IMVU’s methods demonstrably worked (“I could see firsthand that they were working”), yet they still would not travel. New employees, investors, and founders of other companies could not make sense of Ries’s explanations. His diagnosis: they lacked a common language for describing the practices and concrete principles for understanding them. Without that language, a working practice remains bound to the people who embody it. Hiring cannot scale it, investors cannot evaluate it, and other people’s experience cannot improve it.
The direction of causality matters: the practice worked before the theory. Language and principles did not create the method; they made it transferable beyond its original practitioners. That distinguishes this claim from the overstated version, “there is no practice without theory.” Here, theory is a technology of transfer, not the source of the practice’s effectiveness.
Language has a second function: improving the practice itself. Ries refined his theory “in the process of being called on to defend and explain my insights” - in blog posts, talks, and arguments with skeptics (“That could never work!”). An explicit practice becomes available for reasoning, criticism, and refinement. This is thinking by putting things into words at the scale of an entire movement. An inability to articulate concepts and relationships is a reliable sign that the practice has not been worked through; the pressure to articulate them is what deepens it.
How this supports Startup success can be engineered: the right process can be learned
The claim is the load-bearing premise in the transition from “learned” to “taught” in the chain “success can be engineered through a process -> the process can be learned -> it can be taught.” Without codification, that chain ends with the person who already knows the method.
Parallels in the vault (the connections are made here):
- A Ubiquitous Language in Software Development is the DDD version of the same claim: a shared language within a bounded context removes translations, centralizes knowledge, and lets it survive turnover from one generation of team members to the next. The genesis also matches: the language arises in direct conversation with short feedback loops, just as Ries’s language was forged while defending it to skeptics.
- Different languages in software development holds that languages determine thought; the higher the level of a language, the closer it is to domain experts’ mental models.
- Characteristics of a mature process says that a mature process produces a predictable result regardless of its performer and requires documentation intelligible to everyone: transferability without relying on a particular practitioner is a sign of a mature practice.
- Team standards as infrastructure is a modern version of the claim, with AI agents in place of new employees: a senior developer’s tacit knowledge is invisible to an agent until it has been externalized into explicit artifacts; a standard “scales senior intuition.” Related notes include Knowledge priming and Agent readability.
- Following an explicit method improves the chances of success compared with acting on a myth uses the Dreyfus model: a novice can enter a practice only through explicit rules; a tacit practice gives a novice no entry point.