I don't think it's quite ignorance of the implementation domain is the key here, but rather outside-in vs inside-out thinking. Nearly all engineers solve problems from the inside in my experience.
It's not just that though. Something may be impossible to do in the application, except with knowledge of x, but trivial for someone who knows x. If you know x you might totally forget that without this knowledge the task is basically impossible for the user.
So if building the application engenders learning x, then you might totally miss this design problem, unless you step away long enough to forget x again yourself - and that may never happen.
It is a bit too much to ask for one person to learn all of the details and totally own the implementation space, and yet be able to instantly forget all those details and go back to when they didn't know anything, to look at things from the user's perspective.
