Skunk Works: Who Gets to Say No?

When Skunk Works was short of orders, Ben Rich was offered a paid study for a hypersonic passenger aircraft. He declined. In his recollection, the Reagan administration’s proposal required solving problems for which he could see no workable technical solution. Taking the research money was possible; honestly promising the desired result was not.

That is a more useful starting point for discussing autonomous teams than a call to gather talented people and leave them alone. In Skunk Works, by Ben R. Rich and Leo Janos, engineering freedom begins with a concrete decision: which project a team agrees to take on at all.

Autonomy before the work begins

At the start of “The Two-Billion-Dollar Bomber”, Rich writes that he continued Kelly Johnson’s rules and turned down proposals he considered misguided. He explains this through reputation: do not promise customers something you do not believe in, and do not conceal your own mistakes.

The important detail is that work was scarce. A busy team might choose among orders because it lacks time. In Rich’s account, something else mattered: disagreement with the project outweighed an opportunity for funding. Autonomy concerned whether to accept an assignment, as well as how to deliver it.

This is a manager’s account of his own decisions. It does not establish that all his judgements were correct. His doubts about one proposal, in particular, are not a present-day assessment of hypersonic technology as a whole. The episode does, however, pose a precise organisational question: can a team say no when money is already on offer?

Permission to fail needs a boundary

In “Drawing the Right Conclusions”, Rich describes the model more fully. It needs a difficult but specific goal, freedom to try unusual solutions and permission to fail. He also insists on low overheads, limited investment and a small series of prototypes. Failure must remain an acceptable research risk for the company.

“Allow mistakes” therefore does not mean “fund every idea indefinitely”. Management gives up some control over execution while limiting the size of the bet. A prototype should help evaluate a technology and identify its applications before large spending on production.

These are distinct decisions easily hidden behind the word “trust”: which problem to take on, how much an attempt may cost and what its result must demonstrate. If those remain vague, a small team solves nothing by itself.

The team does not start in an empty garage

There is another condition that romantic retellings easily miss. Rich explicitly discusses the parent company’s resources: premises, tools and workers can be provided for a project and returned when it ends. Independence in his model does not mean self-sufficiency.

That changes the meaning of “create your own Skunk Works”. He describes work organised separately within a larger company capable of supplying resources and absorbing a failed attempt. Copying team size alone leaves out what makes the experiment possible.

Applied to an engineering team in Ukraine, the idea can start with three questions: who may reject a project, who sets the spending limit and where do resources come from? These questions examine the distribution of authority. They are not a ready-made defence procurement model.

For this particular problem, read chapters 15 and 16: first the refusal, then the conditions for organising an experiment. The memoir is interesting here because it shows how many decisions must be entrusted to a team before independence becomes more than a name. It does not establish that small teams always win.

Source

Ben R. Rich and Leo Janos, Skunk Works: A Personal Memoir of My Years at Lockheed. Author’s copyright 1994; ebook edition used, 2012. Chapters 15, “The Two-Billion-Dollar Bomber”, and 16, “Drawing the Right Conclusions”.