The least initial deviation from the truth is multiplied later a thousandfold.

Small errors at the start do not stay small. A slightly wrong assumption gets built on, and every subsequent decision inherits it, so by the time trouble shows up it is expensive and hard to trace back to a difference nobody thought worth arguing about. Direction matters more than speed at the beginning of anything.

Which is a case for spending disproportionate care on the first decisions. Before a project, a hire, a definition, ask whether the premise is exactly right rather than close enough. Corrections are cheap now and will never be this cheap again.