Ergodicity

Nov 22 / Network Capital
Understand the role that history (time), state (condition) and statistics together play in making decisions. For example, if you have to decide the average time it takes to bake a cake, you can either have 100 different people bake a cake at the same time or have one person bake hundred cakes. For each of the two cases the average time would not be the same. That one person baking a cake builds upon their experience of baking each time.

Subscribe to read more

Systems and processes are irrevocably connected to data, numbers and patters. Therefore, our ability to make sense of the numbers and data is extremely critical. Nassim Taleb in his book Skin in the Game beautifully explores different methods of understanding systems and processes. One of the key principles he examines is of ergodic and non-ergodic systems.

To put it simply, ergodic systems are those which have no ‘deep sense of history’. The average outcome of a sub-set is same as the outcome for the entire unit. Ergodic systems follow the same pattern across time, and return to every possible state infinite number of times.

Example of an ergodic system would be the production of an Apple iPhone 7. Each iPhone 7 would follow the same pattern of assembly and manufacturing. Similarly, all batches or consign- ments of iPhone 7s produced will follow the same loop of produc- tion. However, the usage of an iPhone 7 would not be an ergodic system.

Non-ergodic systems are those that are dynamic. They change due to multiple factors, including but not limited to—time, external stimuli, barriers, etc. Most human systems that we know are non- ergodic. This ranges from the biosphere that we live in to the iPhone 7 that we might use.

An understanding of ergodicity is therefore extremely critical to our understanding of the world’s functioning. Taleb in his book notes how policy-makers, economists, businesspersons, etc., make crucial decisions based on data. However, while doing so they tend use non-ergodic data to make future predictions. This is inherently inaccurate.For us to avoid the pitfalls of data and to fundamentally understand systems, we must start by understanding ergodicity. We must also understand that one system can be a combination of both ergodic and non-ergodic systems.