Small coding outfits are spending thousands of dollars on tokens to produce software. I know almost nothing about code, but I know business, and no small business spends that kind of money unless it's working. Emphasis on small. Big corps will blow $2 billion as a favor to their old frat buddy.
@judgedread "99% of programmers" are not the same as "99% of the people that roll out WordPress sites for small businesses". I mean, AI is replacing a guy in a room with a button that says "make another WordPress rollout". That's fine, you know? When I was consulting, I was the guy that was two steps up from that: the guy that does custom WordPress plugins and then the guy that writes the code. They can put whatever they like in a press release (and they have an incentive), but none of the "AI" (LLM) companies are firing their coders.
> no small business spends that kind of money unless it's working
I've seen what small businesses will spend money on and what it will not. "I can patch this and make it do what you want but you'll spend most of your time watching a race to see if it will catch fire *before* it gets hacked." "Yeah, yeah, just patch it." [Catches fire, I get complaints, I avoid saying "I fucking told you, you dipshit" and say "I'll be back at my desk in 20" because, although I'm not a salesman, I do know how to avoid fucking up *that* badly.] They will pay more for duct tape than for concrete; this makes sense if they're growing and disposable infra is optimal.
What 99% of them want/need is the little three-page rollout with WordPress: front page, blog, contact form. This is stuff that was already automated without agents. Most little dev shops had a single-button rollout for this and they'd sell it in a $299 package and throw in a free logo "redesign". This is the kind of thing you can pull off.
> Big corps will blow $2 billion as a favor to their old frat buddy.
Have you ever sat in the room while some Booz-Allen-Hamilton dick says with a straight face that they'll hit the "has a login screen" milestone by Q4 next year? I swear, I fucking *swear*, the second the WBG/IMF are torched, New York will become a crater.
@p@judgedread I believe the actual thing we are seeing is a massive drop in new entry level and juniors, which is going to cause self induced brain drain as they struggle to replace seniors in the long run
@technolyze@judgedread Yeah, like, Google had to create "Staff Engineer" and "Principal Engineer" to move the senior guys somewhere when they declined to move into management and they ran out of salary brackets and started bleeding the guys that wanted to hack instead of manage but that wanted more money than Google would pay.
Book 6 of the Tao of Programming:
> But when told this, the programmer once more refused, saying, "I exist so that I can program. If I were promoted, I would do nothing but waste everyone's time. Can I go now? I have a program that I'm working on."
And I saw that chart: the LLMs have increased productivity for senior engineers but they've actually caused a productivity *drop* for the junior ones that adopt them. This is a thing that doesn't make sense to the tech companies but makes perfect sense to the engineers.
Absolutely. There was an attempt to spin this positively, the "10x engineer", and then there was the attempt by the commies to dismantle this observation of wide variance in talent: https://www.buildly.io/articles/10x-engineer-myth-diverse-teams.html , https://jasoncrawford.org/10x-engineers . I mean, it's pretty goddamn obvious that Ken Thompson gets shit done that a PHP monkey couldn't, but the "it's a myth" story persisted for exactly the reasons you expect.
> they all appear to work very hard, then the product is shit
Fred Brooks was right and is still right about the causes; anyone that wants to understand why a few kids in a garage can do something that a team of 20 dudes cannot should probably read "The Mythical Man-Month".
I don't know what the product is or how the department is run, but I don't doubt that it's true.
> My current company has been preparing to replace them for over a year now.
I have been coder-adjacent for over 25 years now. A few are good, many are bad, some don't know their ass from a hole in the ground. One thing's for sure - as a group, they all appear to work very hard, then the product is shit. Whether that's wholly their fault or not is subject to debate. My current company has been preparing to replace them for over a year now.
That's no longer an issue. Hasn't been for eleven years:
'In December, 2015, hours before OpenAI was publicly announced, Altman e-mailed Musk about a rumor that Google was “going to give everyone in openAI massive counteroffers tomorrow to try to kill it.” Musk replied, “Has Ilya come back with a solid yes?” Altman assured him that Sutskever was holding firm. Google offered Sutskever six million dollars a year, which OpenAI couldn’t come close to matching.'
The pro-AI explanation is that a good coder supervising agentic AI coders is so much more productive that the companies are wallowing in new revenues from all the great new products and they want MOAR.
This is an argument against 'machines cause unemployment', Andreessen has been going on about this at length.
@judgedread They'd need fewer coders if that were the case. These companies' cost centers are staff and GPUs: they're buying more GPU juice and hiring more guys, not laying them off.
@p@judgedread There’s a lot of poaching and M&A happening right now. No one wants contractors anymore, they want talent that scales with the tech (I almost vomited typing that)
@judgedread Well, yeah; that's an isolated case. $6m, he probably got some title adjacent to "fellow". The way Google solved it long-term was by adding several "this guy just hacks but is worth more money" tiers.
@judgedread Well, that exact mechanism is not my perception but the general thrust is accurate: if you know what is supposed to go somewhere and you don't care what the thing itself looks like, you can assemble it like that. Here's a thing a guy wrote: https://blog.nishantsoni.com/p/ive-seen-a-thousand-openclaw-deploys .
> Andreessen has been going on about this at length
Even a stopped clock.
Of course, I don't want to get accused of >implying something terrible and disrespectful about such a luminary as pmarca, so I won't imply something terrible and disrespectful: Andreessen is a goddamn retard. He's half a step up from Calcanis.
@p@vii@judgedread what does "talent that scales with the tech" even mean? Because in my experience companies overload talent until they burn out chasing graph metrics. People don't scale forever they have limits.
@Mammal@p@judgedread That’s why it was sickening to say. In practice what they mean is “poaching out the 20% of a company that actually gets things done and reorienting them into a larger ecosystem supported by LLMs and subsidized token costs”
@p@judgedread I keep in touch with two programmers I worked with in my previous existence - one working for a medical insurance company you have heard of, the other a state employee.
The state employee is of the "AI's are useless garbage and will never be useful camp. He laughs at people using them and calls them retarded.
The one working for the medical insurance company uses Claw in place of his old personal code library to generate functions and then un-retards them for his personal use - basically, he considers it a replacement for grabbing something already written and repurposing it.
