Decentralized Decision Making
Top down management is shown to be inefficient time and time again
First off, this is not a post on crypto. Not everything with the word "decentralized" means crypto.
This is a post on how organizations are structured and how decisions get made in those organizations. One thing I've observed in many organizations is that the natural inclination of most people is to have decisions made at the top. As an organization grows, more layers need to be added and each lower layer is allowed to make less and less impactful decisions. Those layers act as proxies for the decision makers at the top.
This model *feels* more organized. We crave organization because it feels like we're taking things seriously. It's why we idolize neatness or try to get our email inboxes to "zero". When things are controlled from the top, the organization feels like it is better coordinated. If it is not coordinated with top down decisions, then some of the layers need to be corrected to better apply the decisions from the top. Unless a conscious effort is made to do otherwise, we as humans will default to accepting this type of organization.
I'd argue that the conscious effort should always be made to allow more independant decisions made throughout the organization. Take this example of Barnes and Noble. At one point, the company controlled everything from the top. Everything from what books to carry, to where they were placed, to even the angle of the tables those books were on. The turn around came when the CEO changed the organization to allow each bookstore to make their own choices. The staff at each store takes into account who their customers actually are in their particular location and adapts the content of the store for those customers.
This is a model that has worked in a variety of other situations. The notion of franchises and franchisees itself is one that runs on this type of decision making. McDonald's was built on this kind of structure, as a whole if not the main corporate office. Capitalism itself is based on the idea that allowing multiple competing organizations to make their own decisions results in better economic outcomes than a powerful centralized body making choices for everyone. Yet despite many business leaders identifying as capitalists, they choose to run their company using the centralized model.
Why doesn't centralized decision making work? I think the best way to understand the limits of centralization is to look at chaos theory. This is the notion that some things are so complex that the outcome *seems* random, but was actually completely deterministic. Take a roll of dice. We all act as if rolling dice produces a random result. However, consider these factors:
dice have a set weight
they are thrown from a given height
they are thrown with a certain amount of force
there is a certain amount of air resistance where they are thrown
the surface they land on has a fixed shape and amount of friction
If you could account for all of these factors, you could determine what the result of a die roll will be 100% of the time. Accounting for all of these factors is completely beyond human capability though and so the roll of dice becomes random because it is too complex for us to predict the result.
Consider what the CEO of a company with 100 locations spread across the US must account for. People in NYC are not going to have the same preferences as people in Alabama. People in NYC don't even have the same preferences as people in the rest of New York. Hell, people in Brooklyn don't even have the same preferences as people in Manhattan. What people like is affected by all sorts of things:
who their parents are and how they were raised
who their friends (and how their parents raised those friends)
age
socioeconomic status
personal tragedies, if any
genetics
probably a thousand more factors I'm failing to think of
Now add 100 locations to the mix, all with their own local culture and preferences. How reasonable is it to expect a single human, or even a group of several humans, to be able to account for all the factors in those locations that result in the preferences for those people? No matter what those people do, the complexity involved will make every result seem random.
Now, what about companies with 1000 locations? Or ones with 10,000 locations?
Being able to make good decisions that account for all these factors in a large number of locations is likely impossible for humans. Reducing the scope of how many factors to consider is also not a guarantee of success, but it does result in a higher probability of success. Allowing the staff that is closest to the conditions for a location to make the decisions for that location significantly reduces how many factors a single decision maker has to think about.
Software development organizations suffer a similar problem. The natural inclination for most people is to organize by job function. Create a front end team, a backend team, a product management team, an SRE team, etc. The person running that team understands that particular job function and can best manage other people with that function.
The problem with this organization is that no one team can decide what to do by themselves. They're all dependant on each other. That requires someone overseeing every team to make decisions for all of them.
Contrast that with a team that is organized by product instead. A team for product A contains front end devs, backend devs, a product manager, etc. That team has everyone they need to make decisions for product A and execute on those decisions. The factors they need to consider are not those of the entire company, but scoped down to just the concerns of product A. Similar to companies with physical locations, allowing the people closest to the customers to make the decisions will result in better decisions.
Doing this is hard though. Most people, especially business leaders, like to feel in control. If you let staff make decisions, they will almost certainly make decisions that are different from yours. That is an uncomfortable thought. Leadership is hard though. It is not hard because controlling everything is hard, controlling everything is impossible. Leadership is hard because you have to come to terms that your job isn't control, but to enable the rest of the organization to do things on their own.