If the premise is “A roadmap is a promise”, I agree completely with your article.
But I’d challenge that by saying that “A roadmap is a forecast”. This completely reverses the sentiment.
A bad estimate is better than no estimate. Estimates are built to change.
Humans suck at estimating things, but the magic sauce comes from the remarkable consistency of that suckiness. We usually end up failing our estimations in very similar ways on average and when you get comfortable with planning in a way that acknowledges that huge margin of error, you end up significantly improving on the quality of output.
So… I’d recommend to not be fearful of roadmaps. Be fearful of the level of assumed commitment you put on those roadmaps.