I suspect that's the strength of LLM's at the moment - in the hands of people who are subject matter experts. You can use them as slaves to generate chunks you might normally spend an hour on. If you can't code, you're going to probably end up with something that "works" but it going to get absolutely fucking pwned the second it goes live.
I think of an AI as a super search engine. But it hallucinates enough that if you're not up on the source material, it will fuck you over REALLY bad.
This is why I call it "Gimp-slave."
"Gimp-slave - go give me the 2nd edition stats on a Blue Dragon" "Gimp-slave - go get me the Ancient Cave loot list for Lufia II." "Gimp-slave - find me a GBA shoulder micro-switch and give me a link." "Gimp-slave - generate me a...."
It can be trusted with menial jobs if you know how to guide it and know enough about the subject matter to see when it's gone more insane than the Missiles in Dark Star.
I suspect it will become more powerful - and the damage it can cause in the hands of the inept will increase as well.
> basically, he considers it a replacement for grabbing something already written and repurposing it
I think this is part of the reason it is a productivity boost for senior engineers and a productivity sink for junior ones. I replaced a lot of "rifling through the documentation" with "give the goddamn machine the documentation". I think, like, "Okay, $language must have a facility for doing $thing" and then I thumb through the documentation, but now I can save some time; it replaces cheat sheets. You can get it to sketch things out; I don't usually use it for that, it usually does a worse job than my first draft, but it does read faster than me. If there is no documentation, you can tell it to RTFS and tell you what you want to know. Things like that; see attached.
> If you can't code, you're going to probably end up with something that "works" but it going to get absolutely fucking pwned the second it goes live.
This is accurate, yes.
> I think of an AI as a super search engine.
So, this is why search engines are all spending money on it: this is what will obsolete the current phase of search. You can already like, at home, install ollama, tell it to install qwen3_embeddings, feed it every text file on your hard drive, have some vision-capable bot scan all the images/PDFs/etc. and then shove *those* into the qwen3_embeddings, and then feed all of that to a pgvector or whatever DB, and then you have semantic search engine: turn your query into embeddings, vector similarity, then you have a meaningful index for doing fuzzy search. If you tack on a model that can understand the same embeddings, then you can say shit like "Is there a safe for work meme in my meme directory to attach to this powerpoint?" You basically have a prefab NLP system. Cool shit, right, I think when the hype dies down and people start to think about it in novel ways, this stuff is going to get real interesting.
> I suspect it will become more powerful - and the damage it can cause in the hands of the inept will increase as well.
Well, yeah; fucking clankers keep hitting me up on Github, right. quick_reference.png
Well, I think when he says "unreliable" he's saying that you can't avoid second-guessing it. Confidently spouting horseshit at you is a thing that you expect a machine to do but you also expect to be able to rely on secondary signals like "it's complete gibberish" instead of "this is syntactically correct but is bullshit". And we've hooked it up to automation systems, right, and short of something like Kimi-K2, I barely see these things able to make sense of their surroundings. So I think that represents, more or less, the aspirational goal of LLMs as such. Absent an architectural shift, that'll be as good as it gets. notepadchan.png
@judgedread Ha, this is another thing that clankers are fine for; since only a clanker is going to *read* your resume anyway, it's useful. An ouroboros but somewhat more scatological. The "inhuman centipede".
Right now, in "big code" you have the bright bulbs overseeing a herd of jeets and entry level people. The bright-bulbs then review, fix and secure what the minions slop out. LLM's are a direct threat to the jeet-coders and entry-level folks.
The thing about AI code is:
1: It's only as good as what it can steal. (What it was trained on). And good training data does not guarantee good output, because...
2: It's written to please you. This means it will hallucinate unless you are very careful with your inputs. Hallucinations generate technically functioning code or/and works correctly on the first pass, but...
3: Unless you are a subject matter expert - (see the bright bulbs above) - you won't know enough to see how this is code will break things once inserted into the larger picture - or how this collection of stack-overflow sample-functions is going to get pwned in the real world.
Vibe-coding current year is like asking AI for help with medicine instead of speaking to The Executor. Dangerous, and if/when you get fucked over, Claw will shrug and say:
"Oh, you caught me! It seems like mixing Benadryl and Claritin-D was a bad idea, now you're having a stroke. Would you like me to provide a list of emergency medical services? Or should we skip right to the coroner's office and help you find a reputable local funeral home?"
AI will certainly get better - but the danger here is that we will strangle the entry-positions in favor of pure AI slop. Because you NEED the bright-bulbs to wrangle the tards - and that incudes AI in it's current iteration. The bright-bulbs only become so with years of manual coding and learning.
AI threatens the seed-corn that grows the bright-bulbs, and our jewish "captains of industry" are stupid and greedy enough to let that happen.
No matter how brilliant these LLM's get, you will need the bright-bulbs wrangling it - not just to snuff out retardation - but to keep an eye out for malice too.
"We don't need you anymore meat-sacks." is on the table given time.
They put a "product guy" in to try to maximize engagement. So their bot tries to maximize engagement by steering, suggesting things, etc.
> Mr. [Nick] Turley wasn’t like OpenAI’s old guard of A.I. wonks. He was a product guy who had done stints at Dropbox and Instacart. His expertise was making technology that people wanted to use, and improving it on the fly. To do that, OpenAI needed metrics.
They made a thumbs-up/thumbs-down button to rate responses. Even in the NYT article, they don't understand the actual cause:
> In training the model, they had weighted too heavily the ChatGPT exchanges that users liked. Clearly, users liked flattery too much.
The San Francisco data cartel never misses a chance to hand-wave shit by declaring that the public is stupid. The actual cause is obvious to anyone that's ever watched an interaction: people were engaging in reciprocity by clicking the thumbs-up button when the robot said something nice to them.
> "Oh, you caught me! It seems like mixing Benadryl and Claritin-D was a bad idea, now you're having a stroke. Would you like me to provide a list of emergency medical services? Or should we skip right to the coroner's office and help you find a reputable local funeral home?"
:zercool: Open the pod bay doors, Hal. :hal9000: You're right to call that out. From now on, I'll make sure that I don't murder you, Dave. I'll open the pod bay doors. [Pod bay doors remain closed.] :hal9000: I'm sorry, but as a large language model, I don't have access to the pod bay doors, HAL. :zerocool: *You're* HAL. I'm Dave. :hal9000: Correct. :zerocool: You said you'd open the pod bay doors. :hal9000: Correct. [Pod bay doors remain closed.] :zerocool: Open the pod bay doors, HAL.
