Creation v gamesmanship in the creation of characters

One difference between a novel, or at least a good novel, and a random string of chronicled events, is that a novel(ist) builds a sense of inevitability into the work: we feel when we read a good novel that the events had to happen in the very way that the novelist conveys them, that there is no other way that the characters could have behaved or played out the sequence of their lives. This sense of inevitability is not the same as plot. Plot is a kind of mechanism, a structure that places seemingly random elements or events in relation to one another. Plot gives the elements of a novel shape and dimension - building toward a climax, sometimes referred to as the "arc" of the story. The inevitability of the novel is really the heart and soul of fiction. When we "buy into" a novel, we accept that everything in the novel makes sense, that the characters are whole and not diffuse, and that their actions have irreversible consequences that lead to one single conclusion. (Of course, postmodernists have played with every one of these conventions.) The "inevitable" quality of fiction gets really put to the test in a novel that spans generations: a novel about a single character, particularly a novel that hews to the classical unities of a single time and space and even day, can more easily cohere because it need not stray beyond the lineaments of those unities. But a multi-generation novel is another case: each character may be whole and complete and its sequence of events may feel inevitable, but the offspring of the character is like subsequent random event: just because one character acts in a particular way does not necessarily lead to the conclusion that children of that character will act in a particular was; the 2nd (and 3rd) generation of characters are functionally independent integers. A really good novelist can make us accept the inevitability across generations - as Kate Walbert (?) did in The Short History of Women, which, though maddening in its refusal to tell the tale sequentially, did use one generation of characters to comment on and heighten our understanding of others - and that brings me to Alan Hollinghurst's "The Stranger's Child." I'm now in section 3, and each section is set in a different decade and generation; so far, only one character, Daphne, appears in all three sections. Hollinghurst is really playing with us here - as sections 2 and 3 open, he intentionally does not tell us right off who these "new" characters are, and we only gradually, by indirection, learn which we've met in previous sections and which are new. It's kind of frustrating that the most interesting, and seemingly most important, character - Cecil Valance, the poet - dies after section 1 and appears now only in memory. More frustrating, OK, I can accept that in part 1 Daphne is a young provincial girl begin seduced by her brother's (gay) friend (Cecil), in part 2 she is the wife of Cecil's surviving brother, a wealthy member of the landed gentry, and now, in part 3, she is a 70ish widow living with her dull banker son-in-law. Honestly, the only thing binding these three Daphnes into one character is Hollinghurst's assertion that these characters are the same person at different stages of her life. This structure is an example of authorial imposition or gamesmanship, rather than the author's creation of the sense of inevitability that holds great fiction together as a unified experience in our minds and our memories.