Read this today, and it’s something that I’ve known for a long time, but have never really articulated.
Kanban smooths out the aggregate, but the individual work request may still vary widely.
– Corey Ladas, here.
The entire thread (2 posts currently) is worth a read, but the main point does boil down to that quote… There is nothing that eliminates variance entirely in software development, and the cone of uncertainty is always a reality. What the methodologies with smaller batch sizes accomplishes is to reduce the overall impact of any one piece hitting the high end of its variance.