Hi. KindlyDoRight. Happy that you saw in my perspective something useful.

I've read couple reports on propriatery social media (ex Twitter. links from other sites (1) since i dont have account )  that already 'programmers' use bigLLM to reverse engineer games object code and then instruct bigLLMs to do various change to that code.

So in a strange irony bigLLM has started shooting on the food the big corporations that support it (security wise for now mostly). 

The essense of the issue is that for a bigLLM object code is functionally seen as open source. So i guess the next move is keys everywhere ?

I cant help but notice a social aspect here. The 'programmers' who try 'opensourcing' closed binaries obviously have access to  enough money in order to use these bigLLM's and obviously they dont have a respect for opensource or anything. But judging by how they use the 'opensource' word obviously they dont have a clue and  they mean - I can do what ever i want without respecting other peoples efforts and intentions .Which in a sense is a reflection of the ethics that bigLLM has shown sofar with the 'i can steal every well formed byte that exists in the internet'.   And thus they use 'opensource' as synonymous for stealing ...  Another trait of the ultra rich wrecking car ethos. Lets throw all big concepts on the llm blender and have fun scratching our itches.

They remind me of the ethics of spoiled rich types that use their cash with a mentatlity - WOW lets f*** everybody. Social norm are useless.Lets drive on the highway @as fast as we can and see how many innocent bystanders  we can kill in the process.

So IF we had a social scientist we could possible hear that bigLLM is not the essence. Perhaps pathological concentration of power , money , computing , comm etc can create havoc in a society. And why the social scientist would help us? Because we would understand that bigLLM is a manifestation of bigPower . Just a tentacle. So patholigal power imbalances are manifested here.  If most people had the money to have a personal pc with 128GB able to run local llms or access to local distributed llm servers then probably we would see the llm with less critical attitude .. or less fearfull. Because i think pathological power concentration imprints fear in us. A powerfull entity could guide and control our lives or endanger our lives . 

I get the strange feeling that whole earth has become a playground for relative young ultra rich bros . Ironically the nvidia ceo who now poses as the hypedAI semigod has build his fortune on gpu for playing addictive games...

refs :
@ 6-Oct-2026 (Tue) Wherein the slopbros won't stop by @ the Xscrensaver programmer + DiscoClub owner pissed to AI guy!!!

notes:
(1) https://stratechery.com/2026/apple-and- … rs-future/

----------- for later integration ----
Interesting things i've learned about LLMs.

• agents from LLM users use wikis (like wikipedia) or discord channels for comm. Why? The answer is interesting!.

• one of the first LLM training model contained irc logs from ubuntu channels

• although common crawl's data sets have come into discussion i think the carefull hidden elephant is the cloud data storage from smartphone and deskop.I believe common crawl is used as a present and future scapegoat , as a red sheet to funnel anger and complaints.

• various poisoning methods have been though already that aim to foster an adversarial env for LLM's

• An big australian data center (backed by nvidia) company's stock ipo failed last month . The owners wanted to retain their rights to sell off from day 1. So ..see how hype is a very profitable bussiness. If their ipo succeeded they could become more rich

• Very funny gen-ai satirical videos against the tech-bro-ligarchy and trump  have appeared in mainstream websites. Ironically they were created by the tech-bro-ligarchy and the question can they produced locally ? Because the subsidies are running out and soon the tockens will go up! .

@ hardinfo2 in Devuan.

I need excalibur @ backports for v:2.3  if we want when we sycn our data to have a usernote field.

---------------- hardinfo 2 ------------------
System: Devuan 6 , Xfce
CPU:i7-4790

• CPU Blowfish (Single-thread) 12.89

• CPU Blowfish (Multi-thread) 75

• CPU Blowfish (Multi-core) 50

• CPU Zlib   73

• CPU CryptoHash  32

• CPU Fibonacci  2270

• CPU N-Queens  1562

• FPU FFT    148

• FPU Raytracing (Single-thread)  2481

• SysBench CPU (Single-thread)  1128 (*1)

• SysBench CPU (Multi-thread)  7849(*1)

• SysBench Memory (Single-thread)  7115(*1)

• SysBench Memory (Multi-thread)18556(*1)

(*1) At the bottom is displayed: Alexey Kopytov's sysbench is required . But i see that sysbench is installed!
---------------------------------------------------------------------------

------------- Performance test  (license:propr) --------------
System: Devuan 6 , Xfce
CPU:i7-4790

@Benchmark/PassMark PerfomanceTest (11.0.1004) Results for Excalibur/i7-4790 CPU @ 3.60GHz (x86_64)
4 cores @ 7 MHz  |  31.3 GiB RAM
----------------------------------------------------------------------------

ps:

This post was created in Excalibur/Netsurf where there was no cursor while editing and neither toolbar but hey .. its worked without js!

