scope and/or increase the price of their rate-limited subscription offering
L56[13:37:03] <as​ie> furthermore, the legal situation of LLM generated code remains precarious: that is why the FSF is staying away, for example, as is OpenJDK on Oracle's lawyers' advice (ironically, Oracle itself is all in on the LLM datacenter craze)
L57[13:37:31] <as​ie> it will probably clear itself up, but not in all jurisdictions at the same time, and not in the same ways. my open source code should be legal globally
L58[13:38:15] <as​ie> furthermore, there are all the ethical problems with how large language models are trained and monetized. mass scraping of internet data without consent or compensation, relying on low-paid labor for RLHF, uncontrolled datacenter buildouts which lead to citizen complaints
L59[13:38:47] <as​ie> but that's the *general* problem
L60[13:39:16] <as​ie> the problem with such a porting project specifically begins when the author admits he doesn't have much modding experience or Scala experience. LLMs can only help you insofar as you know when they are wrong, and they cannot solve every issue that may crop up in a mod port.
L61[13:39:24] <as​ie> so, it is concerning in that sense
L62[13:41:34] <Re​nno> to be honest I don't think it really matters much about a mod
L63[13:41:38] <Re​nno> like, it's a mod
L64[13:41:43] <Re​nno> all of the rest of that is basically overkill
L65[13:41:45] <as​ie> oh, i'd argue on the contrary: it matters *the most* in places like modding
L66[13:41:58] <as​ie> modding, and hobby activities, were historically the places where new programmers were minted
L67[13:42:02] <as​ie> teenagers who caught a coding bug this way
L68[13:42:02] <Re​nno> I'd rather have OC than not
L69[13:42:45] <as​ie> sure, but you wanted to know why some people are concerned about the trend
L70[13:43:00] <as​ie> everyone at this point is relying on some LLM generated code. doesn't mean i have to like it
L71[13:43:12] <Re​nno> I want to know why they're overly concerned more specifically
L72[13:43:20] <as​ie> i mean i gave you my answer
L73[13:43:27] <Re​nno> right
L74[13:43:31] <Re​nno> still seems overly concerned
L75[13:43:58] <as​ie> i mean in 2014 i was concerned about curseforge's encroachment on the modding scene
L76[13:44:11] <as​ie> most of what i felt at the time did not come to pass, curseforge is nowadays criticized but for very different reasons than i had anticipated
L77[13:44:16] <as​ie> so maybe my concerns will not come to pass here either!
L78[13:44:50] <Re​nno> I mean some of those are valid sure enough, it just seems like a lot of the OC people hear AI and immediatly kinda stuff their fingers in their ears
L79[13:45:01] <as​ie> "a lot of the OC people" what's your sample size?
L80[13:45:07] <Re​nno> about 10 or so
L81[13:45:10] <Re​nno> of the few that remain
L82[13:45:12] <as​ie> the largest OC communities are the GT:NH Discord, then the Minecraft Computer Mods Discord
L83[13:45:14] <as​ie> this is a distant third
L84[13:45:24] <as​ie> dunno GT:NH, but the MCM Discord is very much in favor of LLMs
L85[13:45:46] <Re​nno> seems to be more of the old school OC players
L86[13:45:50] <Re​nno> that feel that way
L87[13:45:59] <as​ie> "older people are more concerned about new technologies" tale as old as humanity itself
L88[13:46:11] <Re​nno> when the fact is that it can actually write code and it is really dependent upon the process
L89[13:46:24] <Re​nno> so it's not as simple as "it's written by ai so it must be slop"
L90[13:46:26] <as​ie> I can actually write code too, but if I decide to charge you money for my efforts suddenly that's not that useful innit
L91[13:47:17] <as​ie> see: the recent crashout people are having over GitHub Copilot ditching the subsidized model
L92[13:47:28] <Re​nno> that particular port will most likely need significant review and testing for obvious reasons, however they are going through the trouble of iterating and testing it actively even if they themselves are not coders
L93[13:47:35] <Corded> > <Re​nno> so it's not as simple as "it's written by ai so it must be …
L94[13:47:35] <as​ie> I never said that
L95[13:47:47] <as​ie> I said "the developer has little experience and is embarking on a complex mod, they might reach issues which the LLM is not equipped to solve"
