# Legible games enable people outside of an institution to control it If you want to make sure that someone isn’t up to any shenanigans, you need a way of measuring their outputs. In order to have measurable outputs you need to make the game they’re playing at least somewhat legible. The importance of legibility is the idea behind KPIs, Taylor’s scientific management, and a lot of modern economics. Similarly, if you want to make sure that an institution isn’t corrupt, you want a legible way of seeing how money is spent and people are hired to make sure that they aren’t spending money on fancy dinners and cigars or hiring their unqualified buddy from the golf club’s son over the highly qualified immigrant kid. At the end of the day, all of these metrics make an entity (individual, group, or institution) appear legible to outsiders. Of course [[Many things cannot be measured well]] and [[The Map is not the Territory]] so the metrics will vary on the accuracy of the picture they give. However, the the appearance of legibility is enough for outsiders to make judgements on how an entity is performing. Generally this corresponds to making some numbers go up and other numbers go down. Once you know how an entity is performing, you can then start to form opinions about what sort of inputs you should give it to try to make the numbers do what you want. More money and awards if the numbers are doing what you want; less money, shame, or sackings if the numbers are not doing what you want. It’s a control system: inputs informed by measured outputs. Bureaucracies depend on legible games because in effect [[Bureaucracies are human algorithms]] that need clear inputs. As a result, the bureaucratisation of an institution is usually coupled to making that institution more legible - academia, government, and large companies are clear examples of this. [[Scott Alexander]] argues that a lot of the thesis of [[Seeing like a State]] is that creating legibility is a [significant part of how governments exert control over their citizens](https://slatestarcodex.com/2017/03/16/book-review-seeing-like-a-state/). The desire to measure and control an institution often comes from a good place and can lead to good results - reducing corruption, incentivizing inefficiency, and making them more fair. The benefits of legibility are legion. However, making an institution more legible can also have many downsides. [[Optimizing for fair gatekeeping games is often disjoint for optimizing for the real game]], [[Fair games can be gamed]], and external control of an institution makes it more political: [[Politicization of an institution means that the people in the institution are more aligned with external goals than the institutions mission]]. Over time, people outside of institutions seem to have become more interested in controlling what happens inside of them. There are many possible explanations: * Perception that institutions are not delivering on their perceived mission ([[Institutions have implicit, explicit, and perceived missions]]), * A trend towards caring about behavior in all situations, * A trend towards wanting more fairness, * more value placed on measurability, * A trend towards wanting more transparency. In order to create this control, institutions have been pushed to become ever more fair and legible. My biggest concern is the impact that this trend has had on high variance institutions - specifically research. [[Making a game more fair reduces its variance]]. ### Related * [[Streetlight Effect]] * [[Formal process lets people outside the organization trust in the process instead of the people]] * [[Opacity is important to DARPA’s outlier success]] * [[Shared values allow you to break rules without breaking the spirit of the rules]] <!-- #evergreen --> [Web URL for this note](http://notes.benjaminreinhardt.com/Legible+games+enable+people+outside+of+an+institution+to+control+it) [Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Legible+games+enable+people+outside+of+an+institution+to+control+it)