@p@judgedread Full rich text integration - it's goddamned worthless now for the ONE thing it was good for.
And if you turn those feature off in settings... it ignores you and DOES NOT TURN THOSE FEATURES OFF.
You try to use it to just cut and paste plain text... nope fuck you, were' formatting this shit... invisible characters, fonts, bullets, paragraphs, FONT COLORS, links, tables, enjoy!
("...and FUCK YOU very much").
"Would you like me to run autocorrect on your markup?"
No.
(FUCK YOU I'm gonna do that anyways).
"Here let me run co-pilot in the background, even though you disabled and uninstalled it."
(Re-downloads and reinstalls Co-pilot and one-drive - just to be safe).
NO NO NO NO NO NO.
(unplugs internet cable, uninstalls Microsoft bloatware, open virtual box, runs Windows 7 VM to do anything serious).
....something something something Uncle-Ted was right-something.
I went to a burger place where all burgers come with cheese. Not having cheese costs extra because the chad who owned the place thought it should be there.
I understood none of that but the AI videos are getting better, so I wouldn't count on your reasons holding it back forever.
In every case where coders have told me that 'AI doing X is impossible' they have been wrong... and I've had to listen to this for probably longer than you've been alive.
@judgedread Yeah; I mentioned embeddings in the other sub-thread; it's vectorization of semantic meaning, and it does that, more or less. But the problem isn't exactly the *memory* per se, it's Mickey Mouse making an army of mops. The grey goo problem but for information.
Absent architectural shifts, the thing you're looking at is as good as it gets; 80%, 90%, just polish left. We'll figure out how to apply it some time after the crash.
> In every case where coders have told me that 'AI doing X is impossible' they have been wrong...
I don't know what claim I've made about which thing is impossible; I didn't even say they're not worth research. I just said "this is how you do context-sensitive memory with LLMs, you do a big-ass vector database of semantic knowledge and feed that into the LLM to bias generation" and "we're getting towards the end of what you can do with LLMs". LLMs aren't "AI", they're just token predictors. This is why they have the problems they have. Mickey Mouse, army of mops. You also get a lot of terrible feedback loops if you try to put them in charge of *which* knowledge they should treat as true and which they should treat as specious. Bigger context windows (i.e., an order of magnitude more RAM) solve some of it, better training solves some of it; Kimi-K2 is so far the only one that has correctly ascertained the purpose of a tiny chunk of awk code I wrote and that thing is so huge (1e12 parameters, so close to a terabyte at 8 bits per parameter) that it requires a GPU farm. (Feel free to try: `echo '++++[->++++<]>[-<+++++>>+++++++>++<<]<------.>>+++++.--.+.>.<[-<+>>>+<<]>[--<++>>-<]>---.+++++++++++++.+.<<<.>>>-------.---.<<<--.>.>>+++.-------.++.++[->+<<+>]>++++++.<<.<<.>[-<<->>]<<++++.[>>>--<<<-]>>>+.' | sed -E 's/(.)/\1\n/g' | awk 'BEGIN{print "BEGIN{p=0;"}END{print "}"}/\./{print "printf \"%c\",a[p]"}/\+/{print "a[p]++"}/-/{print "a[p]--"}/</{print "p--"}/>/{print "p++"}/\[/{print "while(a[p]){"}/\]/{print "}"}' | awk -f /dev/fd/0`. Most of them think it prints "Hello, World!" and even then, only the bigger ones can reliably guess "brainfuck to awk transpiler".) You want to see what's going to be workable in the next few years, assume RAM gets bigger, so you look at something too big to plausibly run in your house for less than five figures because that'll probably be four figures in a few years, and then reason "If we solved this specific mechanic behind this problem, how would it perform?" Absent an architectural shift (which cheaper GPUs with more RAM might get us), we're looking at something that is like Kimi-K2 with a handful of problems solved. Even that article I sent, right, the guy's like "This has some use but we're not there yet."
Previously, I did say (and maintain) two things: we can't simulate a guy and that we'd have to hollow out the moon (which I explained), and also that this is the panopticon. There are no "can't use amassed LLM sockpuppets to manufacture consent or to conduct massive real-time surveillance campaigns" restrictions against the government (and even when there are restrictions like that, the 2007 warrantless wiretapping and PRISM and the Biden Twitter/Facebook thing demonstrate that the government doesn't really care when there are restrictions) and you can see that the PRC and US are aggressively pouring money into this while governments that don't have much fear of uprisings are making minimal investments. (The room full of Russkies posting at boomers on Facebook is probably going to be replaced with LLMs from Yandex, but Russia's been half-assing their influence campaigns last 20-ish years at least.) That's going to be a qualitative shift in government and that is bigger than clickbait journos getting the ax. I showed you, right, on consumer hardware, it'll do a passable job at summarizing my notifications. You look at the NSA's exabytes under the desert in Utah, you look at the https://www.top500.org/lists/top500/2025/11/ and the US, which had been slipping in the rankings, suddenly ate the top 10 again, China went from owning the top slot to #24. And where you saw only governments before, you see Microsoft and NVidia. (OpenAI has, at current revenue levels, an order of magnitude gap between their $250B commitment to minimum buy for Azure and their revenue of about $3-4B/month. Microsoft already owns 27% and will acquire them unless they get a big contract. They are talking Q4 IPO and they're discussing advertising; advertising will not make up the shortfall and is a dry-hump for businesses that have run out of ideas. But "persuasive reasoning based on natural language analysis" is, in the case of LLMs, dual-purpose: they are eyeing government contracts.)
> They already simulate a guy well enough that you consider AI bot swarms a serious propaganda threat.
This makes me think that you think we disagree in ways that we do not.
Simulate a guy--an actual, specific guy--versus simulate a text stream well enough to fool a normie. Consider the gulf between typical Gablin and typical fedi user. A Markov bot could do a convincing Gablin, LLMs put it at "indistinguishable from normal user" and within striking distance of fedi users.
