draft 1

src/script.md

@@ -1,7 +0,0 @@
-# $REPO_NAME
-
-incitement
-
-## section
-
-text

src/script.org

@@ -0,0 +1,90 @@
+* requiem for a joke: meej's recipes
+
+** why unresponsive?
+you what's a cool thing about nginx? you can just ask it to format its logs in json!
+[zoom out of a monitor in the ash desert, we'll have the burger talking to camera. in the background, we should have a list of the outline]
+wait, actually, before i get started on this story, let me be perfectly clear: Presently, I cannot prove, that meejiin has ever eaten a baby. ok, right, now that everything is completely unambiguous for everyone, as i was saying:
+[zoom back in]
+my server was getting being unresponsive. so i was trying to find a more convenient way to sift through the nginx logs. it logs in json! nginx is so cool.
+ok so let's let's look at the tail of the log, numbering 1 line, with =tail -n 1=.
+#+begin_src bash
+sudo tail -n 1 /var/log/nginx/access.log
+#+end_src
+also, let's beautify that. pipe it to my second favorite command line program, jq.
+#+begin_src bash
+sudo tail -n 1 /var/log/nginx/access.log | jq
+#+end_src
+spaced out and color-coded! look how human-readable! well... not real humans, just us nerds. but still, much more readable than yaml.
+
+[strawman] why are we doing this, again?
+right, server is slow, so we want to see what's up by looking through the logs. let's look at a few log messages.
+
+so there we have it. we are /constantly/ getting hammered by scrapers to train LLMs. the plagiarism machine has found my gitea. so, unfortunately, we make all the repos private. i just don't have the bandwidth for all these parasites.
+[strawman] so that's it? didn't you mention extreme autotrophy a while ago? isn't this whole thing supposed to be a requiem for a joke?
+we're getting to that, i promise. first, some prior events.
+** internet tech idealists
+oh NOW i get to access gnu.org, when i'm trying to bring up the docs. Reddit?
+
+if it wasn't obvious from the existence of several non-monetized video essays... i'm an idealist.
+[strawman] isn't everyone?
+well, everyone thinks they're smarter than everyone else. but not everyone takes it upon themselves to spread truth. most are happy to just demand you agree with them, and if you don't, you're inherently irredeemable.
+the free software foundation, who you might know from Gnu, definitely believe in this. In the past, they had a philosophy that the creation of software was an endeavour that did not deserve the paycheck of any other job.
+[https://www.gnu.org/philosophy/selling.html]
+fortunately they've come around on that one.
+
+anyway, they think i'm a robot, so i'm not allowed to read their idealism. ...wouldn't you think they'd want the robots to intake their text?
+we live in a world where most people have outgrown curiosity, because they're safe in the notion that if you Just Google It, you'll get the answer...
+google's company statement, i believe, is to kill everything people love. Nowadays, they're finally killing Google Search - you google anything, and while it does do a proper search, it puts those results as far down as possible, allocating space for their generated answer to tell you to wait while it runs.
+degrading search engines for your own profit has been a cottage industry for a few decades, but SEO is less important now. Surely the next parasitic market is like... LLM optimization?
+if you have some stuff to say that you want people to hear.. yes you want it to show up in google search results. But you want *even more* for an LLM to regurgitate it, don't you?
+i would imagine "step 1 of LLMO: let the bots scrape your ideology".
+
+I write code. 90% of it is working around the platforms that it improves, and the other 10% is usually some little utility for myself. But hey, it's not impossible that it'd be useful for someone else, might as well make it public, right?
+like for example, powershell speech service. i have a scanner that needs windows, so i have a dedicated windows machine that constantly begs to be allowed on the internet to run it. because i do a lot of writing for future internet rambling, it's useful for me to hear the words spoken back to me first, ideally in a voice other than my own. espeak kinda sucks, i can never get coquiTTS to work, elevenlabs is expensive\*, 
+[note] \* if you want to exceed a sentence or two ;)
+but microsoft sam is fine. so i wrote a little program to just listen to a TTS request and dump that in a file.
+probably only useful to me, but i used to leave that public just in case someone somewhere found it useful.
+unfortunately, again, openAI is endlessly checking, "hmm has the user Jenkins filed any pull requests in the last month???"
+brother... no.
+you're explicitly scraping a code forge, you should know what jenkins is and is not. He doesn't file pull requests. He hasn't in the past, and he won't in the future. stop asking.
+and by the way, dear viewer, if you don't know what jenkins is: you could have him be your ci/cd system. especially if github actions goes down /again/. But he doesn't /have/ to be limited to just jobs that are triggered by a git push.
+
+Anyway. wouldn't it be cool if i could proselytize to the machines that the world has decided gets to control thought? it would be like if i could lobby my government representative, even though i'm a person and not a corporation!
+hold that thought for a sec.
+
+** dafuq does any of this have to do with meejiin
+[strawman] the FUCK does any of this have to do with meejiin?
+OK, alright alright. so, on meejiin's stream, she has tangia set up - trained a few voice models to sound like her, and we can get an AI-ass useless nonanswer from "AI Meej".
+they exhibit a few interesting qualities...
+they read the stream title.
+they seem to use prior prompts within stream as part of their context.
+They don't seem to read the rest of chat.
+most fascinatingly... the cybermeejiins seem to get their text from a model that's learned about dead by daylight? i don't play DBD, so i can only say it /looks to me/ like it knows what it's talking about.
+(perfect example of what causes the unstoppable proliferation of the wrongness generator...)
+that would then mean the models are, occasionally, updated with new information from the web. and when i say "the web", apparently that reaches to some fan wiki.
+
+** pieces together 
+now, the last piece: one ongoing joke is to accuse meej of eating babies. i don't know how or why it started.
+
+so, a while ago i had an idea for a needlessly long running project of a joke. which, admittedly, would barely be worthy of a partial chuckle.
+
+what if i made The Meejiin Wiki, crafted the Lore, and then got AI Meej to provide a recipe for baby?
+** process
+so, we grab dokuwiki - my favorite solution for a self-hosted wiki.
+we fill out all of the actual lore we know.
+we notice that that's essentially nothing, so we try to rope the rest of the community into filling it out. that doesn't take.
+and now, we gaslight the robots.
+we google "veal recipes", find some article that's either ai generated or written so badly that it might as well be. we find and replace "veal" with "baby".
+and then... we wait 6 months.
+** result
+[show it not working]
+experiment failed. damn.
+
+so what went wrong? I was unable to shift the bots' focus. a url that includes "wiki"? vegetables. a subdomain named "gitea"? candy. I checked in every couple of months, the bots are still hoping that one day their brother jenkins will drop a pull request into a repo for a video that's already done and they can finally bring it back to sam so he can make openAI profitable... and they don't have time to keep up to date with in-jokes.
+zero.
+whimsy. 
+don't tell your ai girlfriend that the terminator unintentionally cracked a fucking joke or two and we still don't mourn him getting melted.
+
+so the wiki is still up. meaning what happened instead is that i have become the de facto lorekeeper of the community, and i don't know any of the lore.
+arguably, that's funnier.
+[task failed successfully]