It's suspicious that it's a graph of cumulative downloads. Cumulative graphs always go up nicely. But what matters is downloads per month. Someone with good monthly numbers will rarely resort to showing cumulative ones.
It's particularly suspicious the way the number for December, which we're only 6 days into, lines up exactly where it would if they were seeing hyperlinear growth. You'd have to time your post almost to the day to make it look so neat.
I would be very interested to see an ordinary graph (i.e. without graphic tricks to make it steep) of downloads per month. That would show us what's really happening.
I wouldn't be surprised if what that graph showed was linear growth (which is bad, because it means a decreasing growth rate), followed by a massive spike in December due to a special promotion.
I agree with you, but Apple does the same thing. They add not only the downloads since 2008 when they had the App Store for the first time, but also the money they've given since then to developers, as if that would be relevant anymore. What's relevant for developers now is how much they've achieved in the last few months at most.
It looks like the December bar (yellow) is actually for November if you take each bar as a month (which lines up with the rest of the marked points). The labeling on the graph is wrong (the yellow bar should be slightly below 10 billion and labeled as November). I'd like to see the downloads per month (and especially see the paid downloads per month), but as a marketing tool, I'd bet that 10 billion downloads sounds better than whatever their current downloads a day rate is.
"This past weekend, thanks to Android users around the world, Android Market exceeded 10 billion app downloads - with a growth rate of one billion app downloads per month"
I think it's pretty obvious that the graph is cumulative. They do mention the rate is accelerating, and a cursory glance at the graph tells that the delta between months appears to be increasing as well. I agree, though, that complementing it with a non-cumulative, per-month graph would be informative and interesting.
If you grow from 10 to 15, 20, 25 etc., each time, even though you are adding 5, it is a smaller percentage of what you had. In this example, you would be adding 50% to go from 10 to 15, then 33% to go from 15 to 20, then 25% etc. Growth is linear, but growth rate is decreasing.
Is that common terminology? It sounds like you're defining "growth rate" as "exponential growth rate." To me, if you're seeing 10, 15, 20, 25, etc. every month, then your monthly growth rate is a constant 5. You're still growing, it's just that your growth every month isn't a function of your size last month.
... whereas if your downloads per time month is a constant proportion of your total downloads to date, e.g., next month the downloads will total 2.5% of the total downloads to date ... and that keeps happening every month, then:
* the per-month downloads will grow exponentially as you graph it over time
* the cumulate number of downloads to any given date also will grow exponentially
It's suspicious that it's a graph of cumulative downloads. Cumulative graphs always go up nicely. But what matters is downloads per month. Someone with good monthly numbers will rarely resort to showing cumulative ones.
It's particularly suspicious the way the number for December, which we're only 6 days into, lines up exactly where it would if they were seeing hyperlinear growth. You'd have to time your post almost to the day to make it look so neat.
I would be very interested to see an ordinary graph (i.e. without graphic tricks to make it steep) of downloads per month. That would show us what's really happening.
I wouldn't be surprised if what that graph showed was linear growth (which is bad, because it means a decreasing growth rate), followed by a massive spike in December due to a special promotion.