This, in particular is worthy of being burned in the memories of many people:
First, we can somewhat mitigate RMA uncertainty by means of extensive, independent, and competitive experimentation, field trials, and exercises - both single service and joint. We also can seek to sharpen the debate among competing RMA schools, service visions, and branch perspectives - while insulating these from partisan politics and commercial interests.

Second, despite our best efforts, a substantial degree of uncertainty will persist; the only decisive test of vision is war. This humbling fact argues for avoiding over-commitment during periods of great strategic uncertainty, retaining flexibility, and developing our facilities for rapid adaptation. Adaptation is best served by substantial equipment and unit prototyping, which would offer multiple potential paths of development. Also useful is modularization of capabilities, units, and training regimes - an approach that allows rapid change through "add-ons" and supplemental training. (emphasis added, both / kw)

Third, pervasive uncertainty tends to strengthen the position of the status quo, especially during periods of significant strategic change. For institutional reasons, the default position may be to extend the status quo into the future. However, this is not a neutral position. When the world is changing rapidly the preservation (or recapitalization) of the status quo involves a "future vision" as risky and open to question as any - if not more so.

Finally, "uncertainty" by itself does not constitute a strong rationale for either sitting still or moving decisively down a new path. It lends positive support only to efforts to reduce uncertainty or improve our capacity to react, recover, and respond to surprise (that is, to adapt).
All simple basic common sense -- and all far too often ignored totally due to egos and agendas.

That I underlined is harsh reality, that I placed in bold type is the only way to preclude not being prepared for that reality; one or the other will not work -- we have to be able to do either/or. Or even 'whatever'...