[torton]

Some projects, where the constraints are hard (immovable deadlines, inflexible scope, etc.) can still benefit from detailed estimation as far as possible. This way any problems or misalignments have a chance to come out early.

Most stuff I do that is long term isn't that critical and gets broken into reasonable size phases; the closest one is planned in detail, the next one has no major open questions, and the rest have a brief summary of what will be accomplished / what is the goal of that phase only.

That gets rid of a lot of the lack of flexibility of waterfall, and it does happen that priorities change a lot and many projects don't get to the latter phases (often, by definition/priority, the less "immediate fire" ones).