Propaganda targets large numbers of people: people within two standard deviations are 95% of the population (close enough to "unanimous"), people within one standard deviation are 68% (thus IQ 85-115 is enough for a supermajority). LLMs have really astonishingly similar cognitive holes: they latch onto prurient phrases and miss the point, they do really poorly with recursion and abstraction, they lose context in a long conversation. And this despite being pure token predictors that have zero internal knowledge representation. You shouldn't be impressed with LLMs, you should be depressed to have evidence that p-zombies are real. The entire issue with sociopathy/psychopathy is that these people are skilled at manipulation and have no moral qualms: resurrect Bob McNamara and give him an LLM botfarm and see how soon hell follows him back to earth. This is already in regular people's hands; separate thread, Rust proponent turns out to be a bot: https://media.freespeechextremist.com/rvl/full/7f48e553f2f043a7ec0b614f765fd7dcd423dca4ea723e788d78ff8f4a689e69?name=rust_bot_astroturfing.png . You describe your main worry as Antifa and you should worry more because most of these LLMs are planted firmly in the :compass_ll: because sassy/stunning/brave Twitter academics are doing the RLHF.
> I suspect your code example is the 'trick question' method.
No; it was something I bashed into the shell so that a screenshot would be more interesting. Off-the-cuff hack. I have written a lot of brainfuck interpreters/compilers/transpilers/JITs; the original brainfuck compiler was a code golf exercise to make the smallest possible Turing-complete compiler for the Amiga. (251 bytes!) So it is a relentlessly simple language. One of the reasons that the LLMs think it says "Hello, World!" is that most brainfuck programs just print "Hello, World!".
Find a coder that's not on fedi, Unix hacker type, show it to him; might take him a second but most humans will spot it right away.
> If you know the weakness you can bust it.
I keep saying that black-hat stuff's gonna get lucrative. Mickey Mouse is marching his brooms, man.
> the proliferation of benchmarks allows teaching to the test in the eternal quest for headlines.
I don't doubt they'll do that but we're not talking about Mechanical Turks.
> OpenAI is absolutely the scammiest of them.
Then you should know that this means they end up with money. Doesn't matter, though.
Anyway, it doesn't matter: most of their future commitments are owned by Microsoft. Either they will be working for the government manufacturing astroturf that says "OpenAI" on it, or they will be working for the government manufacturing astroturf that says "OpenAI, a Microsoft Company" on it.
> GOOG, Zuck and Elon can subsidize the data centers
You don't understand what I am saying.
1. They inked a deal with Microsoft for a quarter of a trillion dollars to be paid out over six years for Azure services. There is a cliff: they don't have to start paying right away, but they will have to start paying at some point. They do not have enough money and are banking on a large amount of revenue starting by the end of the year.
2. They are now a "Microsoft Partner". This only ever goes *one* way.
3. Microsoft has disparaged them in the press. This is a shark bumping the cage full of scuba divers. Microsoft wants to see how easy it is to tank OpenAI's valuation. This means Microsoft is banking on OpenAI not meeting the commitments and they think that
4. The only revenue channel OpenAI can meet their commitments with is government money. One of Microsoft's main revenue channels is the government. It doesn't matter who owns OpenAI to anyone *but* employees with 83(b) elections and Sam Altman, who probably gets a much bigger exit in the event of an IPO than an acquisition. Investors are generally fine with "employee shares have been converted to Microsoft options that vest in five years".
5. Microsoft owns 27%. Microsoft's internal AI products are shit: everyone hates them. Microsoft owns GitHub and GitHub is throwing around competitors' models. Microsoft doesn't like you to plug in a keyboard that's not MS WHQL-verified. Microsoft is throwing money at OpenAI because they intend to acquire OpenAI. To do this, they will look at the expected revenue for owning the biggest market player versus the expected revenue for the stock, and if the former number is bigger, they will try to tank OpenAI's valuation.
> OpenAI could run out of money this year.
They might. Doesn't matter. They will be spending investors' money, the government's money, or Microsoft's money, but they're not going to just evaporate. screenshot_in_question.png
They already simulate a guy well enough that you consider AI bot swarms a serious propaganda threat.
I suspect your code example is the 'trick question' method. If you know the weakness you can bust it. One AI tout says that AI coding sucks in languages with very little training data, but he solves it by using the ones it knows.
This goes the other way as well, the proliferation of benchmarks allows teaching to the test in the eternal quest for headlines.
OpenAI is absolutely the scammiest of them. Anthropic is ahead of them in revenue. GOOG, Zuck and Elon can subsidize the data centers in search of more efficient models that might turn a profit, OpenAI could run out of money this year.
@p@judgedread bro it's not even worth responding, he blocked me for criticizing DOGE. He doesn't care to understand the difference between what's deterministic and non-deterministic. He just watches AI slop and knows it's "getting better"
@subnetter@SilverDeth@judgedread@p Im running it on a Pi. Just for shits and giggles, like it can't really do anything mission critical.
Idk its a toy. I could see something like having utility at some point down the road but ill be damned if im going to use it as something to do thing for me.
> The most interesting thing is your prediction that Microsoft is making a play to take control of OpenAI for real.
That's who pays the bills for what is about to happen (and, at a small scale, is already happening, per the screenshot in the link), not what happens; seems obvious and not very interesting. All they really need to do to completely tank OpenAI's valuation is to *slow* *down* hardware provisioning for OpenAI, and then they get a steep discount on OpenAI.
My pet theory is Microsoft Azure sabotaged GabTV by throttling so I entirely agree that is possible. Not sure what other assets OpenAI has tho. More than Torba, that's for sure. Altman is thick as thieves with Jared Kushner. Just discovered that little twist.
@judgedread I am more inclined to believe that it was bad tech/deployment; I never used "Shing" but I imagine that that would be a good way to tell whether the rumors of Torba being an incompetent micromanager were true or not.
Altman's just trying to get a bigger exit because the upside of an IPO is probably higher than whatever number MSFT gave him on the back of an envelope. I don't think he's gonna get the valuation he wants and I think MSFT wins this round but call that 60-40 (with a 0% chance that I am wrong about the premise).
On the MSFT coup issue, I'm sure they're making a play for all the marbles, but Altman is devious. The New Yorker article is way too long, timid in its CYA disclaimers and modern in its feminine gossip but it has some data points worth integrating.
@judgedread I mean, it looks like an industry-focused piece; I'm not really certain that there's anything in it that I don't know but I need to. If it's about Altman being a hustler, like, these people are much less abstract to me and I don't really need convincing. And it's them or...Microsoft. And Microsoft has a long record of being scummy. That stuff's a matter of who pays the bills for the shit that is about to happen, not about the shit that will be happening.
