Sorry for the long delay in responding to your query. I have recently moved jobs and had to relocate. Unfortunately my ISP made an utter shambles of configuring my internet so I've been offline for so long I am getting withdrawal symptoms.
I don't have a copy of the book at the moment (packed away in boxes due to the house move) to understand the full context of those statements, but I generally agree with your assertion. Optimistic concurrency is about allowing overlapping reads as an alternative to locking the data. If multiple of those readers then try to update, the first write wins and the others fail (because they are operating on stale data). A version check might be used to determine if the data is stale, but that's just the implementation detail
I've actually worked in domains where last write wins was used as the concurrency strategy. However, it is distinctly different from optimistic concurrency. I think we probably made a typo there and meant "first write wins". I'll double-check when I've unpacked the books, and if so will get it added to the errata.
I do apologise for that, and hope it didn't spoil your enjoyment of the book. I wish I could show you how much effort we put into the book trying to make everything perfect, and removing all mistakes. I really am sorry that this got through.
If you have any other questions or comments, please let me know.