I am pretty much a total novice - so the bulk of my "design" would be in the tedious adjust, measure, listen, repeat process of setting up the system. The complexity of building from scratch is one of the reasons I am planning on modifying my existing speakers first - it seems like other than adding 2 pairs of binding posts, everything would be easily reversible as well.
My plan for working around the electrical transfer function/filter which as I understand it, is tailored to each driver to adjust the frequency response to get the desired performance on & off axis, will again be in software. In addition to running the crossovers via software, I plan to first get everything as close to how I want with trial and error, using free testing methods like Room EQ Wizard, then when I feel I am close I plan on using room correction software to adjust for anomalies in the speakers and speaker room interaction.
I have already treated my room, and will do more in the future, at which time I can re-calibrate - hopefully, requiring less and less correction to the signal as I make improvements.
I think the model I am aiming for is a poor man's DEQX, which I can test with the equipment I currently have and then add dedicated amps in the future, if I like the results.
One of my big concerns still is protecting the drivers - since if I open the wrong application it will send a full range signal to every driver. I know that doing everything in software - requiring specific applications, will also limit a fair amount of the flexibility of a computer based system.
Jim C