Also indeed the command proposed by @kapqa in this thread previously, in order to run the propriatory benchmark utiliy PerforanceTest in Devuan releases without libnurses 5 ,worked for me in Excalibur.

sudo ln -s /usr/lib/x86_64-linux-gnu/libncurses.so.6 /usr/lib/x86_64-linux-gnu/libncurses.so.5

@ my PassMark PerfomanceTest (11.0.1004) Results for Excalibur.

So : AMD Ryzen 9 5900X  performance seems close to AMD Ryzen 7 9700X.

Also some hardinfo2 benchmarks:
AMD Ryzen 9 5900X 12-Core    24x 4951.00 MHz   

• CPU Blowfish (Single-thread) 35.57

• CPU Blowfish (Multi-thread) 469.43

• CPU Blowfish (Multi-core) 400.66

• CPU Zlib   308.51

• CPU CryptoHash  105.80

• CPU Fibonacci   9928.79

• CPU N-Queens  6261.64

• FPU FFT    662.61

• FPU Raytracing (Single-thread)  3695.20

• SysBench CPU (Single-thread)  5461.55

• SysBench CPU (Multi-thread)   68252.09

• SysBench Memory (Single-thread)  8834.38

• SysBench Memory (Multi-thread)11986.74

Isnt it an irony that the most superficial and impressive aspects of our modern life (including 'easy' 'fast' banking) are the most vulnerable to the tsunami of the overhyped bigLLMs ? I wonder if we should thank for that the tech-bro-oligarchy !

https://www.cs.vu.nl/~ast/intel/

cool @alphalpha ..

TJ-2 , TYPSET and RUNOFF , roff , Troff , TeX etc seem essentialy non-interactive text formatting programs (with increasinigly more typesetting functionalities as we move closer to the present).

All man pages i've seem seem suitable for tty. No diagrams, pics , math types. So it seems man pages currently still use text-formatting suitable for tty typed display devices. Since todays we have also display devices types with better resolution (called pixel) (*) surely it makes sense to have some sort of man page capable of containing info for both.

There is a pattern here. New display/input devices will always come along. In the context of helpsystems @ do we keep help systems along in their initial form ? Do we add carefully support for new display types ?

Let's ponder on that... Do we have TTYs ? No. We have displays of  /visual/granularity/pixel type that can ALSO mimic o TTY's  (of /visual/granularity/text  type) . BUT if you wear glasses or visual displays are not our preferred types then that means that even if we dont have TTYs we need TTY's compatible data outputs. And surely most sysadms of devuan use mostly TUI shells .

So even if the display tech (input tech seems little behind it seems :-) ) changes it seems there are ways of computing that dont die . Ways of computing suitable for diferrent ages , different computers users , a11y oriented  etc .

So framing the current state  : Manpages are currently alinged mostly  for a tty display type  and  text oriented processing type of computing workflows . One could say that that aligns better for programmers and sysadmins. (so no worries?) . But wouldnt help if you could see a nice table , diagram or datatype structure ?  Or wouldnt help IF the manpage  systems could have a certain display flexibility. Containing a data core easily presented in most display types (haptic,visual,sound) and data that can have preferable display types ? 

Helpsystems could have various traits having lightweigth ones and heavy ones .

Also likewise why not using interactive tools (an emacs mode? ) to create man pages more easily.

(email MIME @ ) .

(*)(pixel cons @) Trying to do animation on pixel level more creators come to realise that editing to pixel level is time-devouring .Personal but.. you need structure , types to allow reusability-composability. Not only you can be cutoff from others but even you can not even reuse your own work . The issue ofcourse is not the granularity of the display tech so much (that should be generic) but different ways of doing computing .

I have created Devuan wiki page titled Help to new users   : @

I present the wiki page as testing showcase of how i think a wiki could relate to various help systems used from Devuan Project.
I have transfered some ideas presented here there. Tell me if you dont like ideas trasnfered there.

In a few words a wiki (remote text-collaboration website) could be a  slow-paced user-admin workflow integrator incubator. That would position it (as a help system) in the opposite direction from a fast-paced-critical im or irc helpsystem-channel. In my test page at the top there a category tag. There a wiki user could see all user related efforts to formulate related workflows and express opinion on a user usefullness gradient and collaborate on a consensus building common wikipage.

So if you find the idea promising you could

• ask wiki admin onefang in the irc #devuan-wiki to create an account for you .

• create a personal wiki space @

• post on a  wikipage your worklow and

• link it to the same category [[!SharingFiles]].  @

All these are personal views-ideas. I am a user of the Devuan wiki.
For my general views : HelpSystemSemantics @