"Big picture" design one release cycle ahead?

[Thomas Pfeiffer]

Hey folks,
I have a suggestion:
I fully support Aaron's idea of small concentrated effort sprints with 
integrated design and development, which comes pretty close to agile development.
However, product managers in the industry have found that such short design 
cycles do not work well for all kinds of design. They work well for small 
features/improvements, but for introducing larger changes or completely new 
concepts, longer research/strategy/design phases work better.
One of these big changes for me is fleshing out the general concept of 
task-driven design for PA. Since the big picture won't be implemented in PA4 
anyway (we already got our hands full with small improvements/polish), I think 
it would make sense to use this release cycle to work design-wise on the big 
picture of a task-driven system parallel to designing the detailed improvements 
for PA4 in each focus sprint.
The big disadvantage is of course that design capacity is split between the two, 
but i think this is pretty much unavoidable since designing for a new metaphor 
requires more then just a few weeks and we probably don't want to have a full 
release without any actual development. In case of insufficient design capacity 
for both, the focus sprints would always have higher priority, of course.
The biggest advantage to me is that ideally we have a broad concept for 
task-driven design ready by the end of the PA4 cycle and can implement the 
details for PA5 if we wish to do so.
So what do you guys think? Is the path to task-driven design worth putting some 
effort into it ahead of time?

Cheers,
Thomas