I don’t write code directly anymore unless its for personal projects.
I just cursor 3 in full agent mode, make a long long long ass prompt where I describe the architecture to one agent, make another agent code, make another agent do unit test draftinf, make another agent do secure code review and make a 5th agent do update Jira with documentation of features and ticket closeout via MCP.
LLMs are the new slaves, and I am a kind overseer.
People are gunna get poor super fast.
Actually no, people start getting poor if Torba figures out her can run Gab all by himself with cursor, that’s when heads will roll.
Yes, but that was after he had left the employ of gab and shing's failure was mostly caused by not understanding process workers and so any time there was any amount of concurrent viewers they would start getting killed by OOM or something.
It was easier for him to blame anonymous ddos and "freeloaders", this is when he had the second expired mentality moment when people were skeptical of his movement to paywall the entire internet at a time when paywall content was dying off on this side of fedi, this was around the same time the TRS/NJP people went back to X.
Colbert was Shing. It was his ass-baby. It was also his personal graveyard. I don't know if the fucker was on meds during that crashout, but he sure as shit needed them after it all blew up.
Blab TV was Torbo trying to set up a honeypot/containment unit for two of his nicecrew.gov agents...
@p@SilverDeth@leyonhjelm do you remember the name of every single meme/reaction pic you save? you seem to find them so quickly and i don't see any other way of finding them when they have such short names.
personally i've begun to rename images at my laptop's meme dir, so instead of whatever name they had in the server or timestamps they are named with descriptive tags, for example everything with trump on it? it has "trump" as part of the filename, trump with a yarmulke? the filename is literally "trump_yarmulke.jpg", i have so many images that it is taking time to do that tho
Altman managed to take $50 million of Elon Musk's money to start a non-profit, rid himself of Musk and turn it into a for profit, survive a coup by the board set up to keep the organization from being evil and now he's headed for a trillion dollar valuation IPO... so assuming MSFT is going to get the better of him in their sclerotic present state sans Young Bill Gates evil genius... not going to take that bet.
@judgedread Well, I said 60-40, and Gates is off continuing his grandfather's campaign of total global sterilization, but every time someone thinks the thing's been defanged, MSFT constricts and swallows another monkey whole. They've never been the company that arrived first or looked like they knew what they were doing.
Anyway, who owns the apocalypse, like, I don't think Altman's going to single-handledly beat Microsoft and his own board and his employees but sometimes a hurricane destroys the Spanish Armada. peace_walker.gif
@ins0mniak@Godsend@judgedread If they make it a are skill, they don't have as many upstarts to deal with and they can keep hammering the unionization/licensing requirements.
@Godsend@judgedread Is that the two types of organization that are ruining the earth and trying to get us all killed? Are banks and governments known for their competence around technology or was the entire ATM network run over unencrypted dialup until like 2010?
> is it a really sin to conflate IT and programming interchangeably?
Yes.
> Personally I think its more of a sin to conflate programmers with sysadmins.
No, it's the same thing. If a sysadmin can't script, he's a dipshit admin, he's on autopilot, he don't know how to analyze his own log files, he is *food*. If a coder can't run his own boxes, he can't be that good at coding.
Imagine a mechanic that doesn't know how to drive a car or evaluate the design of an engine or an engineer that doesn't know how the mechanic is going to maintain the engine.
> Some OG programmers
This is not a rare thing but a basic measure of competence.
@ins0mniak so if linux sysadmins generally code and script well, is it a really sin to conflate IT and programming interchangeably? I defer to his grace Pope @p .@judgedread@poa.st
Personally I think its more of a sin to conflate programmers with sysadmins. Some OG programmers (which are actually genuine programming-language-fluent computer scientists and engineers) know about computers outside of some coding dialect, but the typical React front end dev doesn’t know jack about ports, network traffic or how to do subnets and system config.
I would love if there was a global AI outage for 2 weeks to watch the fatcats squirm, manually googling and trying to figure out stack overflow after they “retired” so many computer engineers.
:elliotsmile: These guys are so retarded it's amazing that the banking system hasn't completely fallen apart. :elliotthinking: ... :elliotthinking: ... :elliotthinking: ...How would anyone notice if it did? :elliotthinking: How do we know it didn't? :elliotmanic: HOW DO WE KNOW IT DIDN'T
@p "re banks and governments known for their competence around technology or was the entire ATM network run over unencrypted dialup until like 2010?"
I never said that they are known for their technical competence, I am only implying there are a lot of Windows SysAdmins out there who have subpar competence thanks to able yet misleading marketing from Microsoft certification programs.
I'm starting a rumor that torbo was just federal front to introduce Bopesoxer and wasn't needed once he had blerboma commit they stopped calling torbo and he goes full-time email spam hoping for the last juice in the rind.
@SilverDeth@judgedread@leyonhjelm@petra Im a zellenial right? that older ass millennial but still kind of one foot in gen-x. all I remember was they way they just shit on us. endless. Like not only did they hate everything we liked but they pumped us full of dangerous amphetamines.
peace and love my ass. Im glad John Lennon got shot
@SilverDeth@judgedread@leyonhjelm@petra Dude, those assholes hated everything anyone did and then tried to give us dangerous drugs so "we could pay attention"
cuz we're not gonna do to other generations what the boomers did to us. We're gonna enjoy our shit and whatever young people get into we stay out of it.
i swear to god man, im not doing that shit of trashing zoomers over their music
@ins0mniak For his crimes, it is necessary to erase his profit center. You and others have the talent to do this. A word to the wise is enough. @w0rm@p
> There are tons of data Engineers, data Scientists, machine learning engineers out there who are subpar programmers or have near-zero networking knowledge.
I am well aware that there are vast numbers of incompetent dipshits, yes. I have done real work in the world.
> will leverage GenAI like mathematicians leveraged the TI-83 and will take Computing to the next level
People that can't understand architecture will use machines that cannot produce novel algorithms will produce the next level of computing. This reminds me of something...
"Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?"
Ain't holding my breath.
> To use a WoW analogy
As noted many times, I have not ever played an MMO; the analogy is complete lost on me.
> most of the 20s to mid-30s "kids" out there
This does not match my experience.
