From a reader: "I will NOT upgrade my D3, NOR buy the next one, even if it's a D700x or so. The LIFETIME of all these DSLR are VERY (too) short and quite impossible to resell for a reasonable price."
Therein lies the dilemma for both camera user and maker. We're headed towards an auto-like market, I think, where new model introductions are regular and predictable events, but only a fraction of the customers update each year. If, as I predict, the DSLR market has flattened and reached near peak size (~10m units a year), this implies a full penetration number of somewhere less than 50m units (possibly significantly less if more of those previously sold units were upgrades than I thought). Thus, if the total DSLR user base is 50m users and these folk update once every five years, the current sales volume will never improve. If they update every two-and-a-half years, the maximum annual sales is 20m units, roughly 2x the current volume.
But let's put that "five years" in perspective. That's roughly the time period from D100 (2002) to D300 (2007). I've long suggested that skipping a generation is the most logical update cycle for most users. For many DSLR models, that puts us at or near the five year mark, thus the 10m annual unit shipment mark wouldn't be budged unless something changes.
From a user standpoint, now that most DSLRs are capable of doing excellent work up to the maximum size of the desktop inkjets, the urge to update in the future is low. The number of users that need to print larger than 11x19" is few. The burden is thus on the camera makers. They have only a few choices to change volume in that scenario: shorten product cycles, increase product improvements each generation, or add/change things that attract new users to the DSLR world.