Brandon, I'm really not trying to beat that dead horse here . . . well really I am . . . but I will make one last attempt to point out why your math is wrong. This is what you state in the article:
9.1 million auto sold times 50% penetration rate = 4,550,000 new installations.
4,550,000 new installations times 50% [take rate] <- my edit (currently running below that) = 2,275,000 new oem subscribers.
21,000,000 subscribers times 1.7% monthly churn = 4,284,000 deactivated subscribers. <--- We've pointed out ad naseum your error here r.e. using self-paying subs to calculate churn vs. gross subs, using EOY gross vs average monthly gross, etc. Next, you already included OEM losses in the above when you calculated the 2,275,000 number. You are in essence double counting the OEM churn in your calculation when you calculate churn from the gross after already applying the take rate!
4,284,000 - 2,275,000 = 2,009,000 subscribers to be lost in 2009
Here's what you should have said using your own thought logic:
OEM gross adds = 9,100,000 x 50% = 4,550,000
OEM deactivation = 4,550,000 x 50% = 2,275,000 (these are trial sub deacts, not self-pay deacts)
15,000,000 (average monthly self-pay subs starting at 19 mil gross subs after Q4 and EOY 17 mil assuming 2 mil net decrease) x 1.7% churn = 3,060,000 deactivated self-pay subs.
4,550,000 - 2,275,000 - 3,060,000 = 785,000 subs to be lost in 2009
Your number is off by at least 1.224 million not even getting into the issue of retail subs. That is fact, not opinion. Note that 21 million never enters into the math (I'm assuming you got this number from 19 mil current + 3 mil average gross OEM adds for the year).
You could have posted the article stating "700,000 net lost subscriptions" and had the same effect while having a leg to stand on.
Lastly, retail subs are NOT negligible. Since you've already included the retail loss in the churn (which is about 50% of churn), every new retail sale is essentially a NET SELF-PAY ADD! Gross retail sales have historically been 1,000,000+ (rough low estimate here). The number may come down given the current economic conditions, but hardly something you should dismiss as neglgible . . . how much do you think retail will decrease? Auto sales are considered a disaster at 40% down. I find it hard to think SATRAD sales will decline more than auto (or any other industry). There's considerable merit to the argument that cheap entertainment services will actually increase sales in a poor economic environment. Look at Netflix for example. Assuming a drastic 50% decline you are calling some 500,000+ net self-pay adds negligible.