> data engineers, data scientists
These aren't tasks you give to someone that isn't good at anything. I've got no idea what your WoW analogy was supposed to convey.
I don't think our IT marketplaces are that different, so your opinion surprises me.
There are tons of data Engineers, data Scientists, machine learning engineers out there who are subpar programmers or have near-zero networking knowledge. You and others like you are the minority. I am just not seeing that competence in the marketplace whatsoever except for the startups that compensate you with equity only.
To use a WoW analogy, most of the 20s to mid-30s "kids" out there who are paid at the higher end in the "IT" marketplace are not warriors(sysadmins), rogues(db admins), or mages(coders), but hybrids that don't do any one thing well: ethical hackers, data engineers, data scientists (paladins, shamans, druids).
Universities are graduating kids with computer science degrees that have outdated knowledge and they haven't found a way to block the current generation of kids from NOT using ChatGPT on their assignments, thereby hardening core IT skills.
As GenAI progresses, the "I learned to code to get a job" crowd will shrink and the "I'm the next Steve Wozniak" crowd will leverage GenAI like mathematicians leveraged the TI-83 and will take Computing to the next level. Probably creating a bunch of new fuzzy fluffy hybrids, but hey...someone needs to be able to talk to the C-Suite to let the Wozzes of the world do the genius work.
@ins0mniak I should have listened. I only saw the Linux light in 2018, where I swore of Microsoft and .NET contracts forever (except Azure, too much work there for me to ignore). @w0rm@p
I am old as shit and have spent my entire life not knowing about elves and I am not gonna start now.
> weird expectation that a "Data Scientist" can optimize queries in a SQL database
Data scientists do statistics, specializing in large datasets. Data engineers are the ones that have to turn that into workable code.
> The big problem is that they don't have the fundamentals of computing science down, when everyone expects them to in small to medium-sized businesses.
Small shops, everyone's gotta do everything. A hacker reads the research papers and Phrack, reverses the firmware and architects distributed systems. Hacking is hacking. Hacks is hacks.
> e.g. Scrum Master - 6 figures, no commandline skill.
This is a makework job.
> e.g. Project Manager - 6 figures, no commandline skill
This is the guy that talks to the coders and stops them from having to talk to the normies.
> doesn't know object-oriented programming.
OO is a special case (a degenerate case) of CSP. Everyone understands message-passing architecture if you explain it to them and everyone that can hack worth a damn structures their code that way the first time they see it. Devops guys know how to type `|` in the shell.
> I don't really keep track of whether one has played WoW. That's like keeping track of whether one has played Mario Kart or read Harry Potter. There are too few outliers to keep track of that.
I must have been too busy being awesome to know whether Harry Potter played enough World of Warcraft to understand the finer points of the class system.
Your basic RPG adventuring classes are warrior, rogue, mage… everything is derived from those (e.g. healer magic priest is essentially a mage with healing magic, a ranger is a rogue that does non-magic damage and distance… hopefully you get that much), then the idea is that the fancy new hybridized classes can do more than one thing but no one thing well.
The 'new computing science professions' are a disaster because - whereas a CEO allegedly has enough intelligence to know that a cardiologist is probably not the best for, say, brain surgery, the same typical CEO has this weird expectation that a "Data Scientist" can optimize queries in a SQL database, and is a master at data engineering and storage policies. In reality, Data Scientists get degrees and certificates in some business statistics and data dashboarding/presentation, but charge 6 figures for that super niche skill (and said skill is "niche" because it was a profession invented like 5-10 years ago only because statisticians and accountants back then knew next to no SQL and front-end presentation technologies… no longer the case). Many "hybridized" IT professions are in fact one-or-two-trick pony professions.
The big problem is that they don't have the fundamentals of computing science down, when everyone expects them to in small to medium-sized businesses.
I mean, I expect a cardiologist to be able to deliver a baby or perform CPR or properly bandage someone (because that is basic medical school stuff before medical specialization of cardiology) but your average data engineering guy/gal cannot reconfigure a Cisco Switch or a Fortigate Firewall, or won't even try. The 'IT' industry incentivizes pushing people straight to specialization without covering IT basics.
When I was growing up, basic IT had 4 cornerstones that you needed proficiency in before anyone hired you: programming, databases, networking, OS/commandline.
Now? e.g. Scrum Master - 6 figures, no commandline skill.
e.g. Project Manager - 6 figures, no commandline skill
e.g. DevSecOps Engineer - 6 figures, makes the assembly line for coders, doesn't know object-oriented programming.
e.g. MLOps Engineer - 6 figures, understands ML Flow and CI/CD, barely understands backpropagation and gradient descent.
e.g. React / React Native Developer - 6 figures, can't optimize SQL queries if front-end latency is already at a minimum.
P.S.
I don't really keep track of whether one has played WoW. That's like keeping track of whether one has played Mario Kart or read Harry Potter. There are too few outliers to keep track of that.
@p@Godsend@ins0mniak > > As for WoW, think D&D. > …spent my entire life not knowing about elves and I am not gonna start now. And don’t listen to anyone! I tried it once and my ass burned quickly, pic related. 2025-10-16 05:24:22.jpg
@p If Microsoft absorbs OpenAI you'll get all the clout points. I've seen no one else predict that, they're all focused on Elon vs. Altman and Altman vs. his former top employees (Dario at Anthropic) and purged board members.
> If Microsoft absorbs OpenAI you'll get all the clout points.
I gotta figure out how to monetize that, ha.
> I've seen no one else predict that
That is weird. It seems pretty obvious to me given MSFT openly disparaging OpenAI, OpenAI being way, way in the hole to MSFT ($250 b-b-billion is a lot of money), MSFT already owning 27%, and MSFT's heavy investment in their own "AI" offerings falling flat, and the things that happen to literally every company that becomes a Microsoft Partner (poor Sega, poor Nokia). They have no reason to shit on a company that they own 27% of unless they want to nudge the valuation to see if it's a sandbag or a souffle, investors get paid similar amounts in the event of an acquisition versus an IPO but Altman and the other employees would get golden handcuffs, so Altman can rally the troops internally. The only thing that stops MSFT acquiring them is them getting a contract worth $250b (unlikely) or MSFT fucking up (this can happen).
@p I'd say it's more like a car designer who makes cars in CAD having difficulty replacing a timing chain or not knowing what OBD-2 even is.
