I'm not sure you can label obfuscation of an interpreted language as pointless (I'm unable to add a comment to Schwern's post, so here goes a new entry).
I think it's a little shortsighted to assume you know all the possible scenarios where someone would like to obfuscate code, and you assume that anyone will actually be willing to go to whatever necessary lengths to view that code once obfuscated. Consider my current scenario:
I work for a consulting company that is developing a large and fairly sophisticated PHP-based site. The project will be hosted on a client's server that is hosting other sites developed by other consultancies. Technically any code we write is owned by the client, so we can't license it. However, any other consultancy (competitor) with access to the server can copy our code without getting permission from the client first. We therefore have a genuine reason for obfuscation - to make the effort required for a competitor to understand our code more than the effort of creating a copy of our work from scratch.
