But it fails if you are talking about any changes greater than 1 order of magnitude. For instance, from the article: "we're still about one order of magnitude from making a measurable dent in the search market and two from a major one". The terminology of "orders of magnitude" lets you talk in terms of the logarithm of the value rather than the value itself, and for certain kinds of thinking this is useful. The terminology "10x" does not help do this (or at least I have never seen it used that way).
This is exactly the same "trick" as plotting graphs on a logarithmic scale - it gives you a different point of view on the data which sometimes (but certainly not always) reveals useful insights.
I wonder if there are interesting scaling/growth/startup point of view changes that are equivalent to other data visualization transformations? Is there a growth-data version of changing from cartesian to polar coordinates perhaps?