# Externalized research is a buffer against B players If a performer underperforms, it is easy to never work with them again after that specific project. By contrast, employees are hard to get rid of. The reason it can be hard to get rid of an employee can be legal (especially in Europe, the US government, etc.) Fuzzier reasons why it’s hard to get rid of an employee can perhaps be bigger than the precise legal reasons. There is an implicit assumption that an employee is going to be around ‘forever’ so they can gain essential knowledge about a critical system. The emotional burden of getting rid of an employee can be high. Externalized researchers are ideally uncorrelated except through the person finding them. That is, you can’t get quality contagion (the second part of [[A players hire A players and B players hire C players]]) Externalized research could ideally lead to an asymmetric upside situation. If you really do like working with a contractor and they like working with you, you get most of the lowered [[Transaction costs]] of hiring an employee without the downsides. Speculatively, there is reason to believe that an organization made out of temporary people could actually have a *stronger* culture. See colleges, lab groups, etc.[[Intergenerational culture is like a standing wave]] ### Downsides Of course, there are downsides to externalized research. * It’s hard to have as tight feedback loops with external performers - this makes it crucial to have clear goals and clean interfaces. However, [[Creating legible interfaces is expensive]]. * Some research really just needs the ability to play, which is hard to do when you know you’re on a clock. [[Feedback loops and play are important for breakthroughs]] * Some great people want the security that comes from a default-alive job. ### Related * [[DARPA does not do any research in house]] * [[It is especially hard to capture value from externalized research]] * [[R+D orgs lie on a spectrum between externalized and internalized execution and benefits]] * Employment contracts are a subset of [[What if all entities had expiration dates?]] - in this case the “entity” is the employment term. <!-- #evergreen --> [Web URL for this note](http://notes.benjaminreinhardt.com/Externalized+research+is+a+buffer+against+B+players) [Comment on this note](http://via.hypothes.is/http://notes.benjaminreinhardt.com/Externalized+research+is+a+buffer+against+B+players)