# Institutions either make things more legible or act as trust brokers [[Michael Nielsen characterizes institution building as making previously illegible things legible]]. This is then in tension with the fact that [[Many knowledge generating activities are inherently illegbile]]. How do we resolve this tension? What Michael describes as “inherent illegibility” seems to be high [[Knightian Uncertainty]] about the *purpose* of an activity in [[Activity Space]]. (Using the term *purpose* instead of *goal* because the action itself is often a complex system). New institutions are trying to make <thing X> legible to <entity Y> ([[Legibility is not an inherent property of a thing — it always involves a subject and an object]]. e.g. Making potential actions legible to researchers; making projects that would be impactful but poor startups legible to funders; making talented people legible to the rest of the world; making researchers in different fields legible to each other; etc. Is it possible that the entire process of creating new knowledge is a process of making things more legible? In the sense that if I know a thing and die, or create a thing and it’s burned, knowledge is not accumulated. Unless it remains entirely tacit, passed on through apprenticeship, knowledge must become legible for people to build on it. On the flipside, trying to make something legible to too many people too soon can lead to [[Legibility poisoning]]. Building an institution is about enabling actions. If [[The two ways to enable action is through legibility and trust]] then one of the two strategies an institution can employ is to (as [[Michael Nielsen]] pointed out) create legibility where none existed before. The other strategy is to act as a “trust broker” — people outside the institution trust the institution to place trust correctly. Examples of this strategy include early VC firms, and [[JCR Licklider]]-era DARPA. Perhaps these institutions also created legibility around the fact that there was a set of actions that required a different environment/inputs/expectations to flourish. “Hey, this black box exists. I understand this black box. Trust me to bring the results forth from the black box.” In a way you could look at this through the lens of how [[Systems depend on where you draw the boxes]] — the legible pieces of the process are the boxes you can see and their connections and you trust the boxes that are closed. DARPA makes the quality of the program manager and the bull case for a program legible to people outside the organization but then black boxes/trust brokers the rest of the program. [[Opacity is important to DARPA’s outlier success]] Most institutions do a mix of trust brokering and legibility creation. However, especially for new institutions in the low-trust 21st century, trust needs to be earned so you need to create legible results first before an institution can do trust brokering. Let’s circle back to the tension that Michael called out between the fact that there is inherent illegibility in many great discoveries and the coupling between institutions and legibility. Through this framework, an institution that supports these discoveries would need to do some combination of making the quality of the people doing the research legible completely separate from the research itself (this is close to what Braben did) or act as a trust broker where people trust the institution without demanding legibility at all. [Web URL for this note](http://notes.benjaminreinhardt.com/Institutions+either+make+things+more+legible+or+act+as+trust+brokers) [Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Institutions+either+make+things+more+legible+or+act+as+trust+brokers)