Hi all,
There’s a lot going on in this thread and I simply can’t touch on every train of thought, but I wanted to pop in and say a few quick words. I’ve been a technical designer for Blizzard for almost eight years now, six of those on the Heroes team. I’ve had a hand in implementing probably half of the Heroes in the game and have been the primary designer for about 10 reworks now.
The biggest time “expenditure” or “bottleneck” in the design process is not the implementation, but the iteration. Assuming I had an entire design ready on paper (four basic abilities, two heroics, 21+ talents, etc) and was only focused on making the hero playable ASAP, I could probably bang out a hero using existing assets in a week or two. That’s just to have a controllable character with placeholder visuals that can interact with their world and be mostly bug free. The problem is, we don’t just cobble together our initial thoughts and call it a day! We try an idea out, tweak it a bunch of different ways, scrap it, try some new ideas, tweak those, scrap them again, try something new and so on and so on.
This iteration process can take weeks or months, depending on the situation. Often we have to throw out multiple days worth of data work because an idea didn’t pan out like we thought it would or we found an alternative idea that was better. This is not a failure of a designer’s capabilities or aptitudes, just the nature of the beast. Even after this iteration process is done, there’s still tons of work needed to make sure all our new systems are bug-free and mesh well with every other existing system (there are hundreds at this point!). There’s also optimization, data cleanup, and refactoring involved, all of which take expertise and time.
And that’s just touching on the design process! There’s still artists, animators, sound engineers and a whole host of people whose work requires a designer’s time to implement/modify.
To draw a rough metaphor, looking at a final design is like looking at an iceberg on the open ocean; you’re only seeing 5% of the total structure.
Hopefully this sheds a bit of light on a largely invisible process 
Cheers,
David