> If you're lucky enough to work on large software projects where there are no managers or other stakeholders asking "and how long will this take, exactly?" then maybe the design is enough.
How well can your project managers answer how long will this take? Because on my experience, 10000% delays are all but routine (ok, on construction it's usually bounded to something around 1000%). And how much value do people that can give you an estimate between 1% and 10000% of the target?
> But pretending planning ahead on large projects that is something that will just happen by osmosis
I'm telling you that planning ahead has a small value, following up with the plan has a high negative value, and the thing with a high positive value that is reexamining the plan is fought against by the practitioners.
If they are good project managers, they plan for non-routine delays and account for them in the project schedule. Before people are allowed to commit to schedule, the project managers make sure that these people have properly scoped and estimated their part of the work, by asking pointed questions, using their domain expertise and general experience to figure out if they're being bullshitted or not.
If things start to slip, they find out where and why and apply what pressure they can to get them on track. If they cannot get them on track, they then liaise with everyone downstream to make them aware of the slippage, adjust the entire schedule, and get everyone reorganised.
It's not just putting things in the calendar and forgetting about them - it's a constant, ongoing herding of cats to get them all going the same way.
It sounds like you just might not have been lucky enough to work with really great project managers!