Nodes in Bitcoin do not care if these transactions were reversed across chains, they only care about the transactions on the one most work valid chain, where "valid" is a set of rules expressed in the software binary that node is running. Two nodes can disagree on validity rules and literally never find out. Or a block can trigger a fork in the 2 nodes and then we need developers to tell us which chain is valid Bitcoin (see 2017 fork). While this is happening, transactions can go from being settled in a valid most-work chain to suddenly being considered invalid because everyone else is rejecting that chain.
It's a faulty design for a global payment network. We'd expect it to be predictable in how it handles transaction reversal and for human entities to be held accountable for ruleset changes.
A) >It’s impossible to count how many of those abandoned transactions were successfully replayed on the new chain
now it's
B) >nodes don't know if re-org's are happening elsewhere in the network
Claim A is false as it is indeed entirely possible:
I have a chain of height 100. It is fully synced. I am sent a series of blocks diverging at height 80. I have all the information I need to ascertain which of the transactions in the two competing chains are being replayed or have been doublespent.
Claim B is true
The comment I was replying to claimed that transactions can never be reversed. I'm just highlighting how this represents a poor understanding of how Bitcoin really works.
>It’s impossible to count how many of those abandoned transactions were successfully replayed on the new chain
To detect if transactions were double spent, you would need some sort of database. Like a data base with _really_ good integrity. Maybe with periodic and limited deltas. Increase the some of the integrity with decentralized time stamping, sleeping-beauty hashed chaining. hmmm I wonder if there is anything like that...
Cheekiness aside, it's possible a node could ascertain whether or not transactions were replayed and definitely not impossible. The steel-man reply is: node software doesn't have great resources to do it.