Event sequencing - Topics

Granger causality asks whether the past of one series helps predict another Sorting by local clock time breaks the moment a message crosses a time zone Stratigraphy tells you what came later but never how much later Why the same list of dated events can support two defensible orderings Why auto-increment sequence numbers beat timestamps for ordering transactions Two events stamped in the same millisecond: which one actually happened first How ancient historians cross-dated events using an eclipse both sides recorded A chain of events is not a chain of causes The myth of the first event: how do you prove nothing came before Concurrent does not mean simultaneous: some events are simply unordered Reordering a transaction log silently changes what the database believes happened Sequence diagrams record who sent what to whom but not what actually caused what How Google Docs decides whose keystroke came first when two people type at once Vector clocks detect concurrency but the vector grows with every node that touches the data Version vectors in CRDTs are vector clocks with a different bookkeeping job A Lamport timestamp of 1 followed by 2 proves nothing about causality