Would you clarify a bit what claims you're making?  Is this a legal question (in what jurisdications is a copyright claim likely to hold up based on plot analysis when the text content is not copied)?  Or a broader IP advice (copyright isn't sufficient, you should patent or trademark parts of your work as well)?  Or something else (a moral claim - this is legal (defined as: courts don't provide relief), but shouldn't be done)?

I agree with, but am confused by, your description of important differences between the book, along with important similarities.  I expect MANY readers enjoyed both, and bought both.  Which rather reduces the reasoning for a court to enforce royalties on or prevent publication of the later book.  

People liked Brown’s book not just because of the meaningful ending that revealed the story as an allegory about the pursuit of knowledge and love. They liked it because they felt like they were learning something about religious symbols. I was fascinated to learn that Venus traces out the shape of a star relative to the ecliptic every eight years and I also enjoyed the few pages Brown wrote about PHI.

In constast, I liked Dunn’s book because of the plot and because I felt like I learned something about how banking criminals operated. Nevertheless, after reading both books, it is clear to me that Brown borrowed heavily from Dunn and lied about it. He used Dunn’s story as a template or road map and the lying is where he crossed a line, in my estimation. Then again, maybe his publisher wouldn’t allow him to tell the truth. I’ve heard that they can be bullies.

I don't know much about book publishing, but in software, it's well-known that algorithms are not covered under copyright.  Expression and literal code is.  Algorithms and designs can sometimes be patented, but there's shockingly little case law about where the lines are drawn.  Similarly in boardgames - one can copyright text, graphics, names, etc.  But not the actual mechanics - there are TONS of copycats of any popular game.  They mostly don't succeed because they're not actually an improvement, and there's no need being served by them.

I'd expect the same of books - significant chunks of text are covered by copyright, but not plotlines or even plot event sequences.  Unlike boardgames, though, there's a whole lot of authorial variation and expression in the prose that describes and reveals the plot, so there is value in reading multiple superficially-similar novels.

This is an interesting analysis. The Midnight Library was a shocker though. Never expected such a well promoted book to be plagiarized. How do they get away with it?

Hi, interesting stuff - I do NLP type research and will follow up on the links you provided.  But a minor issue - who is "Gunn"?  I assume that is a typo.