He holds that such theory is useful to know in general, but that real-world negotiations are carried out by emotional, irrational humans. As a result, these rational theories often break down in the face of hubris, anger, fear, love, shame, etc, and as such, relying solely on these theories will not get you the results you want. Thus, you should "never split the difference."
Instead, he then goes on to suggest a whole variety of ways one can achieve better negotiation results, starting with the most obvious: really, truly listening to the needs of the other. The book is fascinating. Highly recommended.
Imagine if it was a 24-slice pizza. The argument is that the "original" 6 slices should be divided according to the original preferences of 4 and 2, and the "extra" 18 slices should be divided evenly into 9 and 9.
You're the first person I've seen suggest that this is a thing and I'm kind of shocked. What reasoning would there be for unequal division of the additional slices in A's favor?
As an edge case, let's say A has 1M, and B has 1k and there's 500k new units. Should B get 250k same as A? Would A go for that? Would B be happy with 100k?