I think this gives "people" too much credit.
In my experience when people say they're looking to grow by an order of magnitude, they don't know that has a very specific meaning and they're just talking about getting bigger by some vague undefined number that could be anything from about two times on up.
An important upshot of the article is really the idea that you should think logarithmically about scaling. It's a scale-independent way to think. It's an easy way to get a website that gets 10 hits/day on the same plot as one that gets 10^9 hits/day.
An example is here: http://measuredmass.wordpress.com/2012/10/20/more-hn-numbers...
I don't even really blame them, because I'm sure they simply hear the phrase a few times from technical people without explicit explanation of meaning and then adopt it themselves to mean "any big growth" which is a fairly reasonable assumption contextually, prior to someone pointing out to them that it means something more specific.
FWIW, in my CS education (both at school and self-directed via books, system manuals and such) I was taught to worry about whatever base is relevant to my current data and level of abstraction to the machine. 2, 10, 16 and even 8 were the usual culprits, but certainly nobody tried to teach me to forget about the ones other than 2. I worry about base 2 far less than base 10 in my day to day code despite the fact that I'm old and know assembly (6502, 68k, x86, MIPS, ARM) and still do a lot of bit-twiddling.
As Gabe points out, decimal orders of magnitude are good sign posts because they often separate the space into meaningful chunks, what ever it is.
Think about your own earning potential, the differences between earning $100/month, $1000/month, $10,000/month and $100,000 a month. Those are very different places to be.
The argument given was that you should plan infrastructure expansion with orders of magnitude. For example, increasing disk space from 10tb to 100tb. Or replacing a network switch with one that has 10x traffic capacity. The reasoning goes that the time spent engineering the expansion is a better investment at 10x, rather than having to increase infrastructure incrementally.
Considering problems/situations from an orders of magnitude point of view is an under appreciated approach IMO.
Of course the orders of magnitude argument get abused by marketing people (but what doesn't?). That aside, they're extremely valuable for personal or team understanding of the problems and goals.
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?
