Agree with the plainest-of-plain-text approach. For me, the more structured the note system, the less it is used. I have a simple script called `n` that operates on a `notes.txt` on my NAS. If given arguments, it appends a timestamp and the arguments (verbatim) to the notes file. This serves as a super quick way to jot something down or log an action (`n determined Krylon adheres best for project X`, `n timestamp placeholder for server X reboot`). Alternatively, calling it with no arguments appends a timestamp and opens vim in insert mode at the bottom of the file. Finally, I have an alias `nt` to open `less` at the bottom of the notes file, for quick review (`alias nt=less +G ~/share/notes.txt`).

The main script is:

  ~$ cat share/bin/n
  #! /usr/bin/env bash
  cd $HOME/share # or wherever you want to put your notes
  echo >> notes.txt
  echo -n '# ' >> notes.txt
  date '+%a %d %b %Y %T %Z' | tr -d "\n" >> notes.txt
  echo ' #' >> notes.txt
  if [[ $# -ne 0 ]]; then
    echo $@ >> notes.txt
    tail notes.txt
  else
    vim "+normal Go" +startinsert notes.txt # or nano: nano -R +-1 notes.txt
  fi

This script is accessible from all my personal machines as well as appropriate work machines. It works well enough. I've devised a small system of tags and keywords, but that part is not particularly interesting. I just put the keywords at the beginning of a line and search for them with /^thekeyword/.

More structured notes go into dokuwiki or just a LibreOffice file in the project folder.

All this is almost guaranteed to be readable in 20 years. The LibreOffice files may be at more risk, but at least there is more than one tool available to read them.