Further Reading

The Challenge of Consistency in Distributed Systems

Achieving consistency across distributed systems is a notoriously difficult problem. The key reason is that, in a distributed environment, multiple nodes can independently make changes to the same piece of data. When different nodes hold different versions of this data, deciding how to reconcile these differences without losing valuable updates or introducing conflicts becomes a complex challenge.

Traditional approaches often require coordination mechanisms, such as consensus algorithms (like Paxos or Raft), to ensure consistency. However, these methods can be resource-intensive, require high communication overhead, and often struggle with scalability, especially when dealing with frequent updates across many nodes. The famous CAP theorem even states that distributed systems can only guarantee two of three properties (Consistency, Availability, and Partition Tolerance) at any given time, making it hard to achieve strong consistency while keeping a system always available and partition-tolerant.

In the 1960s psychologist Stanley Milgram conducted an influential experiment that revealed something amazing about human relationships. Milgram chose people at random in cities like Kansas and gave each a letter with the address of someone they didn’t know in Boston, Massachusetts. They were instructed to get the letter to that person but only by sending it to someone they know personally, who would send it to someone they know personally - and so on. Milgram repeated this letter-sending experiment nearly 200 times. On average, these letters reached their target in just six steps, this is where we get the term ‘six degrees of separation.’ Milgram’s findings demonstrated that despite the vastness of the world, most individuals are only a few links away from each other, highlighting the surprisingly small number of intermediaries connecting us all.