There's a great blog entry over on the SATURN Network Blog entitled Software Architecture and "The Principle of Small Decisions" that talks about how even those small day-to-day decisions can cause havoc if you don't keep on top of them.
inattention to the small decisions can cause problems ... many decisions made at the individual level can end up having broad system impact
I'm in agreement and it's worth bearing in mind that not all day-to-day decisions will be "architectural" in nature. Yes, software architecture is about the significant decisions, but the software architecture role is needed on an ongoing basis throughout the entire project to ensure that decisions made complement the architectural vision rather than fight it. If you have enough of these small decisions waiting to be solved or solved poorly (perhaps because software development has been treated like a relay sport), you have what the SATURN Network Blog refers to as "landmines in the code".
Imagine you need to build a road from A to B, which will need to cross a number of different terrains and environments. After figuring out the overall route at a high-level (the significant decisions), you're left with the day-to-day decisions related to much smaller sections. Assuming that the overall route is correct, failure to address the day-to-day issues properly can result in a road that ultimately ends up being littered with potholes, traffic cones and debris. Sure, the road might look good from the 30,000 ft view and it might even get you from A to B, but we all know that the journey will be a bumpy ride.