The challenges described in the article are primarily distributed systems problems, not inherent to all event-driven architectures.

Because long and hard to reason about chains of events form in a complex event driven system. It separates functions from the call site of the function. Its just much easier to only break this coupling when you start a new thread/task/request because you are doing a long running or network operation.