A simple way to understand CRDTs
There's a simple way to understand CRDTs: It leverages algebra to unmix the inevitable mixing of data when syncing over an unreliable network.
Why are networks unreliable? For one, it'd be really expensive to build a network that assured 100% only-once delivery. But also, the internet was designed to allow variable routing paths and dynamic topologies to adapt to failures and congestion. This flexibility allows the network to stay up. The downside of this robustness is the variability of time delays and mixing of data when sending it from node to node.
There are three ways a network can mix up your data, and CRDTs constrain operations on the data to be algebraic to address each of them when syncing data.
As an aside, For state-based CRDTs, the operation is called a merge, where given a data packet coming in, we can merge it with the existing state. For op-based CRDTs, the operation is called an apply, but it's effectively a merge with one less required property. We'll get to that.
When data comes in out of sequential order, we can neutralize this mixing if our merge op is associative. I'll denote our operation as a dot.
