Minor point, but that's actually an application, not a patent. It might be a patent later, but it isn't yet.

Interesting that the patent was filed 17 years after initial publication of the algorithm.

The patent is for Metaphone 3, a new version of the algorithm that was released in October 2009 and improves on the previous ones. The differences are why the Ruby gem, which already has Metaphone and Double Metaphone implemented, got a discussion going about porting Metaphone 3 from Java.

Edit: wording.

If he's the holder of that patent, "Lawrence Brooke Frank Philips", then he can enforce it no? At least in the United States.

That patent has actually been abandoned. Can't find any continuations filed on it either. So as far as I can tell, he doesn't have any patents on the algorithm/process

I thought best practice for software developers was to not look at patents?

Especially if they are working in the same field. I think so, yes.