L96[13:47:50] <Re​nno> I know, I wasn't quoting you. that's just the general sentiment it seems
L97[13:48:08] <as​ie> because most people have developed their opinions on LLMs primarily in the context of labor rights and ethics
L98[13:48:25] <as​ie> which is perfectly fine, but also, makes it hard to talk to people who are right now benefitting from LLM generated code
L99[13:48:29] <as​ie> because they don't believe such a thing could exist
L100[13:48:43] <Re​nno> that cat is out of the bag tbh
L101[13:48:48] <Re​nno> has been for a while
L102[13:48:49] <as​ie> it is and isn't
L103[13:48:59] <as​ie> so long as local models are not mainstream all of this can go away
L104[13:49:46] <as​ie> in addition, communities can still separate concerns and use honor codes
L105[13:49:54] <as​ie> if they want
L106[13:50:21] <as​ie> i know many people who use LLMs at work but refuse to use LLMs outside of work, for personal reasons, like "in my spare time i want to learn things myself, in my work time i want to get work done quickly"
L107[13:50:40] <Lizzy> IMO, i don't care for LLM-generated stuff and will generally avoid using something if it's made using it. If anyone wants to use LLMs to help them, then that's their choice, just don't bring it anywhere near stuff i work on
L108[13:51:09] <as​ie> i've given up - the amount of open source projects using LLMs in some capacity is too large to really run away from it now
L109[13:51:25] <as​ie> linux, freebsd, netbsd, openbsd, and probably windows/macos all ship LLM generated code as part of the base system
L110[13:51:37] <as​ie> not to mention the many applications
L111[13:52:03] <as​ie> the no-ai lists are comparatively miniscule - some nice projects on there too, yes, but in general, if you compare the "ai-contaminated" vs "ai-resistant" lists, the victor seems obvious
L112[13:52:19] <as​ie> however, in modding, very few old-timey modders have adopted LLMs in general
L113[13:52:21] <as​ie> i have been doing research
L114[13:52:38] <as​ie> there are a few who have! but most LLM usage is *new* creators and there the inexperience and sometimes overconfidence concerns me
L115[13:52:51] <as​ie> this appears to be the same in other spaces, like game boy homebrew (which has good data via itch.io disclosures)
L116[13:53:28] <as​ie> my point is: i refuse to make my open source projects assume, as part of their day to day development, tools which require prospective contributors to pay good money for a subscription
L117[13:53:46] <as​ie> whether this means "no LLMs allowed" or "i'm building a local LLM server" remains to be seen
L118[13:54:20] <as​ie> it's the same reason for which my projects are Linux first: Windows and macOS are much pricier 😛
L119[14:04:16] <Forec​aster> ooh, what a reveal!
L120[14:04:18] <Forec​aster> https://discordembeds.pc-logix.com/live/1511369873618501733/1780409057937-0904c55db60f4.png
L121[14:04:32] <Forec​aster> A ship launched "fighter" that can land!
L122[14:04:39] <Forec​aster> designed for exploration
L123[14:04:45] <Forec​aster> exciting
L124[14:11:26] <Lizzy> oh good, getting spam via the forum's contact form
L125[14:14:03] <Lizzy> perhaps turning off reCAPTCHA wasn't the best idea
L126[14:14:07] <as​ie> probably
L127[14:16:28] <Lizzy> turned it back on using reCAPCHA v2, we'll see if this helps
L128[14:24:13] *** Joins: Vexatos (~Vexatos@2001:4647:af75:0:2143:c483:ae52:16bb) Vexatos
L129[14:24:13] zsh sets mode: +v on Vexatos
L130[15:21:21] <Va​ur> %tonk
L131[15:21:21] <MichiBot> Woooo! Va​ur! You beat Forec​aster's previous record of <0 (By 2 hours, 22 minutes and 4 seconds)! I hope you're happy!
L132[15:21:22] <MichiBot> Vaur's new record is 2 hours, 22 minutes and 4 seconds! Vaur also gained 0.00237 tonk points for stealing the tonk. Position #3. Need 0.177256 more points to pass Forec​aster!
L133[17:58:55] <Forec​aster> %tonk
L134[17:58:56] <MichiBot> Hooray! Forec​aster! You beat Va​ur's previous record of 2 hours, 22 minutes and 4 seconds (By 15 minutes and 30 seconds)! I hope you're happy!
L135[17:58:57] <MichiBot> Forecaster's new record is 2 hours, 37 minutes and 34 seconds! Forecaster also gained 0.00078 (0.00026 x 3) tonk points for stealing the tonk. Position #2. Need 0.277458 more points to pass Spider​EveryOS!
