If we want to produce machine readable texts:

• we have a reasonable notation, XML
• we have a character encoding system, Unicode
• we can use whatever elements we like in XML
• we know how to write XML processing applications

• the ability to interchange texts and tools with others
• a formal way of expressing constraints about our markup
• (preferably) some way of documenting our markup vocabulary
• (possibly) some way of expressing the meaning of our vocabulary in some formally processable way

Some, but not all, of this is the task of a DTD or schema