Oh this is loads of fun. Usually solved in the past by cheapest solution of precisely synchronised clocks and having a time stamp on each message. This timing & reconstruction problem is not unique to computer networks but has been seen in various forms in a number of systems.
Your first question should be
Is it even possible to know the precise order of events and/or is it sufficient to use the majority view of the sequence. Of course there maybe no useful majority view since a sequence of 5 events could be viewed more ways then are nodes in a group in a worse case situation.
As to events and states question, as you already know the question is really what do you want to know. if you know what you want to know then working backwards you will be able to work out what you need to know in order to finally get what you need to know.
It always sounds easier than what it is and you need an open mind at all times since preconceived views always clouds the issue and can unfortunately lead to many wasted and passionate discussions.