Ceres is really nothing more than an interpreter capable of transforming plain text that has been marked up using a specific set of tags into an arbitrary format.
The tagging syntax that we use is somewhat similar to Wiki syntax, but much simpler. In fact, our authors really only need to learn a few very simple formatting rules, which will take care of most of their needs.
While new authors do have to learn a new syntax, the latter is much simpler than HTML or XML, and allows for much faster typing—once they've learned how Ceres works, they are be able to focus much better on your writing, as opposed to having to worry about whether their text will be formatted properly.
In addition, Ceres helps the editorial process by minimizing the need for intervention on the content in the layout phase. Because the process that transforms the author's input into the final product—be it HTML code, a book or a laid-out article for the print edition of php|a—is almost entirely automated, the chances of errors being introduced outside of editorial control are minimal.
Also, Ceres is based on a strict LALR(n) parser, which means that—much like an XML parser—it will reject any content that is not syntactically valid. This will help us ensure that the resulting documents are properly formatted, rather than trying to automatically close open tags.