Note what I said earlier about the start-up costs of model lines--they probably aren't significant if you have the technology base and decent sales. And if tech products have taught us anything, it's that you must cannibalize your own products with still more innovative products or risk losing market share. If you don't push forward, like Apple is now struggling to, there's always a Samsung ready to take the lead.
Sure, but it's bad when your consumer models cannibalize your pro models to the brink of irrelevance because you haven't updated your pro crop body in well over four years. Products should predominantly cannibalize sales of previous models in the same general line, and possibly previous models in higher lines, but if they cannibalize current models in higher lines, you pretty much have two choices: rev the higher line very quickly or put the handful of extra features into the lower line and drop the higher line entirely. Then, optionally introduce a lower line.
There really is no good reason to have two consumer crop body lines that are barely differentiated, much less two professional full-frame body lines that each have some features and not others.
Really? Have you been to a grocery store lately? 11 flavors of Cheerios, 25 types of Head and Shoulders, 74 flavors of Campbell's soup... with more than a dozen versions of vegetable soup. Crest and Colgate produce 52 varieties of toothpaste.
The big difference is economies of scale. It doesn't take much R&D to release a new cereal flavor, at least when compared with designing a new camera. And the sales are many orders of magnitude higher for the cereal, so it covers the R&D costs very quickly. In the electronics industry, where the R&D costs are high and the sales volume is low, that approach leads to John Sculley's Apple, where there were hundreds of products differentiated only in tiny ways, leading to consumer confusion, poor sales, and massive financial bleeding. No surprise that the person who screwed up Apple so badly came from a company (PepsiCo) that made just the sorts of products you're talking about.
One of the best things Steve Jobs did when he came back to Apple was slash the number of models. Repeatedly. By the time he passed away, there was one size of MacBook, two (maybe three) sizes of MacBook Pro, the Mac Mini, the Mac Pro, and the iMac. One consumer laptop, a couple of pro laptops, one consumer desktop, one pro desktop, and one all-in-one desktop. There were several configurations that let you change various options (hard drive and RAM sizes, CPU speed, etc.), but as far as actual designs go, there were only a very few.
Also, the high-end model should always be a strict superset of the low-end model. Having features in the 6D and 70D that the 7D, 5D, and 1D don't have is absurd.
I think it would be absurd to wait 8 years for a product to get a feature available today: 5-7 years for the next couple top end bodies to acquire it, and then a couple more for the latest amateur version. Canon, to it's credit, has been willing to try new features on lower-end cameras to see if they drive sales, then push them upward if the feature is successful. Like eye-controlled focus in the 90's, or internal WiFi and GPS today. If you look at the Digital EOS SLR Timeline, you'll see that the 1Ds line has never introduced an image processor; instead, they are introduced below and the software perfected before being moved up. Besides, a pro can always add on the latest external GPS unit or Wireless File Transmitter, with a better quality feature likely resulting (along with possible future upgrades).
The "introduce in low-end hardware and let it bubble up" excuse just doesn't match reality. Well, it does, but only over nearly geologic time scales. Canon first tried GPS on consumer bodies back in 2011, and they were incredibly late to the party even then, having been preceded by dozens of other cameras dating back as far as 2007. The fact that it still isn't built-in across the board in their pro line in 2014—some seven years after you could get cheap point-and-shoot cameras with built-in GPS—means that Canon's rate of DSLR upgrades can only be described as "glacial".
And part of the reason for that slow upgrade speed is that they have to cover their R&D costs before they introduce a new model. The more models they sell (to what is mostly a fixed-size market, give or take), the longer each model has to be on the market before they can upgrade it again. Right now, there are three pro bodies by my count—the 1D-X, the 1D-C, and the 5D Mark III. That's an insane amount of R&D for devices that don't sell very many units. And you can bet the cost difference in the hardware among those three is orders of magnitude less than the price difference. So Canon could drastically reduce their R&D costs by folding them into one. Or, if you'd rather have two different body sizes, use the same guts in different enclosures, and update them both at the same time, every time.
Actually, I'd go so far as to say that Canon's biggest problem is that they're afraid to let their low-end products cannibalize their high-end gear, so they artificially limit the features of their low-end products. If they stopped doing that, they would lose sales of their high-end gear, because there would be less reason for people to spend twelve grand on a 1D-C, but on the other hand, the only reason those cost twelve grand is that so few people buy them, so they're distributing R&D costs over a tiny niche market. If they had only one unified 1D/5D model, they could almost certainly make just as much money selling it at less than the 5D's price, and still significantly increase their rate of updates.