[–] Omnicis 0 points 6 points (+6|-0) ago 

"Styleometry" has its limits. especially if you are heavily relying on stackoverflow code.


[–] roznak [S] 0 points 6 points (+6|-0) ago 

AI tool, you don't need AI for this.


[–] Gargilius 0 points 3 points (+3|-0) ago 

AI tools with VR and block chains ! :-) (that’s how you sell an idea these days, didn’t you get the memo?)


[–] the_sharpest_knife 0 points 0 points (+0|-0) ago 

What do you mean? There's another way to identify a digital fingerprint?


[–] roznak [S] 0 points 5 points (+5|-0) ago 

Of course, it is called "statistics".

Most developers have a very distinct programming style. Mine tends to be:

  • Functions that do not exceed 4 parameters (because the 5th parameter cannot be saved into a register value, so you lose performance when you need to save that 5th parameter onto RAM.
  • Methods that are short, since I write code that normally fits into one screen without need of scrolling.
  • Methods where code can be used into multi threading with almost no requirements in locking.
  • Speed per formant where the compiled code will always reach the end of the method without a jump condition for the most used functionality. This avoids unnecessary jumps in the assembler output and so the CPU does not have a cache miss and needs to reload the instructions from RAM.
  • My code tends to fail gracefully, and I avoid throwing any exceptions.
  • My code barely relies on libraries.
  • Deletes in an list I always have a reverse loop to prevent accidental deleting of the next item.
  • None repetitive assembly code since I do not copy and paste code.
  • Reuse of code that I already developed on other projects.

Other clues will be in the usage of "strings" data.


[–] derram 0 points 1 points (+1|-0) ago 

https://archive.fo/HNRzR :

FYI: AI tools can unmask anonymous coders from their binary executables • The Register

'Essentially, the machine code can be decompiled back into a C-like language, and then traced back to known programmers using machine-learning algorithms. '

'They still managed 88 per cent accuracy in identifying authors. '

'The technique can help identify virus makers as well as unmask the creators of anti-censorship tools and other outlawed programs. '

'Programmers can be potentially identified from the low-level machine-code instructions in their software executables by AI-powered tools. '

' "Randomized pseudonyms with randomized code and malware (unless a malware author is planning to share only one code sample with the entire world), as well as the above methods is a starting step." ®'

This has been an automated message.


[–] redog 0 points 0 points (+0|-0) ago 

Cool, so can we use this to figure out who the real satoshi nakamoto is?


[–] ceopsm 2 points -1 points (+1|-2) ago 

Thats great