Inspired by an EconTalk episode about rational optimism, a Planet Money episode about dairy farming, a lecture about singular value decompositions and perhaps too much coffee, some more thoughts about specialization:

As Russ Roberts and Matt Ridley point out, the positive feedback loop between increasing trade/technology and increasing specialization is how humanity wins. The trend towards specialization was emphasized by the story about dairy farming. Not so many years ago, all the knowledge about how dairy farming worked could be kept in the head of the dairy farmer. Now, he hires a cow-breeding specialist, a cow-nutrition specialist and a cow-milking specialist to advise him on how to maximize output (and we all benefit due to the increased milk production.)

Whenever I hear these stories about the trend of specialization I get a bit worried, because I do not want to be a specialist. There are people who are content to devote their lives knowing everything about one thing, but I am not one of them.  But I’m pretty bullish on the paradigm that I introduced yesterday, in which the world relies not just on specialists, but also on connectors between them.  Many connections are provided by technology, but I don’t think that technology will ever replace human connectors. There is a uniquely human ability to see the products of two specialized area and mentally put them together to form something entirely new.

Think of an aqueduct. The columns are the specialists, rising narrowly ever upward. But if you just had many separate stacks of columns, they would quickly become unstable, useless and then fall over. The arches are the connectors – not increasing the height of the structure as much, per se, but by connecting the specialists, they allow specialization in entirely new areas (like halfway between the two original columns.) In the end, BOTH roles are crucial.

I also think that in a way, being a connector is much more risky than being a specialist. When you are a specialist, more specialization is always better. You know exactly what to do – specialize more. On the other hand, a connector can become too spread out, knowing too little about too many things, or not spread out enough. There’s a fuzzy optimum that can easily be over- or under- shot. The contrast in my mind is between an entrepreneur and a lab researcher. The latter has bounded risks and bounded rewards, while the former has far more variance.

I’m not sure if it’s my bias, but it seems like a majority of history's most impactful people were connectors.