Discussion
RE: https://freeradical.zone/@lew/117349332920194189
more llm booster doublethink from the #reticulum mines?
there is no spare time except for spending months on an llm boondoggle. let's walk in the forest while barraging CO₂ into the atmosphere. building grassroots networks while shovelling money into megacorps
@cleeyv bizzare that ecological concerns are not mentioned here *at all*.
deepseek is standing on the shoulders of claude et al.
and it's just piling on the justifications as per usual...
"painstakingly adjusting every line" - you could have just wrote it yourself then?
"the goal of reaching a higher level of synthesis and symbolic density than I could have achieved without that assistance" - just discuss your ideas with your friends?
🤷♀️ 🤷♀️ 🤷♀️
@d1 It seems that the point is to get away from python to allow more variety of devices to be used by using rust. If it achieves its goals and is tested and works with the existing versions incl the definitive reference then perhaps one should not be so vitriolic? Indeed you are being deliberately offensive which weakens your "case". A difference of opinion is fine. Just make it factual not ad hominem.
@adingbatponder it might be their point but it wasn't mine. toot edited to reflect that, apologies to the author, i was going overboard
@d1 @adingbatponder d1, I agree with your (now edited) point about my morale. Simplified summary:
1. Our consensus: empowering users to be indepentent of big corporation's control is a goal.
2. d1's point: morale and identity forbids collaboration with the enemy. Always stay on the good side, don't support the evil side.
3. My point: everything is a trade-off, the outcome is important. Every day, I make compromises when I interact with people and organizations that I view critically.
@lew @adingbatponder my point is: the end does not justify the means
this is how i explain the cognitive dissonance in your post
for me, there must be unity of means and ends
without this, we are not moving towards our intended goals but end up with something else
in your case, e.g. code which few understand, additional reliance on megacorps and forest fires
you want to empower users but you end up empowering megacorps in the long-term
Install bonfire.cafe
Get the full app experience
@d1 the developer of the python implementation, who originally created Reticulum, has made his opinion on LLM use pretty clear, in a chapter recently added to the manual:
https://reticulum.network/manual/brandolinis.html#assistance-versus-machine-substitution
"The public discussion around machine learning and software development has been derailed, by the loudest voices on both sides, into a farcical fight between “AI enthusiasts” and “AI haters”.
Both positions are absolute caricatures.
The fight is beside the point in the ways that matters most.
It is not an accident, but a very conscious choice that the initial analysis of the Prns case examined herein, and this very chapter, were themself produced with the assistance of a machine-learning based generative system.
Specifically, it was generated using lc managing a Q8_K_XL quant of DeepSeek v4 Flash running in the llama.cpp inference engine."
Brandolini’s Reference - Reticulum Network Stack 1.5.5 documentationContentsMenuExpandLight modeDark modeAuto light/dark, in light modeAuto light/dark, in dark mode