I read it in art school (swiped from my then-girlfriend's bookshelf), and it was nice to have a connection between my school of art weirdos and putting myself through the same school by teaching myself software engineering. I distinctly remembering writing a Perl script to highlight that the Mu puzzle is unsolvable.

    #!/usr/bin/perl -w
    use strict; 


    my %tried = (); 

    main(); 

    sub main { 

        my @rules = (
                     \&rule_one, 
                     \&rule_two,
                     \&rule_three, 
                     \&rule_four,
                    ); 

        my $axiom = 'MI';
    
        my @path; 
        my @tries; 

         solve($axiom, \@rules); 
    }


    sub solve { 

        my @app; 


        my ($axiom, $rules) = @_; 
        my $i = 0; 

        foreach(@$rules){ 
            my $tmp =  $_->($axiom);
            push(@app, $tmp) if $tmp;
        }

        foreach(@app){ 

            next if $tried{$_};
            next unless (length($_) < 1000); 
    
            print $_ . "\n";
    
            $tried{$_} = 1; 
    
            if($_ eq 'MU'){ 
                print "YES!\n\n\n"; 
                exit;
            }
            solve($_, $rules); 
        }

    }



    sub rule_one { 
        # If any string ends in I, you can append U

        my $str = shift; 

        if ($str =~ m/I$/){ 
            return $str . 'U';     
        }else{ 
            return undef; 
        }
    }




    sub rule_two { 
        # If any string begins with M, 
        # you can duplicate the string after M,    

        my $str = shift; 

        if($str =~ /^M/){ 
            return 'M' . substr($str, 1) . substr($str, 1); 
        }else{ 
            return undef;
        }
    }




    sub rule_three { 
        # If any string contains III, 
        # you can replace the III with U

        my $str = shift; 
        if($str =~ /III/){ 
            $str =~ s/III/U/g;
            return $str; 
        }else{ 
            return undef;     
        }
    }




    sub rule_four { 
        #If any string contains UU, 
        # you can delete the UU

        my $str = shift; 

        if($str =~ /UU/){ 
            $str =~ s/UU//g; 
            return $str;
        }else{ 
            return undef; 
        }
    }

I always find the Gödel representations using powers of primes too strange. It's technically correct, I can follow the proof, but there is a small corner of my mind that can't believe it.

The GEB book doesn't have all the details, but it uses the ASCII representations (actually base 20 but whatever). With the ASCII representation the theorem feels obvious.

There is a section in the book where he laments that certain mediums (like books, and movies) leak information about where the story ends (because in a book as you are reading it you will know you are coming towards the end). He explained one way to avoid this and retain some mystery in story-telling would be to finish the story some distance before the end and pad the remainder of the book with blank pages. However blank pages are too easy to spot if someone happens to flick through the book before hand, so instead it would be better to pad the book with text which appears to be related to the primary story and therefore is undetectable unless someone reads through the whole book in order.

C'mon, man. It's an analysis.