I’ve ended up in the position of explicitly designing these organizations in many contexts (government vs private, secrecy, etc). I’ve also had the opportunity to test some ideas to understand the underlying dynamics in practice. The above collection of observations, while directionally correct in my experience, overlooks or undersells some critical relationships when making effective disruptive research organizations a repeatable engineered result. It would be easy to overfit to the identified characteristics and end up with a mediocre research organization.
Many organizations fail to understand what the role of a visionary leader actually is and the characteristics of an effective one in this context. First, these people are always executing their own vision, not the organization’s, and you fill these roles by finding a leader whose personal vision is already aligned with where the organization wants to go. The organization provides a platform for that person’s vision, it doesn’t impose it. The common practice of having a charismatic person execute someone else’s vision leads to poor navigation of the ambiguity inherent in these visions. Second, they must be viewed as true technical peers by the excellent people they are leading or there will be a lack of trust in that vision. Ideally, you want everyone in the research organization to harbor the suspicion that the visionary leader could effectively do their job if required. Too many organizations try to substitute this with someone that is senior, pedigreed, and credentialed but lacks technical gravitas. Finding appropriate leaders is the most difficult part of building these organizations.
Every person on the team should be selected for a unique, critical expertise that does not overlap with any other person. This naturally creates clear ownership of something important, encourages the team to self-organize organically around their individual strengths, and tends to engender mutual recognition of the value other members of the team. This simple practice mitigates a large amount of adverse social dynamics and politicking that destroys the productivity of teams in practice. Excellent researchers bring their own vision of their expertise and having two people responsible for the same thing leads to conflicts of vision (“ought”) that hinder execution and add little value. Concomitant with that, all members of the team (including leaders) have a requirement to be able to clearly articulate and defend ideas and decisions to anyone in the org that asks. There is an element of psychological safety that comes naturally from being the person best equipped to address questions and criticism.
As a general observation, there also needs to be a forcing function toward achieving a concrete objective in a resource constrained way. Without this, there is little incentive to prune the search space.