L136[21:42:26] * Amanda flops back down atop nadja, meows about her adventure across the land called "The Interstate" cutting through another weird land called "Ohio"
L137[21:43:54] <Mic​hiyo> Fuck Ohio. lol
L138[21:43:58] <nadja> Amanda: sounds fun, you should do more normal boring adventures like that that no not involve multiple gravity wells
L139[21:44:33] <Amanda> But it's sooo boring, it gives me a headache!
L140[21:44:53] <Mic​hiyo> Someone tried to pickpocket me in a Greyound station in Ohio... and instead ripped the ass out of my pants. Had to finish the trip with ripped damn jeans
L141[21:52:18] <Izzy> The whole LLM situation is making "write a shim to run Lua directly on top of openfirmware" more and more attractive by the day
L142[21:53:14] <Izzy> Hopefully the slop will slow down now that the cost has gone up like 1000x
L143[21:56:39] <Izzy> Ah well, at least it is kinda funny to see people freak out over the price hike.
L144[22:07:08] <as​ie> *sigh*
L145[22:07:31] <as​ie> It has only gone up for Copilot. The services from the vendors themselves (OpenAI, Anthropic, Google) continue to cost as much as they did.
L146[22:07:47] <as​ie> Copilot was the cheapest option, but plenty of people simply shifted to the lowest Codex or Claude plan
L147[22:08:16] <as​ie> As long as those are in play, the slop won't slow down. Once these go, I think that's when it will slow down.
L148[22:08:38] <as​ie> (Also, the price differential is more akin to 10-25x, not 1000x)
L149[22:09:01] <as​ie> * 10-50x, not 1000x)
L150[22:09:08] <as​ie> * depending on user, not 1000x)
L151[22:10:52] <Izzy> Ahh
L152[22:11:08] <Izzy> I can't say I'm keeping close tabs on it
L153[22:12:05] <Vexatos> idk I code for fun
L154[22:12:07] <Vexatos> AI codes for me
L155[22:12:10] <Vexatos> why would I use it
L156[22:12:21] <Vexatos> I ask the same at work
L157[22:12:24] <Izzy> ^
L158[22:12:30] <Vexatos> every colleague I have uses AI to code
L159[22:12:34] <Vexatos> when it's the best part of the job
L160[22:12:45] <Vexatos> what a waste
L161[22:13:12] <Vexatos> I didn't study colour theory to have a bot make my plotting script for me
L162[22:14:06] <Vexatos> also asie: My work recently denied a request to get a group license for claude, citing security concerns (telling a cloud LLM about your still secret research)
L163[22:14:17] <Vexatos> what happened was that each of my colleagues now has a personal claude subscription
L164[22:14:29] <Vexatos> which they use at work
L165[22:14:33] <CompanionCube> iirc i think i saw microsoft stop paying for claude because it wasn't worth it for them?
L166[22:14:48] <Vexatos> except for my boss
L167[22:14:52] <Vexatos> who has a chatgpt one instead
L168[22:15:47] <Vexatos> for me, LLMs have made my work a few orders of magnitude more difficult and there is nothing I can do about it
L169[22:15:51] <Izzy> Mostly I just want to ignore the LLMs but they keep breaking software I use
L170[22:16:04] <Vexatos> I have to do so much work because of them it's crazy
L171[22:16:11] <Vexatos> I cannot ignore it
L172[22:16:41] <Izzy> Thankfully I am not employed in a field where it's omnipresent
L173[22:16:42] <Spider​EveryOS> %tonkout
L174[22:16:42] <MichiBot> Dad-Sizzle! Spider​EveryOS! You beat Forec​aster's previous record of 2 hours, 37 minutes and 34 seconds (By 1 hour, 40 minutes and 12 seconds)! I hope you're happy!
L175[22:16:43] <MichiBot> Spider​EveryOS has stolen the tonkout! Tonk has been reset! They gained 0.004 tonk points! plus 0.003 bonus points for consecutive hours! (Reduced to 50% because stealing) Current score: 1.782024. Position #1
L176[22:17:11] <Vexatos> before LLMs, you basically ran on good faith in research
L177[22:17:14] <SquidDev> At my old place, we had people who did not understand the codebase checking in slop and breaking things every few days. It was exhausting.
