Loose notes about scaling

• A computer program is a collection of inflexible specifications of processes. So our bureaucracies.
• Primary difference: the executor in a computer program is a machine, in a bureaucracy it's one or multiple people.
• Programmers consistently overlook escape hatches from the fixed path. "I want to talk to the manager", the justice system, etc. Bureaucracies tend to have some mechanism for arbitration/adjucation. Computer programs much less so.
• This is also relevant for blockchain tech: is this lack of arbitration worth it because the current arbitration is so shit?

• Bureaucracy is a tool for scaling processes. You take trust in people and put it in processes.
• Most people (including regulators) want principle-based regulation. Protect your customers, etc. However, this does not scale because it requires unique insight in every individual case. Hence, compliancy bureaucracy.
• That's not to say that we should give up looking for principles-based regulation. There's a difference between functional and sclerotic bureaucracy. An example of the former is renewing your driver's license. An example of the latter is the GDPR compliancy.

• "Scaling" is too narrow a term. A lot depends on how far you want to scale. Millions of users can probably still use a simple websites that's one 1-2 servers.
• A lot of architecture astronauting happens here: people building unnecessary (distributed) things because "it scales".
• "Life Happens in the Pre-Asymptote" - N.N. Taleb. A lot of Taleb's work is about pre-asymptotics, and they matter!

• Great Founder Theory Manuscript (see chapter "How to Use Bureaucracies")
• Do things that don't scale