Long term plans often involve decisions that look stupid over shorter time horizons but if you know what the tech tree looks like you can justify moves that are suboptimal in the short term but will payoff in the long term.
(I am imagining but) In each of the examples of long-term decisions, someone said “what do we expect to be able to do in the future? If we were in that future, what work would we have done in the past to enable this future regardless of how ridiculous it seemed at the time?” The danger is in premature optimization, which is what makes the precision so brutally important.
More precise descriptions of technology could enable faster progress by enabling you to Make precise predictions. This precision is important because history is littered with predictions that were roughly correct but either overshot or undershot the target. “Put AI on all the things!” Is a recent example.
In order to act on the structure of the tech tree, the structure of the tree needs to be legible (Legibility enables action.) This important legibility is why The tech tree model suggests that explicitly drawing out tech trees is important.