Originally Posted by pepak
If you want "one encoding to rule them all", go with UTF-8. Personally, I use the encoding that is best suited to each book in my OS (e.g. us-ascii for english books, windows-1250 for czech books). It helps with your 'readability of source" - it is far easier to read "â" that a sequence of two special symbols.
I'm moving to using UTF-8 entirely. But then, I have a good UTF-8 text editor - BBEdit. This means I have access to all of unicode, and see the characters as they should be, and don't have to use the entities. If any particular reader software needs entities rather than UTF-8, the production process can do the substitution. I much prefer editing with “ and ” rather than “ and ”