Bob:
You give a great summary of the software development cycle of business software. This methodology is one of the problems with software development in America today and one of the reasons Indians are getting a lot of contracts that American contractors are not getting. The Indians have really streamlined the process...and we need to take note. The much lower wages paid to Indians doesn't help us either...but we digress...
I work in an embedded software shop and have gotten my hands dirty with development of a couple of custom moving map GPS devices for large corportations. I can assure you that we don't operate using this methodology -- we would NEVER finish anything if we did. The life-cycle you mention is generally used for consulting firms where EVERYTHING must be in writing. If Garmin is developing their software in-house, the developers have immediate access to the design guys and can ask them any question they need. There isn't this back and forth stuff going on -- it's a get-r-done philosophy.
Our company's current contract is a 5-year deal with SAP and I assure you that we don't use the methodology you mention...and you don't get much more methodical than SAP.
At any rate, if it costs that much for Garmin to write some GPS software, they seriously need to do some housekeeping among management. Remember that most of the software for the 396 was already done for the 296. The new stuff is XM, TIS, SL-30/40 connectivity. Much of the XM is the same as the 376C (XM audio, Nextrad, local weather, etc).
If I was in charge of the 396/376C products, I would make the hardware identical except for perhaps a flag in EEPROM that would tell the 376C not to run the aviation code. This prevents someone from installing the 396 code on the 376C but lowers production costs. This sort of thing is done quite often. This is why I said software is probably the only difference.
I stand by my original assertion that the higher cost is most directly related to market demands. XM weather in a GA aircraft is far more valuable than XM weather to a recreational boater. There are higher software costs in the 396, but not $1600 worth. That's all I'm saying.
I think some of the earlier posts did a great job of outlining some additional cost areas, such as the higher cost for Jeppeson licensing fees, possibly higher insurance rates, a more rigid QA process, etc.
I have nothing against the 396 (or the price). It's a fantastic device and very 'polished', as we say. In my opinion it blows away the competition (anywheremap) at a comparable price point. It will cost you $5000 for the GDL 69 to just add XM WX to a 430...so the 396 is quite a good value.
Now back to earning my $2600 for the 396.
All the best,