L178[22:17:30] <Vexatos> so when you read a paper, it was peer reviewed so you could be sure of the scientific soundness
L179[22:17:43] <Vexatos> and then you assumed they would publish their research in good faith
L180[22:17:46] <Vexatos> i.e. not lying on purpose
L181[22:18:06] <Vexatos> nowadays, they are still peer-reviewed, so still scientifically sound in theory
L182[22:18:12] <Vexatos> and they are still in good faith by default
L183[22:18:27] <Vexatos> except now, they might have asked an LLM to generate their plotting script based on a csv file
L184[22:18:40] <Vexatos> and you cannot be 100% sure that they checked the script to ensure it actually used the data from the csv
L185[22:18:46] <Vexatos> and didn't make up random numbers
L186[22:18:58] <Vexatos> because this happens a lot
L187[22:19:13] <Izzy> Tangentially if anyone submits a LLM generated merge request to any of my OC projects I will go to their house and eviscerate them with a camera mount
L188[22:19:16] <Vexatos> so now when doing research I have to look at the raw data for every single graph in the paper
L189[22:19:19] <Vexatos> and compare numbers
L190[22:19:30] <Vexatos> to make sure the graph actually has the same contents as the data
L191[22:19:48] <Vexatos> and THEN I have to hope the data, if it was processed using an LLM-generated script, was processed correctly
L192[22:19:55] <CompanionCube> Izzy: what would be your platform of choice for the lua-ofw shim, thugh?
L193[22:20:04] <Vexatos> so it's basically just all a lot more shitty and a lot more work
L194[22:20:10] <Vexatos> and it doesn't matter if you use LLMs or not
L195[22:20:14] <Vexatos> because everyone else does
L196[22:20:17] <Izzy> I have recently obtained another PPC mac
L197[22:20:39] <Izzy> They're not cheap but they are cheaper than modern computers
L198[22:20:41] <CompanionCube> isn't the openfirmware implementation on some of those....not great?
L199[22:21:09] <Izzy> all I need from it really is the ofw console / framebuffer and malloc
L200[22:21:44] <CompanionCube> what, no no i/o?
L201[22:22:16] <Izzy> I can muddle through that myself if need be
L202[22:23:24] <Izzy> Though I meant more for getting "a lua environment" running. Ideally I'd be able to talk to ofw from the Lua side at that point
L203[22:24:48] <Izzy> Wrappers for calling into ofw and for submitting PCI ... requests?
L204[22:29:53] <CompanionCube> (ah, apparently the 'good' version is 3.0 on 'new world' macs)
L205[22:30:10] <Izzy> These are G4s so that's new world
L206[22:30:34] ⇦ Quits: Vexatos (~Vexatos@2001:4647:af75:0:2143:c483:ae52:16bb) (Quit: Insert quantum chemistry joke here)
L207[22:31:23] <Izzy> https://cdn.shadowkat.net/media/6a/55/26/6a5526172102f7482e66d5d2ebeac86f591cc23229b604ea5a95344b313f9a2e.png
L208[22:32:25] <CompanionCube> aesthetic
L209[22:34:00] <Izzy> If I can scare up a GPU with any hardware acceleration it will be very usable
L210[22:34:18] <luna​r_sam> oh ac8 is a $70 game with a battlepass
L211[22:34:28] <Izzy> The GF4MX is too old for proper support so even moving windows around can be slow
L212[22:34:30] <luna​r_sam> abysmal dogshit
L213[22:34:58] <Mic​hiyo> I remember my MX440...
L214[22:35:35] <Izzy> Because this is Apple I can't just stick any AGP 8x card in
L215[22:35:49] <Izzy> I need to fuckin kapton tape some pins and it will only run at 4x
L216[22:36:21] <Izzy> Also flash the card firmware but that's not a big deal
L217[22:42:16] <Izzy> but on the upside, 2x1Ghz G4s is pretty fast!
L218[22:42:44] <Izzy> it is at the very least no worse than the Atom N270 in those IdeaPad S10e netbooks I keep around
L219[23:01:54] <Amanda> %choose candy or pasta
L220[23:01:54] <MichiBot> Ama​nda: A nearby lamp replies "Not candy!".
L221[23:02:05] <Amanda> Hrm... Sounds like it's candy time!
L222[23:28:48] <Ocawes​ome101> Is ofw available on anything non-ppc?
L223[23:29:03] <Izzy> it's on most *nix workstations and can be built for x86
<<Prev Next>> Scroll to Top