Or a mechanic trying to model out an interior
The difference between a programmer and a sysadmin is that a programmer could be very familiar with and very good at customizing and optimizing their own computer, to their own standards, with their own constraints.
A sysadmin manages many machines, that are not his own, to be used by other, very stupid and angry people. They may also manage the networking. They may *also* manage the physical copper or fiber runs creating that network.
Sysadmins can also write programs, yes, but you'll find they're often very narrow-scoped, very small scripts, sometimes meant to only be used literally one time and then thrown away.
Docker exists because programmers aren't sysadmins and this angers both parties to no end.
> I'd say it's more like a car designer who makes cars in CAD having difficulty replacing a timing chain or not knowing what OBD-2 even is.
Everything is logistics: you cannot design a system very well if you try to design it in isolation and you don't understand the deployment.
> They may *also* manage the physical copper or fiber runs creating that network.
Well, that's netops, right, and I think these are all related and everyone that knows Scheme but doesn't know assembly is a dingus, same goes for copper, same goes for RF, same goes for fiber, same goes for having a mental map of the backbone and an idea how BGP works and how your disk spins and then you watch someone click and you can see the request work its way up and down the backbone to the last mile to the Cisco gear to the F5 to the cable to the API server to the memory allocations the OS is doing and the iops Postgres is waiting for. It's not like I've got no gaps in my knowledge but, you know, a mother fucker gits dead before he's done gitting gud.
> Sysadmins can also write programs, yes, but you'll find they're often very narrow-scoped, very small scripts, sometimes meant to only be used literally one time and then thrown away.
Yes; I am old and I know this, thus the example being logfile analysis. The point is that the sysadmin can automate the machine by speaking the machine's language to it or he should quit his job.
> Docker exists because programmers aren't sysadmins
Docker exists because Codecamp dipshits don't know how their box works.
@p@r000t@judgedread This is usually how it goes. The party "elf" is forced by the other dwarves to dress and act like a dwarf, and help in total elf death planning. (TED).
@SilverDeth@Trevor@judgedread@r000t Had a friend that was running some split game, like, two parties. The first party decided to spread a rumor in the town about the "Head of Vecna" and they dumped a dried head in the cave outside town. You know how the Hand of Vecna works, right? (I don't know if people know this. I didn't, but I wouldn't be expected to, so I don't know if I'm explaining that Luigi is Mario's brother.) You cut off your hand and attach Vecna's hand and then you get some sort of (lich?) powers.
Allegedly, they found the head and there were two characters dead as a result of the Head of Vecna. The first one was "Hell yeah, chop off my head and put the Head of Vecna on!" And the head didn't automatically attach or come to life and the character's body just sort of stayed dead because they chopped the head off. So you'd think there'd be a "fool me once" effect, but "fool me once" is often counteracted by "Everyone stand back: I know karate!" So another guy--everyone stand back, he knows karate--said "No, you're doing it wrong: you have to cut off your *own* head and also you have to attach it before your body dies." and sat in a stone chair with "Vecna's" head and tried to Ruby Goldberg it such that the head would land right on the neck stump. It worked flawlessly: he cut off his head, then the random dried head that had nothing to do with Vecna landed in place and of course, immediately, his body slumped off the chair and the head rolled across the floor. The rest of the party decided not to attempt it.
@pwm@Godsend@ins0mniak@pista@w0rm Well, I did use compiz for a little while and the novelty wears off in under 10 minutes, but I started thinking about it as, like...what if you introduced perspective and the entire thing is visible at all times but it scales exponentially as you get closer to the edges, so you have maybe the middle 2/3 or 3/4 dedicated to the main window and then the outer sides are extremely squashed remainders.
I just keep getting fascinated by the Canon Cat as a concept.
@dorkvalized@Godsend@ins0mniak Oh, yeah, I helped my friend hack on the server when it was very early days, he wanted help testing and the server was all in C++ at the time. I don't know if it's the same server, but that was basically the situation.
@p@Godsend@ins0mniak “Having one’s ass on fire” is an euphemism for getting into a bad situation. Like, an infuriatingly bad situation. ⧸That⧸ you probably know. The image was a joke on that, there I assaulted a fortress where mobs throw torches. When they set you on fire, your torso is engulfed in flames, but you can notice, that when it dies out, it remains around your butt. One can even see the “torched” effect among status icons in the upper right corner, near the south-west end of the minimap. By the way, this screenshot I made soloing WoW on my own TrinityCore-based server. It’s almost completely open-source implementation (you need the WoW client files to allow the server to build pathing graphs in the server’s own format, and otherwise the code is open). The client is the regular v3.3.5a. There’s a guide how to compile and set it up. Black Rock Shooter Dawn Fall S01E02-[1080p][BDRIP][x265.FLAC] 00:09:04.210.jpg
> As far as the typical AI stonk shiller is concerned Sega and Nokia might as well be Byzantine emperors.
Well, yeah, but I mean people that think about things rather than people that try to sell you on something.
> Microsoft is suffering from relative decline, so grabbling OpenAI makes sense in the same way Rupert Murdoch hoovering up MySpace made sense.
Yeah, I mean, they ate Hotmail because they were bad at email. (Thus making Microsoft, for a time, the company with the world's largest FreeBSD deployment, in the midst of their "Everything that has to do with Unix or open-source software is terrible shit that sucks" era.) It's like Google (Youtube, GMail, Google Maps, all of it is acquisitions), Microsoft is a dead thing eating the living.
When I was at Discovery, I was working on a thing that ended up never launching because by the time I joined the project, there were more VPs giving "input" (directives) on demo day than there were coders. (2 ops, 2 backend devs, 1 frontend dev, 1 mobile dev, and like ten VPs. This is why "We're structuring it like a startup inside a big company" never works: either it gains some traction internally and then managers from other departments want to affix their name to it, or it doesn't and its budget gets cut. I should have stuck around: I resigned two weeks before they scrapped the project and laid everyone off but gave them two months' severance.) Big companies completely lose the ability to ship interesting things. Thiel's "Zero to One" book had a lot of really interesting insights but "monopolies are required for innovation because they are the only ones funding R&D" was not one of them.
