how do we make computing fun again? fuck jobs, fuck "productivity", fuck money. how do i have fun again?
make the most overcomplicated version of something you can. try to solve a problem in the worst way you can think of.
>>108565980Use AI. With AI you enhance all aspects of yourself by 100-1000x. AI also extends this enhancement to everything you do. With AI you can easily have 100-1000x the fun of not using AI. Face it. AI is the future.
>>108566005AI is the opposite of fun.
>>108565980dig up some old e-waste and overclock it
>>108565980All fun has left the world and the sooner you can accept that the better.
>>108565980Make silly websites
>>108565980I am so glad I have whatever anti-autism makes agentic programming fun.Codex and Claude Code and even Opencode with random chinese models is the most fun I've had in this stupid job for years.I couldn't care less about the joy of manually typing public static void main and making data model classes and ViewControllers and whatnot.
>>108566196this
>>108566142You're using it wrong then.
>>108566196well? is toad okay?
>>108566196What are you working on? I have been creating docs and scripts for almost a week now trying to get AI to finally finish some migration work.The simple approach didn't work but the tryharding approach is slow, boring and also doesn't work so far.I genuinely don't mean to imply your work is easy, I just want to understand wtf I'm doing wrong, this is even more frustrating than combing stackoverflow for hours.
>>108565980>how do i have fun again?Competitive programming
>>108566005>how I can have fun with code again>JUST GO FULL CYBERKEK IT IS SOOOO GOD MAN, I AM CYBERKEKING MYSELF AND WATCH HOW MACHINE HAVING FUN CREATING, TESTING, MAKING ERRORS, I AM JUST SIT HERE IN THE CORNER ON MY CYBERCHAIR AND WATCH BECAUSE IT IS THE MOST LOGICAL AND EFFECTIVE WAY TO DO THINGS, JUST CYBERKEK YOURSELF TOO, TOU CAN EVEN GET THE RESULTS AFTER 900 ITERATIONS, YOU ABSOLUTELY RIGHT, BRUTAL REALITY CHECK, THIS WHOLE TIME I, THE MACHINE, FUCKED YOUR CODE WHILE YOU WATCHED AND NOW YOU TOO DEEP AND DON'T WANT TO UNDERSTAND AND JUST HOPE THAT NEXT TIME ALL BE GENERATED ALLRIGHR, JUST SITTING AT YOUR CYBERCHAIR AND WATCHING WHILE ALL THE THINGS MAKING CREATIVE PROCESS FUN AND LIFE WORTH LIVING ARE HAPPENING WITHOUT YOUwhat a great advice anon
>>108565980Install OpenIndiana
>>108566651I am working on edge services in a distributed environment (robotics) plus some ancient java 8 crap on the sideBy fun I don't mean that it one shots everything first try. I usually use a combination of chatbot kind of as a pair programming buddy to brainstorm through ideas, and an agent on the actual codebase.When dealing with bugs I usually give the description (plus any logs, screenshots, whatever I have available) to my buddy and tell him to give me a list of likely causes. It always reads the architecture document first so it knows what is connected to what.In 9 out of 10 cases the "most likely" cause ends up being the actual cause of the bug.I then give the cause to an agent to do some digging in the codebase and tell me all the potential places it could happen.Where I still have it break down (even bigger models like Opus 4.6) is not very well documented esoteric libraries, like a lot of GIS stuff. The hallucination rate for those is crazy. It will often just hallucinate a function that does exactly what I want to do, but does not exist.In this case the agents handle than the chat because the chat will keep insisting that it exists or hallucinate another one whereas the agents will see an error and dig around the header files or something to realize it's not there.What helped me there is simply putting whatever documentation they have into the context. If it's a PDF it's easy if it's some stupid website it's a bit harder, in which case I have the agent write a scraping script to get me the documentation as markdown.
>>108565980you're like one of those retards that bought raspberry pi that after the fact wonder what to do with it...
>>108567165Thank you. I think my mistake is probably trying to automate it too much, looking at it, most people are still prompting the model relatively often. I have seen one guy who was able to let a few agents run in parallel for a few hours and I tried to emulate that, but maybe it just doesn't work for some tasks.
>>108567712Yeah I haven't seen anything with a ralph loop or similar work in a way that impressed me and I did a lot of digging on youtube.But a lot of the stuff I work on is not well defined enough to even fit into such a way of working. I often change or make up requirements on the spot so I would need to hold its hand anyway
>>108566659This. I remember isolating my brain from real world for 3 days straight to solve some tricky dynamic programming puzzle. I was literally thinking about it every day all day for 3 days, I couldn't be bother by anything else.
