There are plenty solu­tions how to select a ran­dom line from a file. A way to select a ran­dom para­graph was a bit more dif­fi­cult to find, but in the end it was enough to change one com­mand from the always help­ful sed one­lin­ers. Instead of sort­ing para­graphs with sed '/./{H;d;};x;s/\n/={NL}=/g' file | sort | sed '1s/={NL}=//;s/={NL}=/\n/g', we select a ran­dom para­graph:

sed '/./{H;d;};x;s/\n/={NL}=/g' file | shuf -n 1 | sed '1s/={NL}=//;s/={NL}=/\n/g'

Para­graphs are sep­a­rated by blank lines.