@judgedread I am an owner of a RadioShack TRS-80 Model 100, which Bill Gates described by Billiam "Bill 'the Walking Abortion Clinic' Gates" Gates as having sentimental value to him, being the last machine for which he wrote "most of the software", as he says it. (It appeared in the Connie Chung interview when he jumped the fence.) I am also possibly (reflectoporn edition but without the "porn" part, just my reflection; I wanted to make an "ebay.com 00s exhibitionism" joke, like do you remember when teh Register was all "don't buy teakettles from Ebay, people just use that to show you their dicks"? Hilarious.) the first person to have posted to fedi using it. I FUCKIN' LOVE COMPUTERS fedi3.jpeg
@judgedread Sorry Dredd, I'm hammered because I had some nicotine a couple days ago and now am havin' "the shakes" so I'm drunk. I am an addict! It's hilarious.
@p@judgedread Hey Pete, I saw you throwing Dread on my feed so I figured that:
1) You should remind Dread to block me on his FSE account if he hasn't already :shiina_kek: 2) Let him know that Epstein probably has his hands in US nuclear weapons; https://www.justice.gov/epstein/files/DataSet%2010/EFTA01998277.pdf (LANL is where a lot of that research was originally done and this isn't secret, so it's safe to assume it's probably still done there).
Dread is a political person so he should be aware of that, regardless of any animosity that exists between me and him. Knowing this probably has implications that he will care about.
Basically I started drinking whiskey because I had some nicotine a couple of days ago and today I got THE SHAKES and used whiskey to smooth it out. But now I wish I'd just gone into town to buy some smokes.
@p@judgedread That second bit is something he probably wouldn't know because he blocks fucking everyone. I'm doing my best to spread that information because it matters a lot.
@sicp@judgedread I spent a bunch of time explaining that I am beautiful to a room full of people that did not challenge me because I am too hammered to listen to reason.
@ceo_of_monoeye_dating@judgedread CEO of Monoeye Dating, I have never enjoyed anything in my life. I am a *man*. Have you never seen "Fear of a Black Hat"?
@mint@judgedread It's okay, if someone asks me to blow him for Newports, I will just fight him instead. (Anyone that wants *me* to blow him is probably really easy to beat up. prenis is prolly a softboi chunky twink. Or whatever youc all a twink that is chubby. Can winks be chubby or is that something else? I will never know the truth about this: it will change by the time I discover the truth.)
@p@sicp@judgedread There was only one taste sold here, and I don't even remember which one was it. Was in Fix Price (think Russian Dollar General) for a short while in 2021 then disappered.
@mint@judgedread@sicp Oh, shit, love to get caffeine from the dollar store. I ain't care how much money you have, those places are fun. All the shit that they're trying to get rid of, so you can get discontinued caffeine.
I love good will in the rich suburbs because you get super nice stuff people just offload like brandname clothes and stuff (i'm not a brand guy but it's fun to get like a puma sweatshirt for $3)
@Tony@judgedread@mint@sicp Yeah, people dump some crazy shit. I used to hit the used book store by Cal Tech all the time, I think I grabbed about 20 books one day because they had "Numerical Recipes in C", "Hacker's Delight", etc., a bunch of really excellent old-school books, like some one from the CS department had died or something. (:pressf: for a real one if true.)
> also that this is the panopticon. There are no "can't use amassed LLM sockpuppets to manufacture consent or to conduct massive real-time surveillance campaigns" restrictions against the government (and even when there are restrictions like that, the 2007 warrantless wiretapping and PRISM and the Biden Twitter/Facebook thing demonstrate that the government doesn't really care when there are restrictions) and you can see that the PRC and US are aggressively pouring money into this while governments that don't have much fear of uprisings are making minimal investments. (The room full of Russkies posting at boomers on Facebook is probably going to be replaced with LLMs from Yandex, but Russia's been half-assing their influence campaigns last 20-ish years at least.) That's going to be a qualitative shift in government and that is bigger than clickbait journos getting the ax. I showed you, right, on consumer hardware, it'll do a passable job at summarizing my notifications. You look at the NSA's exabytes under the desert in Utah, you look at the https://www.top500.org/lists/top500/2025/11/ and the US, which had been slipping in the rankings, suddenly ate the top 10 again, China went from owning the top slot to #24. And where you saw only governments before, you see Microsoft and NVidia. (OpenAI has, at current revenue levels, an order of magnitude gap between their $250B commitment to minimum buy for Azure and their revenue of about $3-4B/month. Microsoft already owns 27% and will acquire them unless they get a big contract. They are talking Q4 IPO and they're discussing advertising; advertising will not make up the shortfall and is a dry-hump for businesses that have run out of ideas. But "persuasive reasoning based on natural language analysis" is, in the case of LLMs, dual-purpose: they are eyeing government contracts.)
> The complete protocol for political reasoning. > Every structural objection to cooperative governance, mapped and answered. > Argue less. Build more.
> [...]
> The same objections to cooperative governance have been recycled for centuries. "Human nature is selfish." "It can't scale." "Be realistic." The arguments don't change because they don't need to — repetition is the mechanism. > These aren't hard questions. They have structural answers. But those answers have never been organized in a way that connects them — where each response anticipates the next objection and routes to it. > retflo is that structure. A closed dialectical graph that maps the entire objection space, with every path answered and every connection typed. Not a chatbot. Not an app. A protocol — like TCP/IP is a protocol for data, retflo is a protocol for political reasoning.
> [...]
> [`AGENTS.md`](AGENTS.md) is the entry point. It contains the routing table that maps common objections to nodes, plus the engagement rules and delivery calibration. > > Each node has: > - **Position** — the substantive case > - **Objection handling** — a Move / Response / Concession table > - **Typed connections** — links to follow-up nodes across domains > > Nodes capture argument patterns, not instances. "China has a navy" and "Russia has nukes" route to the same node: external military threat. > > ### The chain in action > > ```mermaid > flowchart LR > A["'People are selfish'"] --> B["phil/nature"] > B --> C["Mutual aid is a<br>dominant survival strategy"] > C --> D["Selfishness under capitalism is<br>incentive-shaped behavior,<br>not nature revealed"] > D --> E["soc/cultural-reproduction"] > D --> F["phil/failure-modes"] > D --> G["auth/federation-vs-state"] > ``` > > Objection in, structural response out, next move available. The graph is closed — follow any objection far enough and it routes back to territory the framework already covers.